[PATCH 0 of 3 v4/leftover] Automatic NUMA placement for xl

[PATCH 0 of 3 v4/leftover] Automatic NUMA placement for xl

[Dario Faggioli]

Hello again,

This is a repost of what remained un-committed of my NUMA series. The splat on
1-node-only systems has been resolved as per Ian Campbell's suggestion
(thanks!), and that was the only thing I'm aware that was preventing from
checking-in these last patches. While at it, I also addressed all any other
observation made during review, namely, typos and rewording of some sentences,
as well as an improved filtering solution for avoiding generating a lot of
potential duplicate placement candidates when cpupool are being used.

All the patches have already been acked, although with some minor nits that
should be gone now.

If you have any further comment, I'm here to answer. :-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

[PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

If a domain does not have a VCPU affinity, try to pin it automatically to some
PCPUs. This is done taking into account the NUMA characteristics of the host.
In fact, we look for a combination of host's NUMA nodes with enough free memory
and number of PCPUs for the new domain, and pin it to the VCPUs of those nodes.

Once we know which ones, among all the possible combinations, represents valid
placement candidates for a domain, use some heuistics for deciding which is the
best. For instance, smaller candidates are considered to be better, both from
the domain's point of view (fewer memory spreading among nodes) and from the
system as a whole point of view (fewer memoy fragmentation).  In case of
candidates of equal sizes (i.e., with the same number of nodes), the amount of
free memory and the number of domain already assigned to their nodes are
considered. Very often, candidates with greater amount of memory are the one
we wants, as this is also good for keeping memory fragmentation under control.
However, if the difference in how much free memory two candidates have, the
number of assigned domains might be what decides which candidate wins.

This all happens internally to libxl, and no API for driving the mechanism is
provided for now. This matches what xend already does.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>

---
Changes from v3:
 * fixed a double free on non-NUMA systems (namely, when just 1 node exists).
 * Reword an ambiguous sentence in xl's man page, as requested during review.
 * Do no special case single node systems in libxl__get_numa_candidates(), to
   avoid generating a spurious warning on them.

Changes from v2:
 * lots of typos.
 * Clayfied some comments, as requested during (ijc's) review.
 * Added some more information/reference for the combination generation
   algorithm.
 * nodemap_to_nodes_cpus() function renamed to nodemap_to_nr_cpus().
 * libxl_bitmap_init() used to make sure we do not try to free random
   memory on failure paths of functions that allocates a libxl_bitmap.
 * Always invoke libxl__sort_numa_candidates(), even if we get there
   with just 1 candidate, as requested during review.
 * Simplified the if-s that check for input parameter consistency in
   libxl__get_numa_candidates() as requested during (gwd's) review.
 * Comparison function for candidates changed so that it now provides
   total ordering, as requested during review. It is still using FP
   arithmetic, though. Also I think that just putting the difference
   between the amount of free memory and between the number of assigned
   domains of two candidates in a single formula (after normalizing and
   weighting them) is both clear and effective enough.
 * Function definitions moved to a numa specific source file (libxl_numa.c),
   as suggested during review.


Changes from v1:
 * This patches incorporates the changes from both "libxl, xl: enable automatic
   placement of guests on NUMA nodes" and "libxl, xl: heuristics for reordering
   NUMA placement candidates" from v1.
 * The logic of the algorithm is basically the same as in v1, but the splitting
   of it in the various functions has been completely redesigned from scratch.
 * No public API for placement or candidate generation is now exposed,
   everything happens within libxl, as agreed during v1 review.
 * The relevant documentation have been moved near the actual functions and
   features. Also, the amount and (hopefully!) the quality of the documentation
   has been improved a lot, as requested.
 * All the comments about using the proper libxl facilities and helpers for
   allocations, etc., have been considered and applied.
 * This patch still bails out from NUMA optimizations if it find out cpupools
   are being utilized. It is next patch that makes the two things interact
   properly, as suggested during review.

diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
--- a/docs/man/xl.cfg.pod.5
+++ b/docs/man/xl.cfg.pod.5
@@ -111,8 +111,8 @@ created online and the remainder will be
 
 =item B<cpus="CPU-LIST">
 
-List of which cpus the guest is allowed to use. Default behavior is
-`all cpus`. A C<CPU-LIST> may be specified as follows:
+List of which cpus the guest is allowed to use. By default xl will (via
+libxl) pick some cpus (see below). A C<CPU-LIST> may be specified as follows:
 
 =over 4
 
@@ -132,6 +132,12 @@ run on cpu #3 of the host.
 
 =back
 
+If this option is not specified, libxl automatically tries to place the new
+domain on the host's NUMA nodes (provided the host has more than one NUMA
+node) by pinning it to the cpus of those nodes. A heuristic approach is
+utilized with the goals of maximizing performance for the domain and, at
+the same time, achieving efficient utilization of the host's CPUs and RAM.
+
 =item B<cpu_weight=WEIGHT>
 
 A domain with a weight of 512 will get twice as much CPU as a domain
diff --git a/tools/libxl/Makefile b/tools/libxl/Makefile
--- a/tools/libxl/Makefile
+++ b/tools/libxl/Makefile
@@ -66,7 +66,7 @@ LIBXL_LIBS += -lyajl -lm
 LIBXL_OBJS = flexarray.o libxl.o libxl_create.o libxl_dm.o libxl_pci.o \
 			libxl_dom.o libxl_exec.o libxl_xshelp.o libxl_device.o \
 			libxl_internal.o libxl_utils.o libxl_uuid.o \
-			libxl_json.o libxl_aoutils.o \
+			libxl_json.o libxl_aoutils.o libxl_numa.o \
 			libxl_save_callout.o _libxl_save_msgs_callout.o \
 			libxl_qmp.o libxl_event.o libxl_fork.o $(LIBXL_OBJS-y)
 LIBXL_OBJS += _libxl_types.o libxl_flask.o _libxl_types_internal.o
diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -98,6 +98,106 @@ out:
     return sched;
 }
 
+/* Subtract two values and translate the result in [0, 1] */
+static double normalized_diff(double a, double b)
+{
+#define max(a, b) (a > b ? a : b)
+    if (!a && a == b)
+        return 0.0;
+    return (a - b) / max(a, b);
+}
+
+/*
+ * The NUMA placement candidates are reordered according to the following
+ * heuristics:
+ *  - candidates involving fewer nodes come first. In case two (or
+ *    more) candidates span the same number of nodes,
+ *  - the amount of free memory and the number of domains assigned to the
+ *    candidates are considered. In doing that, candidates with greater
+ *    amount of free memory and fewer domains assigned to them are preferred,
+ *    with free memory "weighting" three times as much as number of domains.
+ */
+static int numa_cmpf(const void *v1, const void *v2)
+{
+    const libxl__numa_candidate *c1 = v1;
+    const libxl__numa_candidate *c2 = v2;
+#define sign(a) a > 0 ? 1 : a < 0 ? -1 : 0
+    double freememkb_diff = normalized_diff(c2->free_memkb, c1->free_memkb);
+    double nrdomains_diff = normalized_diff(c1->nr_domains, c2->nr_domains);
+
+    if (c1->nr_nodes != c2->nr_nodes)
+        return c1->nr_nodes - c2->nr_nodes;
+
+    return sign(3*freememkb_diff + nrdomains_diff);
+}
+
+/* The actual automatic NUMA placement routine */
+static int numa_place_domain(libxl__gc *gc, libxl_domain_build_info *info)
+{
+    int nr_candidates = 0;
+    libxl__numa_candidate *candidates = NULL;
+    libxl_bitmap candidate_nodemap;
+    libxl_cpupoolinfo *pinfo;
+    int nr_pools, rc = 0;
+    uint32_t memkb;
+
+    libxl_bitmap_init(&candidate_nodemap);
+
+    /* First of all, if cpupools are in use, better not to mess with them */
+    pinfo = libxl_list_cpupool(CTX, &nr_pools);
+    if (!pinfo)
+        return ERROR_FAIL;
+    if (nr_pools > 1) {
+        LOG(NOTICE, "skipping NUMA placement as cpupools are in use");
+        goto out;
+    }
+
+    rc = libxl_domain_need_memory(CTX, info, &memkb);
+    if (rc)
+        goto out;
+    if (libxl_node_bitmap_alloc(CTX, &candidate_nodemap, 0)) {
+        rc = ERROR_FAIL;
+        goto out;
+    }
+
+    /* Find all the candidates with enough free memory and at least
+     * as much pcpus as the domain has vcpus.  */
+    rc = libxl__get_numa_candidates(gc, memkb, info->max_vcpus, 0, 0,
+                                    &candidates, &nr_candidates);
+    if (rc)
+        goto out;
+
+    LOG(DETAIL, "%d NUMA placement candidates found", nr_candidates);
+
+    /* No suitable placement candidates. We just return without touching the
+     * domain's info->cpumap. It will have affinity with all nodes/cpus. */
+    if (nr_candidates == 0) {
+        LOG(NOTICE, "NUMA placement failed, performance might be affected");
+        goto out;
+    }
+
+    /* Bring the best candidate in front of the list --> candidates[0] */
+    libxl__sort_numa_candidates(candidates, nr_candidates, numa_cmpf);
+
+    /*
+     * At this point, the first candidate in the array is the one we want.
+     * Go for it by mapping its node map to the domain's info->cpumap.
+     */
+    libxl__numa_candidate_get_nodemap(gc, &candidates[0], &candidate_nodemap);
+    rc = libxl_nodemap_to_cpumap(CTX, &candidate_nodemap, &info->cpumap);
+    if (rc)
+        goto out;
+
+    LOG(DETAIL, "NUMA placement candidate with %d nodes, %d cpus and "
+                "%"PRIu32" KB free selected", candidates[0].nr_nodes,
+                candidates[0].nr_cpus, candidates[0].free_memkb / 1024);
+
+ out:
+    libxl_bitmap_dispose(&candidate_nodemap);
+    libxl_cpupoolinfo_list_free(pinfo, nr_pools);
+    return rc;
+}
+
 int libxl__build_pre(libxl__gc *gc, uint32_t domid,
               libxl_domain_build_info *info, libxl__domain_build_state *state)
 {
@@ -107,7 +207,22 @@ int libxl__build_pre(libxl__gc *gc, uint
     uint32_t rtc_timeoffset;
 
     xc_domain_max_vcpus(ctx->xch, domid, info->max_vcpus);
+
+    /*
+     * Check if the domain has any CPU affinity. If not, try to build up one.
+     * In case numa_place_domain() find at least a suitable candidate, it will
+     * affect info->cpumap accordingly; if it does not, it just leaves it
+     * as it is. This means (unless some weird error manifests) the subsequent
+     * call to libxl_set_vcpuaffinity_all() will do the actual placement,
+     * whatever that turns out to be.
+     */
+    if (libxl_bitmap_is_full(&info->cpumap)) {
+        int rc = numa_place_domain(gc, info);
+        if (rc)
+            return rc;
+    }
     libxl_set_vcpuaffinity_all(ctx, domid, info->max_vcpus, &info->cpumap);
+
     xc_domain_setmaxmem(ctx->xch, domid, info->target_memkb + LIBXL_MAXMEM_CONSTANT);
     if (info->type == LIBXL_DOMAIN_TYPE_PV)
         xc_domain_set_memmap_limit(ctx->xch, domid,
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -2216,6 +2216,140 @@ static inline void libxl__ctx_unlock(lib
 #define CTX_LOCK (libxl__ctx_lock(CTX))
 #define CTX_UNLOCK (libxl__ctx_unlock(CTX))
 
+/*
+ * Automatic NUMA placement
+ *
+ * These functions and data structures deal with the initial placement of a
+ * domain onto the host NUMA nodes.
+ *
+ * The key concept here is the one of "NUMA placement candidate", which is
+ * basically a set of nodes whose characteristics have been successfully
+ * checked against some specific requirements. More precisely, a candidate is
+ * the nodemap associated with one of the possible subset of the host NUMA
+ * nodes providing a certain amount of free memory, or a given number of cpus,
+ * or even both (depending in what the caller wants). For convenience of use,
+ * some of this information are stored within the candidate itself, instead of
+ * always being dynamically computed. A single node is a valid placement
+ * candidate, but it is also possible for a candidate to contain all the nodes
+ * of the host. The fewer nodes there are in a candidate, the better
+ * performance a domain placed onto it should get. For instance, looking for a
+ * numa candidates with 2GB of free memory means we want all the possible
+ * subsets of the host NUMA nodes with, cumulatively, at least 2GB of free
+ * memory. That could be possible by just using one particular node, or may
+ * require more nodes, depending on the characteristics of the host, on how
+ * many domains have been created already, on how big they are, etc.
+ *
+ * The intended usage is as follows:
+ *  1. by, fist of all, calling libxl__get_numa_candidates(), and specifying
+ *     the proper constraints to it (e.g., the amount of memory a domain need
+ *     as the minimum amount of free memory for the candidates) one can build
+ *     up a whole set of suitable placing alternatives for a domain;
+ *  2. after that, one specific candidate should be chosen. That can happen
+ *     by looking at their various characteristics;
+ *  3. the chosen candidate's nodemap should be utilized for computing the
+ *     actual affinity of the domain which, given the current NUMA support
+ *     in the hypervisor, is what determines the placement of the domain's
+ *     vcpus and memory.
+ *
+ * To make phase 2 even easier, a sorting helper function for the list of
+ * candidates is provided in the form of libxl__sort_numa_candidates(). The
+ * only that is needed is defining a comparison function, containing the
+ * criteria for deciding, given two candidates, which one is 'better'.
+ * Depending on how the comparison function is defined, the best candidate
+ * (where, of course, best is defined with respect to the heuristics
+ * implemented in the comparison function itself, libxl__numa_candidate_cmpf())
+ * could become the first or the last element of the list.
+ *
+ * Summarizing, achieving automatic NUMA placement is just a matter of
+ * obtaining the list of suitable placement candidates, perhaps asking for each
+ * of them to provide at least the amount of memory the domain needs. After
+ * that just implement a comparison function by means of the various helpers
+ * retrieving the relevant information about the candidates themselves.
+ * Finally, call the sorting helper function and use the candidate that became
+ * (typically) the first element of the list for determining the domain's
+ * affinity.
+ */
+
+typedef struct {
+    int nr_cpus, nr_nodes;
+    int nr_domains;
+    uint32_t free_memkb;
+    libxl_bitmap nodemap;
+} libxl__numa_candidate;
+
+/*
+ * This generates the list of NUMA placement candidates satisfying some
+ * specific conditions. If min_nodes and/or max_nodes are not 0, their value is
+ * used to determine the minimum and maximum number of nodes that are allow to
+ * be present in each candidate. If min_nodes and/or max_nodes are 0, the
+ * minimum and maximum number of nodes to be used are automatically selected by
+ * the implementation (and that will likely be just 1 node for the minimum and
+ * the total number of existent nodes for the maximum). Re min_free_memkb and
+ * min_cpu, if not 0, it means the caller only wants candidates with at
+ * least that amount of free memory and that number of cpus, respectively. If
+ * min_free_memkb and/or min_cpus are 0, the candidates' free memory and number
+ * of cpus won't be checked at all, which means a candidate will always be
+ * considered suitable wrt the specific constraint.  cndts is where the list of
+ * exactly nr_cndts candidates is returned. Note that, in case no candidates
+ * are found at all, the function returns successfully, but with nr_cndts equal
+ * to zero.
+ */
+_hidden int libxl__get_numa_candidates(libxl__gc *gc,
+                                uint32_t min_free_memkb, int min_cpus,
+                                int min_nodes, int max_nodes,
+                                libxl__numa_candidate *cndts[], int *nr_cndts);
+
+/* Initialization, allocation and deallocation for placement candidates */
+static inline void libxl__numa_candidate_init(libxl__numa_candidate *cndt)
+{
+    cndt->free_memkb = 0;
+    cndt->nr_cpus = cndt->nr_nodes = cndt->nr_domains = 0;
+    libxl_bitmap_init(&cndt->nodemap);
+}
+
+static inline int libxl__numa_candidate_alloc(libxl__gc *gc,
+                                              libxl__numa_candidate *cndt)
+{
+    return libxl_node_bitmap_alloc(CTX, &cndt->nodemap, 0);
+}
+static inline void libxl__numa_candidate_dispose(libxl__numa_candidate *cndt)
+{
+    libxl_bitmap_dispose(&cndt->nodemap);
+}
+static inline void libxl__numacandidate_list_free(libxl__numa_candidate *cndts,
+                                                  int nr_cndts)
+{
+    int i;
+
+    for (i = 0; i < nr_cndts; i++)
+        libxl__numa_candidate_dispose(&cndts[i]);
+    free(cndts);
+}
+
+/* Retrieve (in nodemap) the node map associated to placement candidate cndt */
+static inline
+void libxl__numa_candidate_get_nodemap(libxl__gc *gc,
+                                       const libxl__numa_candidate *cndt,
+                                       libxl_bitmap *nodemap)
+{
+    libxl_bitmap_copy(CTX, nodemap, &cndt->nodemap);
+}
+/* Set the node map of placement candidate cndt to match nodemap */
+static inline
+void libxl__numa_candidate_put_nodemap(libxl__gc *gc,
+                                       libxl__numa_candidate *cndt,
+                                       const libxl_bitmap *nodemap)
+{
+    libxl_bitmap_copy(CTX, &cndt->nodemap, nodemap);
+}
+
+/* Signature for the comparison function between two candidates c1 and c2 */
+typedef int (*libxl__numa_candidate_cmpf)(const void *v1, const void *v2);
+/* Sort the list of candidates in cndts (an array with nr_cndts elements in
+ * it) using cmpf for comparing two candidates. Uses libc's qsort(). */
+_hidden void libxl__sort_numa_candidates(libxl__numa_candidate cndts[],
+                                         int nr_cndts,
+                                         libxl__numa_candidate_cmpf cmpf);
 
 /*
  * Inserts "elm_new" into the sorted list "head".
diff --git a/tools/libxl/libxl_numa.c b/tools/libxl/libxl_numa.c
new file mode 100644
--- /dev/null
+++ b/tools/libxl/libxl_numa.c
@@ -0,0 +1,377 @@
+/*
+ * Copyright (C) 2012      Citrix Ltd.
+ * Author Dario Faggioli <dario.faggioli@citrix.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published
+ * by the Free Software Foundation; version 2.1 only. with the special
+ * exception on linking described in file LICENSE.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Lesser General Public License for more details.
+ */
+
+#include "libxl_osdeps.h" /* must come before any other headers */
+
+#include <glob.h>
+
+#include "libxl_internal.h"
+
+/*
+ * What follows are helpers for generating all the k-combinations
+ * without repetitions of a set S with n elements in it. Formally
+ * speaking, they are subsets of k distinct elements of S and, if
+ * S is n elements big, the number of k-combinations is equal to
+ * the binomial coefficient (n k), which means we get exactly
+ * n!/(k! * (n - k)!) subsets, all of them with k elements.
+ *
+ * The various subset are generated one after the other by calling
+ * comb_init() first, and, after that, comb_next()
+ * (n k)-1 times. An iterator is used to store the current status
+ * of the whole generation operation (i.e., basically, the last
+ * combination that has been generated). As soon as all
+ * combinations have been generated, comb_next() will
+ * start returning 0 instead of 1. It is of course important that
+ * the same instance of the iterator and the same values for
+ * n and k are used for each call. If that doesn't happen, the
+ * result is unspecified.
+ *
+ * The algorithm is a well known one (see, for example, D. Knuth's "The
+ * Art of Computer Programming - Volume 4, Fascicle 3" and it produces
+ * the combinations in such a way that they (well, more precisely,
+ * their indexes it the array/map representing the set) come with
+ * lexicographic ordering.
+ *
+ * For example, with n = 5 and k = 3, calling comb_init()
+ * will generate { 0, 1, 2 }, while subsequent valid calls to
+ * comb_next() will produce the following:
+ * { { 0, 1, 3 }, { 0, 1, 4 },
+ *   { 0, 2, 3 }, { 0, 2, 4 }, { 0, 3, 4 },
+ *   { 1, 2, 3 }, { 1, 2, 4 }, { 1, 3, 4 },
+ *   { 2, 3, 4 } }
+ *
+ * This is used by the automatic NUMA placement logic below.
+ */
+typedef int* comb_iter_t;
+
+static int comb_init(libxl__gc *gc, comb_iter_t *it, int n, int k)
+{
+    comb_iter_t new_iter;
+    int i;
+
+    if (n < k)
+        return 0;
+
+    /* First set is always { 0, 1, 2, ..., k-1 } */
+    GCNEW_ARRAY(new_iter, k);
+    for (i = 0; i < k; i++)
+        new_iter[i] = i;
+
+    *it = new_iter;
+    return 1;
+}
+
+static int comb_next(comb_iter_t it, int n, int k)
+{
+    int i;
+
+    /*
+     * The idea here is to find the leftmost element from where
+     * we should start incrementing the indexes of the iterator.
+     * This means looking for the highest index that can be increased
+     * while still producing value smaller than n-1. In the example
+     * above, when dealing with { 0, 1, 4 }, such an element is the
+     * second one, as the third is already equal to 4 (which actually
+     * is n-1).
+     * Once we found from where to start, we increment that element
+     * and override the right-hand rest of the iterator with its
+     * successors, thus achieving lexicographic ordering.
+     *
+     * Regarding the termination of the generation process, when we
+     * manage in bringing n-k at the very first position of the iterator,
+     * we know that is the last valid combination ( { 2, 3, 4 }, with
+     * n - k = 5 - 2 = 2, in the example above), and thus we start
+     * returning 0 as soon as we cross that border.
+     */
+    for (i = k - 1; it[i] == n - k + i; i--) {
+        if (i <= 0)
+            return 0;
+    }
+    for (it[i]++, i++; i < k; i++)
+        it[i] = it[i - 1] + 1;
+    return 1;
+}
+
+/* NUMA automatic placement (see libxl_internal.h for details) */
+
+/*
+ * This function turns a k-combination iterator into a node map.
+ * This means the bits in the node map corresponding to the indexes
+ * of the given combination are the ones that will be set.
+ * For example, if the iterator represents the combination { 0, 2, 4},
+ * the node map will have bits #0, #2 and #4 set.
+ */
+static void comb_get_nodemap(comb_iter_t it, libxl_bitmap *nodemap, int k)
+{
+    int i;
+
+    libxl_bitmap_set_none(nodemap);
+    for (i = 0; i < k; i++)
+        libxl_bitmap_set(nodemap, it[i]);
+}
+
+/* Retrieve the number of cpus that the nodes that are part of the nodemap
+ * span. */
+static int nodemap_to_nr_cpus(libxl_cputopology *tinfo, int nr_cpus,
+                              const libxl_bitmap *nodemap)
+{
+    int i, nodes_cpus = 0;
+
+    for (i = 0; i < nr_cpus; i++) {
+        if (libxl_bitmap_test(nodemap, tinfo[i].node))
+            nodes_cpus++;
+    }
+    return nodes_cpus;
+}
+
+/* Retrieve the amount of free memory within the nodemap */
+static uint32_t nodemap_to_free_memkb(libxl_numainfo *ninfo,
+                                      libxl_bitmap *nodemap)
+{
+    uint32_t free_memkb = 0;
+    int i;
+
+    libxl_for_each_set_bit(i, *nodemap)
+        free_memkb += ninfo[i].free / 1024;
+
+    return free_memkb;
+}
+
+/* Retrieve the number of domains that can potentially run on the cpus
+ * the nodes that are part of the nodemap. */
+static int nodemap_to_nr_domains(libxl__gc *gc, libxl_cputopology *tinfo,
+                                 const libxl_bitmap *nodemap)
+{
+    libxl_dominfo *dinfo = NULL;
+    libxl_bitmap dom_nodemap;
+    int nr_doms, nr_cpus;
+    int nr_domains = 0;
+    int i, j, k;
+
+    dinfo = libxl_list_domain(CTX, &nr_doms);
+    if (dinfo == NULL)
+        return ERROR_FAIL;
+
+    if (libxl_node_bitmap_alloc(CTX, &dom_nodemap, 0) < 0) {
+        libxl_dominfo_list_free(dinfo, nr_doms);
+        return ERROR_FAIL;
+    }
+
+    for (i = 0; i < nr_doms; i++) {
+        libxl_vcpuinfo *vinfo;
+        int nr_vcpus;
+
+        vinfo = libxl_list_vcpu(CTX, dinfo[i].domid, &nr_vcpus, &nr_cpus);
+        if (vinfo == NULL)
+            continue;
+
+        libxl_bitmap_set_none(&dom_nodemap);
+        for (j = 0; j < nr_vcpus; j++) {
+            libxl_for_each_set_bit(k, vinfo[j].cpumap)
+                libxl_bitmap_set(&dom_nodemap, tinfo[k].node);
+        }
+
+        libxl_for_each_set_bit(j, dom_nodemap) {
+            if (libxl_bitmap_test(nodemap, j)) {
+                nr_domains++;
+                break;
+            }
+        }
+
+        libxl_vcpuinfo_list_free(vinfo, nr_vcpus);
+    }
+
+    libxl_bitmap_dispose(&dom_nodemap);
+    libxl_dominfo_list_free(dinfo, nr_doms);
+    return nr_domains;
+}
+
+/*
+ * This function tries to figure out if the host has a consistent number
+ * of cpus along all its NUMA nodes. In fact, if that is the case, we can
+ * calculate the minimum number of nodes needed for a domain by just
+ * dividing its total number of vcpus by this value computed here.
+ * However, we are not allowed to assume that all the nodes have the
+ * same number of cpus. Therefore, in case discrepancies among different
+ * nodes are found, this function just returns 0, for the caller to know
+ * it shouldn't rely on this 'optimization', and sort out things in some
+ * other way (most likely, just start trying with candidates with just
+ * one node).
+ */
+static int cpus_per_node_count(libxl_cputopology *tinfo, int nr_cpus,
+                               libxl_numainfo *ninfo, int nr_nodes)
+{
+    int cpus_per_node = 0;
+    int j, i;
+
+    /* This makes sense iff # of PCPUs is the same for all nodes */
+    for (j = 0; j < nr_nodes; j++) {
+        int curr_cpus = 0;
+
+        for (i = 0; i < nr_cpus; i++) {
+            if (tinfo[i].node == j)
+                curr_cpus++;
+        }
+        /* So, if the above does not hold, turn the whole thing off! */
+        cpus_per_node = cpus_per_node == 0 ? curr_cpus : cpus_per_node;
+        if (cpus_per_node != curr_cpus)
+            return 0;
+    }
+    return cpus_per_node;
+}
+
+/* Get all the placement candidates satisfying some specific conditions */
+int libxl__get_numa_candidates(libxl__gc *gc,
+                               uint32_t min_free_memkb, int min_cpus,
+                               int min_nodes, int max_nodes,
+                               libxl__numa_candidate *cndts[], int *nr_cndts)
+{
+    libxl__numa_candidate *new_cndts = NULL;
+    libxl_cputopology *tinfo = NULL;
+    libxl_numainfo *ninfo = NULL;
+    int nr_nodes = 0, nr_cpus = 0;
+    libxl_bitmap nodemap;
+    int array_size, rc;
+
+    libxl_bitmap_init(&nodemap);
+
+    /* Get platform info and prepare the map for testing the combinations */
+    ninfo = libxl_get_numainfo(CTX, &nr_nodes);
+    if (ninfo == NULL)
+        return ERROR_FAIL;
+
+    tinfo = libxl_get_cpu_topology(CTX, &nr_cpus);
+    if (tinfo == NULL) {
+        rc = ERROR_FAIL;
+        goto out;
+    }
+
+    rc = libxl_node_bitmap_alloc(CTX, &nodemap, 0);
+    if (rc)
+        goto out;
+
+    /*
+     * If the minimum number of NUMA nodes is not explicitly specified
+     * (i.e., min_nodes == 0), we try to figure out a sensible number of nodes
+     * from where to start generating candidates, if possible (or just start
+     * from 1 otherwise). The maximum number of nodes should not exceed the
+     * number of existent NUMA nodes on the host, or the candidate generation
+     * won't work properly.
+     */
+    if (!min_nodes) {
+        int cpus_per_node;
+
+        cpus_per_node = cpus_per_node_count(tinfo, nr_cpus, ninfo, nr_nodes);
+        if (cpus_per_node == 0)
+            min_nodes = 1;
+        else
+            min_nodes = (min_cpus + cpus_per_node - 1) / cpus_per_node;
+    }
+    if (min_nodes > nr_nodes)
+        min_nodes = nr_nodes;
+    if (!max_nodes || max_nodes > nr_nodes)
+        max_nodes = nr_nodes;
+    if (min_nodes > max_nodes) {
+        rc = ERROR_INVAL;
+        goto out;
+    }
+
+    /* Initialize the local storage for the combinations */
+    *nr_cndts = 0;
+    array_size = nr_nodes;
+    GCNEW_ARRAY(new_cndts, array_size);
+
+    /* Generate all the combinations of any size from min_nodes to
+     * max_nodes (see comb_init() and comb_next()). */
+    while (min_nodes <= max_nodes) {
+        comb_iter_t comb_iter;
+        int comb_ok;
+
+        /*
+	 * And here it is. Each step of this cycle generates a combination of
+	 * nodes as big as min_nodes mandates.  Each of these combinations is
+	 * checked against the constraints provided by the caller (namely,
+	 * amount of free memory and number of cpus) and it becomes an actual
+	 * placement candidate iff it passes the check.
+         */
+        for (comb_ok = comb_init(gc, &comb_iter, nr_nodes, min_nodes); comb_ok;
+             comb_ok = comb_next(comb_iter, nr_nodes, min_nodes)) {
+            uint32_t nodes_free_memkb;
+            int nodes_cpus;
+
+            comb_get_nodemap(comb_iter, &nodemap, min_nodes);
+
+            /* If there is not enough memory in this combination, skip it
+             * and go generating the next one... */
+            nodes_free_memkb = nodemap_to_free_memkb(ninfo, &nodemap);
+            if (min_free_memkb && nodes_free_memkb < min_free_memkb)
+                continue;
+
+            /* And the same applies if this combination is short in cpus */
+            nodes_cpus = nodemap_to_nr_cpus(tinfo, nr_cpus, &nodemap);
+            if (min_cpus && nodes_cpus < min_cpus)
+                continue;
+
+            /*
+             * Conditions are met, we can add this combination to the
+             * NUMA placement candidates list. We first make sure there
+             * is enough space in there, and then we initialize the new
+             * candidate element with the node map corresponding to the
+             * combination we are dealing with. Memory allocation for
+             * expanding the array that hosts the list happens in chunks
+             * equal to the number of NUMA nodes in the system (to
+             * avoid allocating memory each and every time we find a
+             * new candidate).
+             */
+            if (*nr_cndts == array_size)
+                array_size += nr_nodes;
+            GCREALLOC_ARRAY(new_cndts, array_size);
+
+            libxl__numa_candidate_alloc(gc, &new_cndts[*nr_cndts]);
+            libxl__numa_candidate_put_nodemap(gc, &new_cndts[*nr_cndts],
+                                              &nodemap);
+            new_cndts[*nr_cndts].nr_domains =
+                                    nodemap_to_nr_domains(gc, tinfo, &nodemap);
+            new_cndts[*nr_cndts].free_memkb = nodes_free_memkb;
+            new_cndts[*nr_cndts].nr_nodes = min_nodes;
+            new_cndts[*nr_cndts].nr_cpus = nodes_cpus;
+
+            LOG(DEBUG, "NUMA placement candidate #%d found: nr_nodes=%d, "
+                       "nr_cpus=%d, free_memkb=%"PRIu32"", *nr_cndts,
+                       min_nodes, new_cndts[*nr_cndts].nr_cpus,
+                       new_cndts[*nr_cndts].free_memkb / 1024);
+
+            (*nr_cndts)++;
+        }
+        min_nodes++;
+    }
+
+    *cndts = new_cndts;
+ out:
+    libxl_bitmap_dispose(&nodemap);
+    libxl_cputopology_list_free(tinfo, nr_cpus);
+    libxl_numainfo_list_free(ninfo, nr_nodes);
+    return rc;
+}
+
+void libxl__sort_numa_candidates(libxl__numa_candidate cndts[], int nr_cndts,
+                                 libxl__numa_candidate_cmpf cmpf)
+{
+    /* Reorder candidates (see the comparison function for
+     * the details on the heuristics) */
+    qsort(cndts, nr_cndts, sizeof(cndts[0]), cmpf);
+}
+
+

Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Ian Jackson]

Dario Faggioli writes ("[PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
> If a domain does not have a VCPU affinity, try to pin it automatically to some
> PCPUs. This is done taking into account the NUMA characteristics of the host.
> In fact, we look for a combination of host's NUMA nodes with enough free memory
> and number of PCPUs for the new domain, and pin it to the VCPUs of those nodes.

Thanks for this admirably clear patch.

Can you please rewrap your commit messages to around 70 lines ?  Many
VCSs indent them in the log in some situations, and as you see here
mail programs indent them when quoting too.

Indeed I would generally prefer to wrap code to ~75 rather than 80 but
I know that's a bit controversial.

> +/* Subtract two values and translate the result in [0, 1] */
> +static double normalized_diff(double a, double b)
> +{
> +#define max(a, b) (a > b ? a : b)
> +    if (!a && a == b)
> +        return 0.0;
> +    return (a - b) / max(a, b);
> +}

1. This macro max() should be in libxl_internal.h.
2. It should be MAX so people are warned it's a macro
3. It should have all the necessary ()s for macro precedence safety

> +static int numa_cmpf(const void *v1, const void *v2)
> +{
> +    const libxl__numa_candidate *c1 = v1;
> +    const libxl__numa_candidate *c2 = v2;
> +#define sign(a) a > 0 ? 1 : a < 0 ? -1 : 0

Again.

> +    double freememkb_diff = normalized_diff(c2->free_memkb, c1->free_memkb);
> +    double nrdomains_diff = normalized_diff(c1->nr_domains, c2->nr_domains);
> +
> +    if (c1->nr_nodes != c2->nr_nodes)
> +        return c1->nr_nodes - c2->nr_nodes;
> +
> +    return sign(3*freememkb_diff + nrdomains_diff);

The reason you need what sign() does is that you need to convert from
double to int, I guess.

> +
> +    /*
> +     * Check if the domain has any CPU affinity. If not, try to build up one.
> +     * In case numa_place_domain() find at least a suitable candidate, it will
> +     * affect info->cpumap accordingly; if it does not, it just leaves it
> +     * as it is. This means (unless some weird error manifests) the subsequent
> +     * call to libxl_set_vcpuaffinity_all() will do the actual placement,
> +     * whatever that turns out to be.
> +     */
> +    if (libxl_bitmap_is_full(&info->cpumap)) {
> +        int rc = numa_place_domain(gc, info);
> +        if (rc)
> +            return rc;
> +    }

I guess it would be preferable to do this only if the bitmap was full
by default, so that setting the bitmap explicitly to all cpus still
works.

I'm not sure that that's essential to have, though.

> +            /*
> +             * Conditions are met, we can add this combination to the
> +             * NUMA placement candidates list. We first make sure there
> +             * is enough space in there, and then we initialize the new
> +             * candidate element with the node map corresponding to the
> +             * combination we are dealing with. Memory allocation for
> +             * expanding the array that hosts the list happens in chunks
> +             * equal to the number of NUMA nodes in the system (to
> +             * avoid allocating memory each and every time we find a
> +             * new candidate).
> +             */
> +            if (*nr_cndts == array_size)
> +                array_size += nr_nodes;
> +            GCREALLOC_ARRAY(new_cndts, array_size);

This part of the algorithm is quadratic in the number of combinations
divided by the number of nodes.  So the algorithm is
   O( (  C( nr_nodes, min_nodes ) / min_nodes  )^2 )
which is quite bad really.

At the very least this needs to be an exponential allocation, eg
  +                array_size += nr_nodes + array_size;

But if you didn't insist on creating the whole list and sorting it,
you would avoid this allocation entirely, wouldn't you ?

Should we bve worried that this algorithm will be too slow even if it
involves just
  O( C(nr_nodes,min_nodes) )
iterations ?

Thanks,
Ian.

[PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Dario Faggioli]

Hello again,

This is a new version (fixing a small bug) of this series:
<patchbomb.1341932613@Solace>, which in turn was a repost of what remained
un-committed of my NUMA placement series.

As already stated, the splat on 1-node-only systems has been resolved as per
Ian Campbell's suggestion (thanks!), and that was the only thing I'm aware that
was preventing from checking-in these last patches. While at it, I also
addressed all any other observation made during review, namely, typos and
rewording of some sentences, as well as an improved filtering solution for
avoiding generating a lot of potential duplicate placement candidates when
cpupool are being used.

All the patches have already been acked, although with some minor nits that
should be gone now.

If you have any further comment, I'm here to answer. :-)

Thanks and Regards,
Dario

[PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

If a domain does not have a VCPU affinity, try to pin it automatically to some
PCPUs. This is done taking into account the NUMA characteristics of the host.
In fact, we look for a combination of host's NUMA nodes with enough free memory
and number of PCPUs for the new domain, and pin it to the VCPUs of those nodes.

Once we know which ones, among all the possible combinations, represents valid
placement candidates for a domain, use some heuistics for deciding which is the
best. For instance, smaller candidates are considered to be better, both from
the domain's point of view (fewer memory spreading among nodes) and from the
system as a whole point of view (fewer memoy fragmentation).  In case of
candidates of equal sizes (i.e., with the same number of nodes), the amount of
free memory and the number of domain already assigned to their nodes are
considered. Very often, candidates with greater amount of memory are the one
we wants, as this is also good for keeping memory fragmentation under control.
However, if the difference in how much free memory two candidates have, the
number of assigned domains might be what decides which candidate wins.

This all happens internally to libxl, and no API for driving the mechanism is
provided for now. This matches what xend already does.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>

---
Changes from v3:
 * fixed a double free on non-NUMA systems (namely, when just 1 node exists).
 * Reword an ambiguous sentence in xl's man page, as requested during review.
 * Do no special case single node systems in libxl__get_numa_candidates(), to
   avoid generating a spurious warning on them.

Changes from v2:
 * lots of typos.
 * Clayfied some comments, as requested during (ijc's) review.
 * Added some more information/reference for the combination generation
   algorithm.
 * nodemap_to_nodes_cpus() function renamed to nodemap_to_nr_cpus().
 * libxl_bitmap_init() used to make sure we do not try to free random
   memory on failure paths of functions that allocates a libxl_bitmap.
 * Always invoke libxl__sort_numa_candidates(), even if we get there
   with just 1 candidate, as requested during review.
 * Simplified the if-s that check for input parameter consistency in
   libxl__get_numa_candidates() as requested during (gwd's) review.
 * Comparison function for candidates changed so that it now provides
   total ordering, as requested during review. It is still using FP
   arithmetic, though. Also I think that just putting the difference
   between the amount of free memory and between the number of assigned
   domains of two candidates in a single formula (after normalizing and
   weighting them) is both clear and effective enough.
 * Function definitions moved to a numa specific source file (libxl_numa.c),
   as suggested during review.


Changes from v1:
 * This patches incorporates the changes from both "libxl, xl: enable automatic
   placement of guests on NUMA nodes" and "libxl, xl: heuristics for reordering
   NUMA placement candidates" from v1.
 * The logic of the algorithm is basically the same as in v1, but the splitting
   of it in the various functions has been completely redesigned from scratch.
 * No public API for placement or candidate generation is now exposed,
   everything happens within libxl, as agreed during v1 review.
 * The relevant documentation have been moved near the actual functions and
   features. Also, the amount and (hopefully!) the quality of the documentation
   has been improved a lot, as requested.
 * All the comments about using the proper libxl facilities and helpers for
   allocations, etc., have been considered and applied.
 * This patch still bails out from NUMA optimizations if it find out cpupools
   are being utilized. It is next patch that makes the two things interact
   properly, as suggested during review.

diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
--- a/docs/man/xl.cfg.pod.5
+++ b/docs/man/xl.cfg.pod.5
@@ -111,8 +111,8 @@ created online and the remainder will be
 
 =item B<cpus="CPU-LIST">
 
-List of which cpus the guest is allowed to use. Default behavior is
-`all cpus`. A C<CPU-LIST> may be specified as follows:
+List of which cpus the guest is allowed to use. By default xl will (via
+libxl) pick some cpus (see below). A C<CPU-LIST> may be specified as follows:
 
 =over 4
 
@@ -132,6 +132,12 @@ run on cpu #3 of the host.
 
 =back
 
+If this option is not specified, libxl automatically tries to place the new
+domain on the host's NUMA nodes (provided the host has more than one NUMA
+node) by pinning it to the cpus of those nodes. A heuristic approach is
+utilized with the goals of maximizing performance for the domain and, at
+the same time, achieving efficient utilization of the host's CPUs and RAM.
+
 =item B<cpu_weight=WEIGHT>
 
 A domain with a weight of 512 will get twice as much CPU as a domain
diff --git a/tools/libxl/Makefile b/tools/libxl/Makefile
--- a/tools/libxl/Makefile
+++ b/tools/libxl/Makefile
@@ -66,7 +66,7 @@ LIBXL_LIBS += -lyajl -lm
 LIBXL_OBJS = flexarray.o libxl.o libxl_create.o libxl_dm.o libxl_pci.o \
 			libxl_dom.o libxl_exec.o libxl_xshelp.o libxl_device.o \
 			libxl_internal.o libxl_utils.o libxl_uuid.o \
-			libxl_json.o libxl_aoutils.o \
+			libxl_json.o libxl_aoutils.o libxl_numa.o \
 			libxl_save_callout.o _libxl_save_msgs_callout.o \
 			libxl_qmp.o libxl_event.o libxl_fork.o $(LIBXL_OBJS-y)
 LIBXL_OBJS += _libxl_types.o libxl_flask.o _libxl_types_internal.o
diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -98,6 +98,106 @@ out:
     return sched;
 }
 
+/* Subtract two values and translate the result in [0, 1] */
+static double normalized_diff(double a, double b)
+{
+#define max(a, b) (a > b ? a : b)
+    if (!a && a == b)
+        return 0.0;
+    return (a - b) / max(a, b);
+}
+
+/*
+ * The NUMA placement candidates are reordered according to the following
+ * heuristics:
+ *  - candidates involving fewer nodes come first. In case two (or
+ *    more) candidates span the same number of nodes,
+ *  - the amount of free memory and the number of domains assigned to the
+ *    candidates are considered. In doing that, candidates with greater
+ *    amount of free memory and fewer domains assigned to them are preferred,
+ *    with free memory "weighting" three times as much as number of domains.
+ */
+static int numa_cmpf(const void *v1, const void *v2)
+{
+    const libxl__numa_candidate *c1 = v1;
+    const libxl__numa_candidate *c2 = v2;
+#define sign(a) a > 0 ? 1 : a < 0 ? -1 : 0
+    double freememkb_diff = normalized_diff(c2->free_memkb, c1->free_memkb);
+    double nrdomains_diff = normalized_diff(c1->nr_domains, c2->nr_domains);
+
+    if (c1->nr_nodes != c2->nr_nodes)
+        return c1->nr_nodes - c2->nr_nodes;
+
+    return sign(3*freememkb_diff + nrdomains_diff);
+}
+
+/* The actual automatic NUMA placement routine */
+static int numa_place_domain(libxl__gc *gc, libxl_domain_build_info *info)
+{
+    int nr_candidates = 0;
+    libxl__numa_candidate *candidates = NULL;
+    libxl_bitmap candidate_nodemap;
+    libxl_cpupoolinfo *pinfo;
+    int nr_pools, rc = 0;
+    uint32_t memkb;
+
+    libxl_bitmap_init(&candidate_nodemap);
+
+    /* First of all, if cpupools are in use, better not to mess with them */
+    pinfo = libxl_list_cpupool(CTX, &nr_pools);
+    if (!pinfo)
+        return ERROR_FAIL;
+    if (nr_pools > 1) {
+        LOG(NOTICE, "skipping NUMA placement as cpupools are in use");
+        goto out;
+    }
+
+    rc = libxl_domain_need_memory(CTX, info, &memkb);
+    if (rc)
+        goto out;
+    if (libxl_node_bitmap_alloc(CTX, &candidate_nodemap, 0)) {
+        rc = ERROR_FAIL;
+        goto out;
+    }
+
+    /* Find all the candidates with enough free memory and at least
+     * as much pcpus as the domain has vcpus.  */
+    rc = libxl__get_numa_candidates(gc, memkb, info->max_vcpus, 0, 0,
+                                    &candidates, &nr_candidates);
+    if (rc)
+        goto out;
+
+    LOG(DETAIL, "%d NUMA placement candidates found", nr_candidates);
+
+    /* No suitable placement candidates. We just return without touching the
+     * domain's info->cpumap. It will have affinity with all nodes/cpus. */
+    if (nr_candidates == 0) {
+        LOG(NOTICE, "NUMA placement failed, performance might be affected");
+        goto out;
+    }
+
+    /* Bring the best candidate in front of the list --> candidates[0] */
+    libxl__sort_numa_candidates(candidates, nr_candidates, numa_cmpf);
+
+    /*
+     * At this point, the first candidate in the array is the one we want.
+     * Go for it by mapping its node map to the domain's info->cpumap.
+     */
+    libxl__numa_candidate_get_nodemap(gc, &candidates[0], &candidate_nodemap);
+    rc = libxl_nodemap_to_cpumap(CTX, &candidate_nodemap, &info->cpumap);
+    if (rc)
+        goto out;
+
+    LOG(DETAIL, "NUMA placement candidate with %d nodes, %d cpus and "
+                "%"PRIu32" KB free selected", candidates[0].nr_nodes,
+                candidates[0].nr_cpus, candidates[0].free_memkb / 1024);
+
+ out:
+    libxl_bitmap_dispose(&candidate_nodemap);
+    libxl_cpupoolinfo_list_free(pinfo, nr_pools);
+    return rc;
+}
+
 int libxl__build_pre(libxl__gc *gc, uint32_t domid,
               libxl_domain_build_info *info, libxl__domain_build_state *state)
 {
@@ -107,7 +207,22 @@ int libxl__build_pre(libxl__gc *gc, uint
     uint32_t rtc_timeoffset;
 
     xc_domain_max_vcpus(ctx->xch, domid, info->max_vcpus);
+
+    /*
+     * Check if the domain has any CPU affinity. If not, try to build up one.
+     * In case numa_place_domain() find at least a suitable candidate, it will
+     * affect info->cpumap accordingly; if it does not, it just leaves it
+     * as it is. This means (unless some weird error manifests) the subsequent
+     * call to libxl_set_vcpuaffinity_all() will do the actual placement,
+     * whatever that turns out to be.
+     */
+    if (libxl_bitmap_is_full(&info->cpumap)) {
+        int rc = numa_place_domain(gc, info);
+        if (rc)
+            return rc;
+    }
     libxl_set_vcpuaffinity_all(ctx, domid, info->max_vcpus, &info->cpumap);
+
     xc_domain_setmaxmem(ctx->xch, domid, info->target_memkb + LIBXL_MAXMEM_CONSTANT);
     if (info->type == LIBXL_DOMAIN_TYPE_PV)
         xc_domain_set_memmap_limit(ctx->xch, domid,
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -2216,6 +2216,140 @@ static inline void libxl__ctx_unlock(lib
 #define CTX_LOCK (libxl__ctx_lock(CTX))
 #define CTX_UNLOCK (libxl__ctx_unlock(CTX))
 
+/*
+ * Automatic NUMA placement
+ *
+ * These functions and data structures deal with the initial placement of a
+ * domain onto the host NUMA nodes.
+ *
+ * The key concept here is the one of "NUMA placement candidate", which is
+ * basically a set of nodes whose characteristics have been successfully
+ * checked against some specific requirements. More precisely, a candidate is
+ * the nodemap associated with one of the possible subset of the host NUMA
+ * nodes providing a certain amount of free memory, or a given number of cpus,
+ * or even both (depending in what the caller wants). For convenience of use,
+ * some of this information are stored within the candidate itself, instead of
+ * always being dynamically computed. A single node is a valid placement
+ * candidate, but it is also possible for a candidate to contain all the nodes
+ * of the host. The fewer nodes there are in a candidate, the better
+ * performance a domain placed onto it should get. For instance, looking for a
+ * numa candidates with 2GB of free memory means we want all the possible
+ * subsets of the host NUMA nodes with, cumulatively, at least 2GB of free
+ * memory. That could be possible by just using one particular node, or may
+ * require more nodes, depending on the characteristics of the host, on how
+ * many domains have been created already, on how big they are, etc.
+ *
+ * The intended usage is as follows:
+ *  1. by, fist of all, calling libxl__get_numa_candidates(), and specifying
+ *     the proper constraints to it (e.g., the amount of memory a domain need
+ *     as the minimum amount of free memory for the candidates) one can build
+ *     up a whole set of suitable placing alternatives for a domain;
+ *  2. after that, one specific candidate should be chosen. That can happen
+ *     by looking at their various characteristics;
+ *  3. the chosen candidate's nodemap should be utilized for computing the
+ *     actual affinity of the domain which, given the current NUMA support
+ *     in the hypervisor, is what determines the placement of the domain's
+ *     vcpus and memory.
+ *
+ * To make phase 2 even easier, a sorting helper function for the list of
+ * candidates is provided in the form of libxl__sort_numa_candidates(). The
+ * only that is needed is defining a comparison function, containing the
+ * criteria for deciding, given two candidates, which one is 'better'.
+ * Depending on how the comparison function is defined, the best candidate
+ * (where, of course, best is defined with respect to the heuristics
+ * implemented in the comparison function itself, libxl__numa_candidate_cmpf())
+ * could become the first or the last element of the list.
+ *
+ * Summarizing, achieving automatic NUMA placement is just a matter of
+ * obtaining the list of suitable placement candidates, perhaps asking for each
+ * of them to provide at least the amount of memory the domain needs. After
+ * that just implement a comparison function by means of the various helpers
+ * retrieving the relevant information about the candidates themselves.
+ * Finally, call the sorting helper function and use the candidate that became
+ * (typically) the first element of the list for determining the domain's
+ * affinity.
+ */
+
+typedef struct {
+    int nr_cpus, nr_nodes;
+    int nr_domains;
+    uint32_t free_memkb;
+    libxl_bitmap nodemap;
+} libxl__numa_candidate;
+
+/*
+ * This generates the list of NUMA placement candidates satisfying some
+ * specific conditions. If min_nodes and/or max_nodes are not 0, their value is
+ * used to determine the minimum and maximum number of nodes that are allow to
+ * be present in each candidate. If min_nodes and/or max_nodes are 0, the
+ * minimum and maximum number of nodes to be used are automatically selected by
+ * the implementation (and that will likely be just 1 node for the minimum and
+ * the total number of existent nodes for the maximum). Re min_free_memkb and
+ * min_cpu, if not 0, it means the caller only wants candidates with at
+ * least that amount of free memory and that number of cpus, respectively. If
+ * min_free_memkb and/or min_cpus are 0, the candidates' free memory and number
+ * of cpus won't be checked at all, which means a candidate will always be
+ * considered suitable wrt the specific constraint.  cndts is where the list of
+ * exactly nr_cndts candidates is returned. Note that, in case no candidates
+ * are found at all, the function returns successfully, but with nr_cndts equal
+ * to zero.
+ */
+_hidden int libxl__get_numa_candidates(libxl__gc *gc,
+                                uint32_t min_free_memkb, int min_cpus,
+                                int min_nodes, int max_nodes,
+                                libxl__numa_candidate *cndts[], int *nr_cndts);
+
+/* Initialization, allocation and deallocation for placement candidates */
+static inline void libxl__numa_candidate_init(libxl__numa_candidate *cndt)
+{
+    cndt->free_memkb = 0;
+    cndt->nr_cpus = cndt->nr_nodes = cndt->nr_domains = 0;
+    libxl_bitmap_init(&cndt->nodemap);
+}
+
+static inline int libxl__numa_candidate_alloc(libxl__gc *gc,
+                                              libxl__numa_candidate *cndt)
+{
+    return libxl_node_bitmap_alloc(CTX, &cndt->nodemap, 0);
+}
+static inline void libxl__numa_candidate_dispose(libxl__numa_candidate *cndt)
+{
+    libxl_bitmap_dispose(&cndt->nodemap);
+}
+static inline void libxl__numacandidate_list_free(libxl__numa_candidate *cndts,
+                                                  int nr_cndts)
+{
+    int i;
+
+    for (i = 0; i < nr_cndts; i++)
+        libxl__numa_candidate_dispose(&cndts[i]);
+    free(cndts);
+}
+
+/* Retrieve (in nodemap) the node map associated to placement candidate cndt */
+static inline
+void libxl__numa_candidate_get_nodemap(libxl__gc *gc,
+                                       const libxl__numa_candidate *cndt,
+                                       libxl_bitmap *nodemap)
+{
+    libxl_bitmap_copy(CTX, nodemap, &cndt->nodemap);
+}
+/* Set the node map of placement candidate cndt to match nodemap */
+static inline
+void libxl__numa_candidate_put_nodemap(libxl__gc *gc,
+                                       libxl__numa_candidate *cndt,
+                                       const libxl_bitmap *nodemap)
+{
+    libxl_bitmap_copy(CTX, &cndt->nodemap, nodemap);
+}
+
+/* Signature for the comparison function between two candidates c1 and c2 */
+typedef int (*libxl__numa_candidate_cmpf)(const void *v1, const void *v2);
+/* Sort the list of candidates in cndts (an array with nr_cndts elements in
+ * it) using cmpf for comparing two candidates. Uses libc's qsort(). */
+_hidden void libxl__sort_numa_candidates(libxl__numa_candidate cndts[],
+                                         int nr_cndts,
+                                         libxl__numa_candidate_cmpf cmpf);
 
 /*
  * Inserts "elm_new" into the sorted list "head".
diff --git a/tools/libxl/libxl_numa.c b/tools/libxl/libxl_numa.c
new file mode 100644
--- /dev/null
+++ b/tools/libxl/libxl_numa.c
@@ -0,0 +1,377 @@
+/*
+ * Copyright (C) 2012      Citrix Ltd.
+ * Author Dario Faggioli <dario.faggioli@citrix.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published
+ * by the Free Software Foundation; version 2.1 only. with the special
+ * exception on linking described in file LICENSE.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Lesser General Public License for more details.
+ */
+
+#include "libxl_osdeps.h" /* must come before any other headers */
+
+#include <glob.h>
+
+#include "libxl_internal.h"
+
+/*
+ * What follows are helpers for generating all the k-combinations
+ * without repetitions of a set S with n elements in it. Formally
+ * speaking, they are subsets of k distinct elements of S and, if
+ * S is n elements big, the number of k-combinations is equal to
+ * the binomial coefficient (n k), which means we get exactly
+ * n!/(k! * (n - k)!) subsets, all of them with k elements.
+ *
+ * The various subset are generated one after the other by calling
+ * comb_init() first, and, after that, comb_next()
+ * (n k)-1 times. An iterator is used to store the current status
+ * of the whole generation operation (i.e., basically, the last
+ * combination that has been generated). As soon as all
+ * combinations have been generated, comb_next() will
+ * start returning 0 instead of 1. It is of course important that
+ * the same instance of the iterator and the same values for
+ * n and k are used for each call. If that doesn't happen, the
+ * result is unspecified.
+ *
+ * The algorithm is a well known one (see, for example, D. Knuth's "The
+ * Art of Computer Programming - Volume 4, Fascicle 3" and it produces
+ * the combinations in such a way that they (well, more precisely,
+ * their indexes it the array/map representing the set) come with
+ * lexicographic ordering.
+ *
+ * For example, with n = 5 and k = 3, calling comb_init()
+ * will generate { 0, 1, 2 }, while subsequent valid calls to
+ * comb_next() will produce the following:
+ * { { 0, 1, 3 }, { 0, 1, 4 },
+ *   { 0, 2, 3 }, { 0, 2, 4 }, { 0, 3, 4 },
+ *   { 1, 2, 3 }, { 1, 2, 4 }, { 1, 3, 4 },
+ *   { 2, 3, 4 } }
+ *
+ * This is used by the automatic NUMA placement logic below.
+ */
+typedef int* comb_iter_t;
+
+static int comb_init(libxl__gc *gc, comb_iter_t *it, int n, int k)
+{
+    comb_iter_t new_iter;
+    int i;
+
+    if (n < k)
+        return 0;
+
+    /* First set is always { 0, 1, 2, ..., k-1 } */
+    GCNEW_ARRAY(new_iter, k);
+    for (i = 0; i < k; i++)
+        new_iter[i] = i;
+
+    *it = new_iter;
+    return 1;
+}
+
+static int comb_next(comb_iter_t it, int n, int k)
+{
+    int i;
+
+    /*
+     * The idea here is to find the leftmost element from where
+     * we should start incrementing the indexes of the iterator.
+     * This means looking for the highest index that can be increased
+     * while still producing value smaller than n-1. In the example
+     * above, when dealing with { 0, 1, 4 }, such an element is the
+     * second one, as the third is already equal to 4 (which actually
+     * is n-1).
+     * Once we found from where to start, we increment that element
+     * and override the right-hand rest of the iterator with its
+     * successors, thus achieving lexicographic ordering.
+     *
+     * Regarding the termination of the generation process, when we
+     * manage in bringing n-k at the very first position of the iterator,
+     * we know that is the last valid combination ( { 2, 3, 4 }, with
+     * n - k = 5 - 2 = 2, in the example above), and thus we start
+     * returning 0 as soon as we cross that border.
+     */
+    for (i = k - 1; it[i] == n - k + i; i--) {
+        if (i <= 0)
+            return 0;
+    }
+    for (it[i]++, i++; i < k; i++)
+        it[i] = it[i - 1] + 1;
+    return 1;
+}
+
+/* NUMA automatic placement (see libxl_internal.h for details) */
+
+/*
+ * This function turns a k-combination iterator into a node map.
+ * This means the bits in the node map corresponding to the indexes
+ * of the given combination are the ones that will be set.
+ * For example, if the iterator represents the combination { 0, 2, 4},
+ * the node map will have bits #0, #2 and #4 set.
+ */
+static void comb_get_nodemap(comb_iter_t it, libxl_bitmap *nodemap, int k)
+{
+    int i;
+
+    libxl_bitmap_set_none(nodemap);
+    for (i = 0; i < k; i++)
+        libxl_bitmap_set(nodemap, it[i]);
+}
+
+/* Retrieve the number of cpus that the nodes that are part of the nodemap
+ * span. */
+static int nodemap_to_nr_cpus(libxl_cputopology *tinfo, int nr_cpus,
+                              const libxl_bitmap *nodemap)
+{
+    int i, nodes_cpus = 0;
+
+    for (i = 0; i < nr_cpus; i++) {
+        if (libxl_bitmap_test(nodemap, tinfo[i].node))
+            nodes_cpus++;
+    }
+    return nodes_cpus;
+}
+
+/* Retrieve the amount of free memory within the nodemap */
+static uint32_t nodemap_to_free_memkb(libxl_numainfo *ninfo,
+                                      libxl_bitmap *nodemap)
+{
+    uint32_t free_memkb = 0;
+    int i;
+
+    libxl_for_each_set_bit(i, *nodemap)
+        free_memkb += ninfo[i].free / 1024;
+
+    return free_memkb;
+}
+
+/* Retrieve the number of domains that can potentially run on the cpus
+ * the nodes that are part of the nodemap. */
+static int nodemap_to_nr_domains(libxl__gc *gc, libxl_cputopology *tinfo,
+                                 const libxl_bitmap *nodemap)
+{
+    libxl_dominfo *dinfo = NULL;
+    libxl_bitmap dom_nodemap;
+    int nr_doms, nr_cpus;
+    int nr_domains = 0;
+    int i, j, k;
+
+    dinfo = libxl_list_domain(CTX, &nr_doms);
+    if (dinfo == NULL)
+        return ERROR_FAIL;
+
+    if (libxl_node_bitmap_alloc(CTX, &dom_nodemap, 0) < 0) {
+        libxl_dominfo_list_free(dinfo, nr_doms);
+        return ERROR_FAIL;
+    }
+
+    for (i = 0; i < nr_doms; i++) {
+        libxl_vcpuinfo *vinfo;
+        int nr_vcpus;
+
+        vinfo = libxl_list_vcpu(CTX, dinfo[i].domid, &nr_vcpus, &nr_cpus);
+        if (vinfo == NULL)
+            continue;
+
+        libxl_bitmap_set_none(&dom_nodemap);
+        for (j = 0; j < nr_vcpus; j++) {
+            libxl_for_each_set_bit(k, vinfo[j].cpumap)
+                libxl_bitmap_set(&dom_nodemap, tinfo[k].node);
+        }
+
+        libxl_for_each_set_bit(j, dom_nodemap) {
+            if (libxl_bitmap_test(nodemap, j)) {
+                nr_domains++;
+                break;
+            }
+        }
+
+        libxl_vcpuinfo_list_free(vinfo, nr_vcpus);
+    }
+
+    libxl_bitmap_dispose(&dom_nodemap);
+    libxl_dominfo_list_free(dinfo, nr_doms);
+    return nr_domains;
+}
+
+/*
+ * This function tries to figure out if the host has a consistent number
+ * of cpus along all its NUMA nodes. In fact, if that is the case, we can
+ * calculate the minimum number of nodes needed for a domain by just
+ * dividing its total number of vcpus by this value computed here.
+ * However, we are not allowed to assume that all the nodes have the
+ * same number of cpus. Therefore, in case discrepancies among different
+ * nodes are found, this function just returns 0, for the caller to know
+ * it shouldn't rely on this 'optimization', and sort out things in some
+ * other way (most likely, just start trying with candidates with just
+ * one node).
+ */
+static int cpus_per_node_count(libxl_cputopology *tinfo, int nr_cpus,
+                               libxl_numainfo *ninfo, int nr_nodes)
+{
+    int cpus_per_node = 0;
+    int j, i;
+
+    /* This makes sense iff # of PCPUs is the same for all nodes */
+    for (j = 0; j < nr_nodes; j++) {
+        int curr_cpus = 0;
+
+        for (i = 0; i < nr_cpus; i++) {
+            if (tinfo[i].node == j)
+                curr_cpus++;
+        }
+        /* So, if the above does not hold, turn the whole thing off! */
+        cpus_per_node = cpus_per_node == 0 ? curr_cpus : cpus_per_node;
+        if (cpus_per_node != curr_cpus)
+            return 0;
+    }
+    return cpus_per_node;
+}
+
+/* Get all the placement candidates satisfying some specific conditions */
+int libxl__get_numa_candidates(libxl__gc *gc,
+                               uint32_t min_free_memkb, int min_cpus,
+                               int min_nodes, int max_nodes,
+                               libxl__numa_candidate *cndts[], int *nr_cndts)
+{
+    libxl__numa_candidate *new_cndts = NULL;
+    libxl_cputopology *tinfo = NULL;
+    libxl_numainfo *ninfo = NULL;
+    int nr_nodes = 0, nr_cpus = 0;
+    libxl_bitmap nodemap;
+    int array_size, rc;
+
+    libxl_bitmap_init(&nodemap);
+
+    /* Get platform info and prepare the map for testing the combinations */
+    ninfo = libxl_get_numainfo(CTX, &nr_nodes);
+    if (ninfo == NULL)
+        return ERROR_FAIL;
+
+    tinfo = libxl_get_cpu_topology(CTX, &nr_cpus);
+    if (tinfo == NULL) {
+        rc = ERROR_FAIL;
+        goto out;
+    }
+
+    rc = libxl_node_bitmap_alloc(CTX, &nodemap, 0);
+    if (rc)
+        goto out;
+
+    /*
+     * If the minimum number of NUMA nodes is not explicitly specified
+     * (i.e., min_nodes == 0), we try to figure out a sensible number of nodes
+     * from where to start generating candidates, if possible (or just start
+     * from 1 otherwise). The maximum number of nodes should not exceed the
+     * number of existent NUMA nodes on the host, or the candidate generation
+     * won't work properly.
+     */
+    if (!min_nodes) {
+        int cpus_per_node;
+
+        cpus_per_node = cpus_per_node_count(tinfo, nr_cpus, ninfo, nr_nodes);
+        if (cpus_per_node == 0)
+            min_nodes = 1;
+        else
+            min_nodes = (min_cpus + cpus_per_node - 1) / cpus_per_node;
+    }
+    if (min_nodes > nr_nodes)
+        min_nodes = nr_nodes;
+    if (!max_nodes || max_nodes > nr_nodes)
+        max_nodes = nr_nodes;
+    if (min_nodes > max_nodes) {
+        rc = ERROR_INVAL;
+        goto out;
+    }
+
+    /* Initialize the local storage for the combinations */
+    *nr_cndts = 0;
+    array_size = nr_nodes;
+    GCNEW_ARRAY(new_cndts, array_size);
+
+    /* Generate all the combinations of any size from min_nodes to
+     * max_nodes (see comb_init() and comb_next()). */
+    while (min_nodes <= max_nodes) {
+        comb_iter_t comb_iter;
+        int comb_ok;
+
+        /*
+	 * And here it is. Each step of this cycle generates a combination of
+	 * nodes as big as min_nodes mandates.  Each of these combinations is
+	 * checked against the constraints provided by the caller (namely,
+	 * amount of free memory and number of cpus) and it becomes an actual
+	 * placement candidate iff it passes the check.
+         */
+        for (comb_ok = comb_init(gc, &comb_iter, nr_nodes, min_nodes); comb_ok;
+             comb_ok = comb_next(comb_iter, nr_nodes, min_nodes)) {
+            uint32_t nodes_free_memkb;
+            int nodes_cpus;
+
+            comb_get_nodemap(comb_iter, &nodemap, min_nodes);
+
+            /* If there is not enough memory in this combination, skip it
+             * and go generating the next one... */
+            nodes_free_memkb = nodemap_to_free_memkb(ninfo, &nodemap);
+            if (min_free_memkb && nodes_free_memkb < min_free_memkb)
+                continue;
+
+            /* And the same applies if this combination is short in cpus */
+            nodes_cpus = nodemap_to_nr_cpus(tinfo, nr_cpus, &nodemap);
+            if (min_cpus && nodes_cpus < min_cpus)
+                continue;
+
+            /*
+             * Conditions are met, we can add this combination to the
+             * NUMA placement candidates list. We first make sure there
+             * is enough space in there, and then we initialize the new
+             * candidate element with the node map corresponding to the
+             * combination we are dealing with. Memory allocation for
+             * expanding the array that hosts the list happens in chunks
+             * equal to the number of NUMA nodes in the system (to
+             * avoid allocating memory each and every time we find a
+             * new candidate).
+             */
+            if (*nr_cndts == array_size)
+                array_size += nr_nodes;
+            GCREALLOC_ARRAY(new_cndts, array_size);
+
+            libxl__numa_candidate_alloc(gc, &new_cndts[*nr_cndts]);
+            libxl__numa_candidate_put_nodemap(gc, &new_cndts[*nr_cndts],
+                                              &nodemap);
+            new_cndts[*nr_cndts].nr_domains =
+                                    nodemap_to_nr_domains(gc, tinfo, &nodemap);
+            new_cndts[*nr_cndts].free_memkb = nodes_free_memkb;
+            new_cndts[*nr_cndts].nr_nodes = min_nodes;
+            new_cndts[*nr_cndts].nr_cpus = nodes_cpus;
+
+            LOG(DEBUG, "NUMA placement candidate #%d found: nr_nodes=%d, "
+                       "nr_cpus=%d, free_memkb=%"PRIu32"", *nr_cndts,
+                       min_nodes, new_cndts[*nr_cndts].nr_cpus,
+                       new_cndts[*nr_cndts].free_memkb / 1024);
+
+            (*nr_cndts)++;
+        }
+        min_nodes++;
+    }
+
+    *cndts = new_cndts;
+ out:
+    libxl_bitmap_dispose(&nodemap);
+    libxl_cputopology_list_free(tinfo, nr_cpus);
+    libxl_numainfo_list_free(ninfo, nr_nodes);
+    return rc;
+}
+
+void libxl__sort_numa_candidates(libxl__numa_candidate cndts[], int nr_cndts,
+                                 libxl__numa_candidate_cmpf cmpf)
+{
+    /* Reorder candidates (see the comparison function for
+     * the details on the heuristics) */
+    qsort(cndts, nr_cndts, sizeof(cndts[0]), cmpf);
+}
+
+

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Jackson]

Dario Faggioli writes ("[PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
> If a domain does not have a VCPU affinity, try to pin it
> automatically to some PCPUs. This is done taking into account the
> NUMA characteristics of the host.  In fact, we look for a
> combination of host's NUMA nodes with enough free memory and number
> of PCPUs for the new domain, and pin it to the VCPUs of those nodes.

Ian Jackson writes ("Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
...
> This part of the algorithm is quadratic in the number of combinations
> divided by the number of nodes.  So the algorithm is
>    O( (  C( nr_nodes, min_nodes ) / min_nodes  )^2 )
> which is quite bad really.
> 
> At the very least this needs to be an exponential allocation, eg
>   +                array_size += nr_nodes + array_size;
> 
> But if you didn't insist on creating the whole list and sorting it,
> you would avoid this allocation entirely, wouldn't you ?
> 
> Should we bve worried that this algorithm will be too slow even if it
> involves just
>   O( C(nr_nodes,min_nodes) )
> iterations ?

I've been told that boxes with 16 NUMA nodes are coming out about
now.

This algorithm will start to have unreasonable performance with
anything bigger than that.  Eg choosing 16 NUMA nodes out of 32 would
involve searching 601,080,390 combinations.  32 out of 64 gives 10^18.

So I think we need a different approach :-(.

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Campbell]

On Tue, 2012-07-17 at 19:04 +0100, Ian Jackson wrote:
> Dario Faggioli writes ("[PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
> > If a domain does not have a VCPU affinity, try to pin it
> > automatically to some PCPUs. This is done taking into account the
> > NUMA characteristics of the host.  In fact, we look for a
> > combination of host's NUMA nodes with enough free memory and number
> > of PCPUs for the new domain, and pin it to the VCPUs of those nodes.
> 
> Ian Jackson writes ("Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
> ...
> > This part of the algorithm is quadratic in the number of combinations
> > divided by the number of nodes.  So the algorithm is
> >    O( (  C( nr_nodes, min_nodes ) / min_nodes  )^2 )
> > which is quite bad really.
> > 
> > At the very least this needs to be an exponential allocation, eg
> >   +                array_size += nr_nodes + array_size;
> > 
> > But if you didn't insist on creating the whole list and sorting it,
> > you would avoid this allocation entirely, wouldn't you ?
> > 
> > Should we bve worried that this algorithm will be too slow even if it
> > involves just
> >   O( C(nr_nodes,min_nodes) )
> > iterations ?
> 
> I've been told that boxes with 16 NUMA nodes are coming out about
> now.
> 
> This algorithm will start to have unreasonable performance with
> anything bigger than that.  Eg choosing 16 NUMA nodes out of 32 would
> involve searching 601,080,390 combinations.  32 out of 64 gives 10^18.
> 
> So I think we need a different approach :-(.

If 16 are coming out now then 32 nodes are at least a little in the
future, and 16 won't be common just yet.

The 4.2 freeze is starting to drag on and reworking this significantly
at this stage seems likely to take quite a while -- can we perhaps go
with the existing patches (with the known scalability limitation and
workaround of setting the explicit cpu affinity documented in the
release notes) and revisit for 4.3 and/or 4.2.1?

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 2250 bytes --]

On Tue, 2012-07-17 at 21:23 +0100, Ian Campbell wrote: 
> > I've been told that boxes with 16 NUMA nodes are coming out about
> > now.
> > 
> > This algorithm will start to have unreasonable performance with
> > anything bigger than that.  Eg choosing 16 NUMA nodes out of 32 would
> > involve searching 601,080,390 combinations.  32 out of 64 gives 10^18.
> > 
> > So I think we need a different approach :-(.
> 
> If 16 are coming out now then 32 nodes are at least a little in the
> future, and 16 won't be common just yet.
> 
FWIW, I agree. Consider that 16 nodes systems exists right now only
because AMD has an architecture where they're putting 2 nodes on one
socket, which is something pretty special that might require some
special thinking and consideration already. The point being assuming
future 16, 32 and 64 nodes systems to just be like the machines we see
now, only bigger, won't be something letting me sleep at night... :-D

> The 4.2 freeze is starting to drag on and reworking this significantly
> at this stage seems likely to take quite a while -- can we perhaps go
> with the existing patches (with the known scalability limitation and
> workaround of setting the explicit cpu affinity documented in the
> release notes) and revisit for 4.3 and/or 4.2.1?
> 
That would be fine by me.

I already said in the other e-mail how I think we can and should address
the potential scalability issue, at whatever time you think it's best.
I'm fine with putting a limit on the maximum number of nodes a guest
should span on (and that being more a feature than a workaround, and can
definitely go in the release notes) wherever you like.

If it's just a matter of putting a LIBXL__MAX_GUEST_NODES (or whatever)
macro in libxl_internal.h, and enforce that, I'm fine doing it right
now, as I'm reposting these last three patches either way. We just need
to agree on a value to write besides it. :-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Jackson]

Ian Campbell writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
> The 4.2 freeze is starting to drag on and reworking this significantly
> at this stage seems likely to take quite a while

Yes.

> -- can we perhaps go
> with the existing patches (with the known scalability limitation and
> workaround of setting the explicit cpu affinity documented in the
> release notes) and revisit for 4.3 and/or 4.2.1?

I think the existing behaviour of these patches is unreleasable.  In a
couple of years' time people will try to run Xen 4.2 on larger
machines and the result will be a hang (perhaps of their entire
management stack) when they try to create a large guest.

At the very least it should have a safety catch so that it disables
itself, or something.

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 3856 bytes --]

On Tue, 2012-07-17 at 19:04 +0100, Ian Jackson wrote: 
> I've been told that boxes with 16 NUMA nodes are coming out about
> now.
> 
> This algorithm will start to have unreasonable performance with
> anything bigger than that.  Eg choosing 16 NUMA nodes out of 32 would
> involve searching 601,080,390 combinations.  32 out of 64 gives 10^18.
> 
Ok, when designing this thing, I knew it would turn out to be the most
computationally complex part of the whole NUMA thing (hopefully!).

That was on purpose, mainly because this is the *first* (from the
guest's lifetime POV) of the various others NUMA-aware related
heuristics that will come, and the only one that will run userspace and
far from critical paths. That's why, when the suggestion to consider
"all the possible combination of ..." came, during one round of the
review, I liked it very much and went for it.

_That_all_being_said_, the algorithm takes some input parameters, the
most important of which is probably max_nodes. It should contain the
maximum number of nodes one wants his guest to span. I worked really
hard for it not to not be just hardcoded to 1, as that would mean we
were giving up trying to improve performances for guests not fitting in
just one node. However, it is not written in stone that it must be free
to range from 1 to #_of_nodes. That is sure possible, with the current
NUMA systems, and here they are the number of combinations we get for 4,
8 and 16 nodes (n), respectively: 
$ echo "combs=0; n=4; for k=1:4 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  15
 $ echo "combs=0; n=8; for k=1:8 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  255
 $ echo "combs=0; n=16; for k=1:16 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  65535

However, it might well be decided that it is sane to limit max_nodes to,
say, 4, which would mean we'll be trying our best to boost the
performances of all guests that fits within 4 nodes (which is a lot
better than just one!), which would result in these numbers, for nodes
(n) equal to 4, 8, 16, 32 and 64: 
$ echo "combs=0; n=4; for k=1:4 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  15
 $ echo "combs=0; n=8; for k=1:4 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  162
 $ echo "combs=0; n=16; for k=1:4 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  2516
 $ echo "combs=0; n=32; for k=1:4 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  41448
 $ echo "combs=0; n=64; for k=1:4 combs=combs+bincoeff(n,k); endfor; combs" | octave -q
  combs =  679120

Moreover, as another input parameter is a nodemask, telling the
algorithm not only how many but also _which_ nodes it should consider,
one could imagine to have some multi-tier form of the algorithm itself,
or, perhaps, implementing a more abstract and coarse partitioning
heuristics among group of nodes (which will better be done as soon as
we'll see how 32 and 64 nodes system will actually look like :-D) and
then run the current algorithm only on these groups, with numbers that
would then look just like above.

So, to summarize, I can't be farther than saying the current algorithm
is perfect or fast. Rather, I'm definitely saying it has the advantage
of not branching out the problem space too much aggressively at this
early stage, and it is flexible enough for avoiding getting unreasonable
performances, or at least, it looks like that to me, I guess. :-)

Thoughts?

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

On Wed, 2012-07-18 at 02:22 +0200, Dario Faggioli wrote:
> Moreover, as another input parameter is a nodemask, telling the
> algorithm not only how many but also _which_ nodes it should consider,
> one could imagine to have some multi-tier form of the algorithm itself,
> or, perhaps, implementing a more abstract and coarse partitioning
> heuristics among group of nodes (which will better be done as soon as
> we'll see how 32 and 64 nodes system will actually look like :-D) and
> then run the current algorithm only on these groups, with numbers that
> would then look just like above.
> 
Alternatively (or maybe not even that much 'alternatively' :-)), we can
try turning it into something a bit more greedy, e.g., searching for
available placement solutions just as big as min_nodes says (typically
1) as a first step (complexity linear with the number of nodes).

After that, if we did not find anything, we could have some sort of
config switch telling us if we should continue, and maybe investigating
candidates of sizes in the range [min_nodes+1,LIBXL__MAX_GUEST_NODES]. 

At which point, if we haven't found anything yet, we could just give up
or, perhaps, think and implement in the near future a more coarse and
less precise mechanism for proceeding further.

Implementing something like that on top of what I have now (which I'll
resend asap, as soon as I've addressed the minor comments I got) during
4.3 should be pretty easy. Also, if we decide to just statically limit
max_nodes<=LIBXL__MAX_GUEST_NODES (or whatever we could call it) to
something (I was proposing 4 in the other e-mail) for 4.2, backporting
to it will be piece of cake as well. :-)

Thanks again and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Campbell]

On Wed, 2012-07-18 at 01:22 +0100, Dario Faggioli wrote:

> So, to summarize, I can't be farther than saying the current algorithm
> is perfect or fast. Rather, I'm definitely saying it has the advantage
> of not branching out the problem space too much aggressively at this
> early stage, and it is flexible enough for avoiding getting unreasonable
> performances, or at least, it looks like that to me, I guess. :-)

I'm not quite following what you are trying to say with all this.

Are you saying that the code as most recently submitted does not have
the O( C(nr_nodes,min_nodes) ) property which Ian J suggests, which
leads to needing to process hundreds of millions of combinations for a
16 / 32 node guest? Or are you saying that it does have this property
but it doesn't matter? Or are you saying it does have this property but
you intend to ameliorate it somehow? If the latter then how and for
which release(s) (4.2, 4.3 or 4.2.1)?

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 2651 bytes --]

On Wed, 2012-07-18 at 10:13 +0100, Ian Campbell wrote:
> On Wed, 2012-07-18 at 01:22 +0100, Dario Faggioli wrote:
> 
> > So, to summarize, I can't be farther than saying the current algorithm
> > is perfect or fast. Rather, I'm definitely saying it has the advantage
> > of not branching out the problem space too much aggressively at this
> > early stage, and it is flexible enough for avoiding getting unreasonable
> > performances, or at least, it looks like that to me, I guess. :-)
> 
> I'm not quite following what you are trying to say with all this.
> 
Oh, I see, sorry. :-(

> Are you saying that the code as most recently submitted does not have
> the O( C(nr_nodes,min_nodes) ) property which Ian J suggests, which
> leads to needing to process hundreds of millions of combinations for a
> 16 / 32 node guest? 
>
No, I'm not. If you want to place a 16 (32) nodes guest on a 32 (64)
nodes host, it is as bad as he says. However, I'm also saying that on
<=16 nodes host it is able o provide very good placement at a reasonable
cost and that, if we limit the number of guest nodes (e.g., up to 4) it
is able to do the same even on 32 or 64 nodes hosts.

> Or are you saying that it does have this property
> but it doesn't matter? 
>
A lot of people is thinking that trying to address the problem of
placing guests that are bigger (or anyway, guests that does not fit) in
just 1 node should not be even considered. I'm saying that it should and
am also proposing an algorithm that is able to provide good placement
performances for guests that spans up to 4 nodes even on future 32 or 64
nodes systems.

So basically I'm saying that it does have that property but that is
bringing benefits right now, and we have all the knobs to limit its
potential nasty effects for future (bigger) system at any time.

> Or are you saying it does have this property but
> you intend to ameliorate it somehow? If the latter then how and for
> which release(s) (4.2, 4.3 or 4.2.1)?
> 
Yes, I guess I'm sort of saying something like that.
What could be done is restricting automatic placement to guests that
fits on 4 or 8 nodes for 4.2. Then, for 4.3 (and with a very small
backport effort to 4.2.1), implement a lighter solution for dealing with
the ones that does not.

Hope I clarified at least a bit. :-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Campbell]

On Wed, 2012-07-18 at 10:43 +0100, Dario Faggioli wrote:
> What could be done is restricting automatic placement to guests that
> fits on 4 or 8 nodes for 4.2.

8 would mean on a 32 node system considering 10,518,300 combinations?

4 would mean on a 32 node system considering 35,960 combinations? On a
64 node system it would mean 635,376 combinations.

If that's the case then lets go with 4 as the limit for 4.2.0.

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1073 bytes --]

On Wed, 2012-07-18 at 10:53 +0100, Ian Campbell wrote:
> On Wed, 2012-07-18 at 10:43 +0100, Dario Faggioli wrote:
> > What could be done is restricting automatic placement to guests that
> > fits on 4 or 8 nodes for 4.2.
> 
> 8 would mean on a 32 node system considering 10,518,300 combinations?
> 
Yes, I was considering that only because I think we'll have something
ready well before 32 nodes systems becomes common. But fine, let's not
bet on what hardware manufacturers will come up with and in what
timeframe... :-)

> 4 would mean on a 32 node system considering 35,960 combinations? On a
> 64 node system it would mean 635,376 combinations.
> 
Exactly.

> If that's the case then lets go with 4 as the limit for 4.2.0.
> 
That is fine with me.

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Jackson]

Ian Campbell writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
> On Wed, 2012-07-18 at 10:43 +0100, Dario Faggioli wrote:
> > What could be done is restricting automatic placement to guests that
> > fits on 4 or 8 nodes for 4.2.
> 
> 8 would mean on a 32 node system considering 10,518,300 combinations?
> 
> 4 would mean on a 32 node system considering 35,960 combinations? On a
> 64 node system it would mean 635,376 combinations.
> 
> If that's the case then lets go with 4 as the limit for 4.2.0.

What is the maximum number of NUMA nodes we might expect to see on a
single system in the next five years?  I would argue that 32 is too
optimistic.  128 or 256 seem like more reasonable upper bounds.  4 of
256 is 174,792,640 combinations - ie too many.  So there needs to be a
limit on the host size too.  But if you pick an upper host size limit
of 64 then in a hypothetical 128-node system you'd fail to do the
trivial search for a 1-node guest.

An upper bound on log(number_of_combinations) is
log(number_of_nodes_on_the_host) * number_of_nodes_for_the_guest.
This fact could be used to determine more accurately whether the
algorithm is going to terminate in a reasonable time.

Also when this algorithm would be used, but would take too long, we
should print a warning which tells the user they should use cpupools
to assign nodes directly.

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Campbell]

On Wed, 2012-07-18 at 12:00 +0100, Ian Jackson wrote:
> An upper bound on log(number_of_combinations) is
> log(number_of_nodes_on_the_host) * number_of_nodes_for_the_guest.
> This fact could be used to determine more accurately whether the
> algorithm is going to terminate in a reasonable time.

This seems like something which could be done for 4.2.0 and is better
than the more static limit and better than no NUMA at all. Dario can you
do something for this soon, as in this week or not later than next?

> Also when this algorithm would be used, but would take too long, we
> should print a warning which tells the user they should use cpupools
> to assign nodes directly.

Agreed.

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1043 bytes --]

On Wed, 2012-07-18 at 14:14 +0100, Ian Campbell wrote: 
> On Wed, 2012-07-18 at 12:00 +0100, Ian Jackson wrote:
> > An upper bound on log(number_of_combinations) is
> > log(number_of_nodes_on_the_host) * number_of_nodes_for_the_guest.
> > This fact could be used to determine more accurately whether the
> > algorithm is going to terminate in a reasonable time.
> 
> This seems like something which could be done for 4.2.0 and is better
> than the more static limit and better than no NUMA at all. 
>
I think that too.

> Dario can you
> do something for this soon, as in this week or not later than next?
> 
I believe it's feasible for me to release a new version with something
like the above in by the end of this week.

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 3645 bytes --]

On Wed, 2012-07-18 at 14:14 +0100, Ian Campbell wrote:
> On Wed, 2012-07-18 at 12:00 +0100, Ian Jackson wrote:
> > An upper bound on log(number_of_combinations) is
> > log(number_of_nodes_on_the_host) * number_of_nodes_for_the_guest.
> > This fact could be used to determine more accurately whether the
> > algorithm is going to terminate in a reasonable time.
> 
> This seems like something which could be done for 4.2.0 and is better
> than the more static limit and better than no NUMA at all. Dario can you
> do something for this soon, as in this week or not later than next?
> 
Ok, I think I can do something like the below.

Remember there is an external cycle (the while{}) with
i:=[min_nodes...max_nodes] (with, typically, min_nodes=1 and
max_nodes=nr_ndoes). Nested in it, there is another cycle (the for{}),
performing exactly C(nr_nodes i) steps. So:

 1) I'll killing the need for storing the full list of combinations and 
    I'll run the comparisons on-line, so that I always have the best 
    temporary placement candidate, and I can return it anytime;

 2) at the beginning of each step of the for{} (the inner cycle) I
    check if the overall number of steps exceeds a given threshold (I
    was thinking to 2^18=262144) and if it does I just stop, no matter
    whether or not I have a solution;

The --unlikely but-- worst that can happen is the algorithm runs for a
while and at some point it terminates with a suboptimal candidate or
with nothing at all.
However, please consider that placement on an 8 nodes host requires 256
steps, on 16 nodes it takes 65536 steps. Thus, bad things would only
happen on systems with nr_nodes>16, which we now know will probably
never exist! :-)

As an alternative, after thinking quite a bit about it, I found a way to
produce a reasonably good estimation of (an upper bound of) the total
number of steps (as the log(n)*p suggested above is nice, but very very
far from being precise enough, I'm afraid). So, using that to know in
advance whether or not we should run is possible, but calculating it
requires some quite advanced math that I'm not very keen on implementing
here, as it will be ugly and not so easy to understand and maintain
(although I'd do my best in providing good comments :-)). OTOH, doing
like I described above enables the possibility of finding a (maybe
suboptimal) placement in a reasonable amount of time even on those
theoretic large systems, which is something good, I think.

> > Also when this algorithm would be used, but would take too long, we
> > should print a warning which tells the user they should use cpupools
> > to assign nodes directly.
>
> Agreed.
> 
This is definitely possible, although doing it _in_advance_ and being
precise enough would involve the advanced math I was talking about
before.

I can print the warning on-line, as soon as the estimated number of
steps reaches some threshold (lower that the one used to stop the
algorithm. I was thinking at 2^16=65536), would that be reasonable? I of
course can also print the warning in advance basing on the response of
some rough estimation, but I fear it would not be that useful then...

I'm implementing the thing just right now. If you have any feedback,
please shout it loud, and the sooner the better. :-D

Thanks a lot and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Andre Przywara]

On 07/18/2012 01:00 PM, Ian Jackson wrote:
> Ian Campbell writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
>> On Wed, 2012-07-18 at 10:43 +0100, Dario Faggioli wrote:
>>> What could be done is restricting automatic placement to guests that
>>> fits on 4 or 8 nodes for 4.2.
>>
>> 8 would mean on a 32 node system considering 10,518,300 combinations?
>>
>> 4 would mean on a 32 node system considering 35,960 combinations? On a
>> 64 node system it would mean 635,376 combinations.
>>
>> If that's the case then lets go with 4 as the limit for 4.2.0.
>
> What is the maximum number of NUMA nodes we might expect to see on a
> single system in the next five years?  I would argue that 32 is too
> optimistic.  128 or 256 seem like more reasonable upper bounds.

Wow, what are you talking about?
To calm this down from the AMD side:
The current Opteron NUMA architecture is limited to exactly 8 nodes. 
This has ever been the case since the release of Opteron and changing 
this is not trivial and will not happen in any near future. In general I 
don't think we will see much bigger NUMA systems, but more cluster like 
architectures.

Maybe Juergen can comment on the Fujitsu side.

So my suggestion: Get this NUMA placement in if anyhow possible for 
4.2.0. We have much bigger problems without this algorithm than the 
theoretic gigantic machine you are talking about. Since it is an 
internal algorithm, we can fix this later easily in 4.2.1 and 4.3 and 
nobody will ever notice this problem.

Just my 2 cents.

Andre.


4 of
> 256 is 174,792,640 combinations - ie too many.  So there needs to be a
> limit on the host size too.  But if you pick an upper host size limit
> of 64 then in a hypothetical 128-node system you'd fail to do the
> trivial search for a 1-node guest.
>
> An upper bound on log(number_of_combinations) is
> log(number_of_nodes_on_the_host) * number_of_nodes_for_the_guest.
> This fact could be used to determine more accurately whether the
> algorithm is going to terminate in a reasonable time.
>
> Also when this algorithm would be used, but would take too long, we
> should print a warning which tells the user they should use cpupools
> to assign nodes directly.
>
> Ian.
>



-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Juergen Gross]

Am 18.07.2012 15:40, schrieb Andre Przywara:
> On 07/18/2012 01:00 PM, Ian Jackson wrote:
>> Ian Campbell writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable
>> automatic placement of guests on NUMA nodes [and 1 more messages]"):
>>> On Wed, 2012-07-18 at 10:43 +0100, Dario Faggioli wrote:
>>>> What could be done is restricting automatic placement to guests that
>>>> fits on 4 or 8 nodes for 4.2.
>>>
>>> 8 would mean on a 32 node system considering 10,518,300 combinations?
>>>
>>> 4 would mean on a 32 node system considering 35,960 combinations? On a
>>> 64 node system it would mean 635,376 combinations.
>>>
>>> If that's the case then lets go with 4 as the limit for 4.2.0.
>>
>> What is the maximum number of NUMA nodes we might expect to see on a
>> single system in the next five years? I would argue that 32 is too
>> optimistic. 128 or 256 seem like more reasonable upper bounds.
>
> Wow, what are you talking about?
> To calm this down from the AMD side:
> The current Opteron NUMA architecture is limited to exactly 8 nodes.
> This has ever been the case since the release of Opteron and changing
> this is not trivial and will not happen in any near future. In general I
> don't think we will see much bigger NUMA systems, but more cluster like
> architectures.
>
> Maybe Juergen can comment on the Fujitsu side.

I'm not aware of anything with more than 8 nodes. OTOH I'm in the SW
department. I'm not sure I would hear about a larger machine long before
it is announced in public. :-)


Juergen

-- 
Juergen Gross                 Principal Developer Operating Systems
PDG ES&S SWE OS6                       Telephone: +49 (0) 89 3222 2967
Fujitsu Technology Solutions              e-mail: juergen.gross@ts.fujitsu.com
Domagkstr. 28                           Internet: ts.fujitsu.com
D-80807 Muenchen                 Company details: ts.fujitsu.com/imprint.html

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1955 bytes --]

Hi Andre,

On Wed, 2012-07-18 at 15:40 +0200, Andre Przywara wrote: 
> > What is the maximum number of NUMA nodes we might expect to see on a
> > single system in the next five years?  I would argue that 32 is too
> > optimistic.  128 or 256 seem like more reasonable upper bounds.
> 
> Wow, what are you talking about?
> To calm this down from the AMD side:
> The current Opteron NUMA architecture is limited to exactly 8 nodes. 
> This has ever been the case since the release of Opteron and changing 
> this is not trivial and will not happen in any near future.
>
Cool, thanks for bringing your view into this discussion! :-)

> In general I 
> don't think we will see much bigger NUMA systems, but more cluster like 
> architectures.
> 
Yep, I think that too, but hearing this from you is much more important
than hearing it from me as, given your position, your crystal ball is
quite a bit more trained for this kind of foresight than our ones! :-P

> Maybe Juergen can comment on the Fujitsu side.
> 
That would be very interesting.

> So my suggestion: Get this NUMA placement in if anyhow possible for 
> 4.2.0. We have much bigger problems without this algorithm than the 
> theoretic gigantic machine you are talking about. Since it is an 
> internal algorithm, we can fix this later easily in 4.2.1 and 4.3 and 
> nobody will ever notice this problem.
> 
While I basically agree, I think the solution would benefit from some
kind of "safety stop condition" similar to the one IanJ suggested. I'm
just trying to fine tune it a bit, to be sure as much as possible system
might benefit from the placement.

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Jackson]

Andre Przywara writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
> On 07/18/2012 01:00 PM, Ian Jackson wrote:
> > What is the maximum number of NUMA nodes we might expect to see on a
> > single system in the next five years?  I would argue that 32 is too
> > optimistic.  128 or 256 seem like more reasonable upper bounds.
> 
> Wow, what are you talking about?
> To calm this down from the AMD side:
> The current Opteron NUMA architecture is limited to exactly 8 nodes. 
> This has ever been the case since the release of Opteron and changing 
> this is not trivial and will not happen in any near future. In general I 
> don't think we will see much bigger NUMA systems, but more cluster like 
> architectures.

One of these `more cluster like architectures', if it has shared
memory at all, may well end up being most easily represented in the
model supported by Xen as a NUMA host with a large number of nodes.

We need to plan for the software that we are writing today to work for
at least the next 5 years /even if we intend to replace the algorithm
in the next release/.  It is very difficult to foresee what might
happen to hardware in that time.

But it's OK, we don't need to panic.  We just need a safety catch
which stops this algorithm running in situations where it won't work.
Following a discussion with Dario, AIUI he plans to implement a rule
that it will bail on systems with more than 8 NUMA nodes.

That is releasable in 4.2 and ought to be satisfactory for you ?

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Andre Przywara]

On 07/19/2012 04:43 PM, Ian Jackson wrote:
> Andre Przywara writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
>> On 07/18/2012 01:00 PM, Ian Jackson wrote:
>>> What is the maximum number of NUMA nodes we might expect to see on a
>>> single system in the next five years?  I would argue that 32 is too
>>> optimistic.  128 or 256 seem like more reasonable upper bounds.
>>
>> Wow, what are you talking about?
>> To calm this down from the AMD side:
>> The current Opteron NUMA architecture is limited to exactly 8 nodes.
>> This has ever been the case since the release of Opteron and changing
>> this is not trivial and will not happen in any near future. In general I
>> don't think we will see much bigger NUMA systems, but more cluster like
>> architectures.
>
> One of these `more cluster like architectures', if it has shared
> memory at all, may well end up being most easily represented in the
> model supported by Xen as a NUMA host with a large number of nodes.
>
> We need to plan for the software that we are writing today to work for
> at least the next 5 years /even if we intend to replace the algorithm
> in the next release/.  It is very difficult to foresee what might
> happen to hardware in that time.
>
> But it's OK, we don't need to panic.  We just need a safety catch
> which stops this algorithm running in situations where it won't work.
> Following a discussion with Dario, AIUI he plans to implement a rule
> that it will bail on systems with more than 8 NUMA nodes.
>
> That is releasable in 4.2 and ought to be satisfactory for you ?

Absolutely, that's fine. I just wanted to make sure that we don't drop 
the patches at all because of the theoretical fear of a large NUMA box 
lingering somewhere around the corner...

Thanks,
Andre.

-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1500 bytes --]

On Thu, 2012-07-19 at 20:37 +0200, Andre Przywara wrote:
> > But it's OK, we don't need to panic.  We just need a safety catch
> > which stops this algorithm running in situations where it won't work.
> > Following a discussion with Dario, AIUI he plans to implement a rule
> > that it will bail on systems with more than 8 NUMA nodes.
> >
> > That is releasable in 4.2 and ought to be satisfactory for you ?
> 
I did right that. Honestly, I computed the exact amount of steps the
algorithm needs in the worst case (i.e., searching all the combinations
of all the sizes) for 8 and 16 nodes. 8 nodes means 256 steps, 16 nodes
means 65536, so I think both could work, especially after the
optimization I put in the version I just released (it does not uses that
large array of combinations and does not need to always perform all the
steps any longer).

However, as agreed, I put an "if (nr_nodes <= 8)" there... If a 16 nodes
machine will ever appear, we'll decide what to do.

> Absolutely, that's fine. I just wanted to make sure that we don't drop 
> the patches at all because of the theoretical fear of a large NUMA box 
> lingering somewhere around the corner...
> 
:-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Jackson]

Dario Faggioli writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
...
> So basically I'm saying that it does have that property but that is
> bringing benefits right now, and we have all the knobs to limit its
> potential nasty effects for future (bigger) system at any time.

I think that it is essential that we do not have a situation where the
default behaviour is ever to hang while it searches zillions of
combinations.  That would be a release critical bug IMO.

So I don't think "we will fix this after the series goes in" is
acceptable.  We should not be introducing new RC bugs during the
freeze.

> What could be done is restricting automatic placement to guests that
> fits on 4 or 8 nodes for 4.2. Then, for 4.3 (and with a very small
> backport effort to 4.2.1), implement a lighter solution for dealing with
> the ones that does not.

That would be acceptable.

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Ian Campbell]

On Wed, 2012-07-18 at 11:53 +0100, Ian Jackson wrote:
> Dario Faggioli writes ("Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]"):
> ...
> > So basically I'm saying that it does have that property but that is
> > bringing benefits right now, and we have all the knobs to limit its
> > potential nasty effects for future (bigger) system at any time.
> 
> I think that it is essential that we do not have a situation where the
> default behaviour is ever to hang while it searches zillions of
> combinations.  That would be a release critical bug IMO.
> 
> So I don't think "we will fix this after the series goes in" is
> acceptable.  We should not be introducing new RC bugs during the
> freeze.

If this is the case then we need to consider that we may need to release
4.2.0 without this NUMA awareness unless we can find a simple
fix/bandaid soon. 

I don't want this freeze to drag on for weeks or months while we sort
this out.

I'd like to see a solution agreed and implemented ASAP or else we need
to seriously consider deferring the whole thing to 4.3 (perhaps with a
4.2.1 backport)

Ian.

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes [and 1 more messages]

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 2277 bytes --]

On Wed, 2012-07-18 at 02:22 +0200, Dario Faggioli wrote:
> Moreover, as another input parameter is a nodemask, telling the
> algorithm not only how many but also _which_ nodes it should consider,
> one could imagine to have some multi-tier form of the algorithm itself,
> or, perhaps, implementing a more abstract and coarse partitioning
> heuristics among group of nodes (which will better be done as soon as
> we'll see how 32 and 64 nodes system will actually look like :-D) and
> then run the current algorithm only on these groups, with numbers that
> would then look just like above.
> 
Alternatively (or maybe not even that much 'alternatively' :-)), we can
try turning it into something a bit more greedy, e.g., searching for
available placement solutions just as big as min_nodes says (typically
1) as a first step (complexity linear with the number of nodes).

After that, if we did not find anything, we could have some sort of
config switch telling us if we should continue, and maybe investigating
candidates of sizes in the range [min_nodes+1,LIBXL__MAX_GUEST_NODES]. 

At which point, if we haven't found anything yet, we could just give up
or, perhaps, think and implement in the near future a more coarse and
less precise mechanism for proceeding further.

Implementing something like that on top of what I have now (which I'll
resend asap, as soon as I've addressed the minor comments I got) during
4.3 should be pretty easy. Also, if we decide to just statically limit
max_nodes<=LIBXL__MAX_GUEST_NODES (or whatever we could call it) to
something (I was proposing 4 in the other e-mail) for 4.2, backporting
to it will be piece of cake as well. :-)

Thanks again and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Andre Przywara]

Dario,

sorry for joining the discussion so late, but I was busy with other 
things and saw the project in good hands.
Finally I managed to get some testing on these patches.

I took my 8-node machine, alternately equipped with 16GB and 8GB per 
node. Each node has 8 pCPUs.
As a special(i)ty I removed the DIMMs from node 2 and 3 to test Andrew's 
memory-less node patches, leading to this configuration:
node:    memsize    memfree    distances
    0:     17280       4518      10,16,16,22,16,22,16,22
    1:      8192       3639      16,10,16,22,22,16,22,16
    2:         0          0      16,16,10,16,16,16,16,22
    3:         0          0      22,22,16,10,16,16,22,16
    4:     16384       4766      16,22,16,16,10,16,16,16
    5:      8192       2882      22,16,16,16,16,10,22,22
    6:     16384       4866      16,22,16,22,16,22,10,16
    7:      8176       2799      22,16,22,16,16,22,16,10

Then I started 32 guests, each 4 vCPUs and 1 GB of RAM.
Now since the code prefers free memory so much over free CPUs, the 
placement was the following:
node0: guests 2,5,8,11,14,17,20,25,30
node1: guests 21,27
node2: none
node3: none
node4: guests 1,4,7,10,13,16,19,23,29
node5: guests 24,31
node6: guests 3,6,9,12,15,18,22,28
node7: guests 26,32

As you can see, the nodes with more memory are _way_ overloaded, while 
the lower memory ones are underutilized. In fact the first 20 guests 
didn't use the other nodes at all.
I don't care so much about the two memory-less nodes, but I'd like to 
know how you came to the magic "3" in the formula:

> +
> +    return sign(3*freememkb_diff + nrdomains_diff);
> +}

I haven't done any measurements on this, but I guess scheduling 36 vCPUs 
on 8 pCPUs has a much bigger performance penalty than any remote NUMA 
access, which nowadays is much better than a few years ago, with big L3 
caches, better predictors and faster interconnects.

I will now put in the memory again and also try to play a bit with the 
amount of memory and VCPUs per guest.

Regards,
Andre.

-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 5310 bytes --]

On Thu, 2012-07-19 at 14:21 +0200, Andre Przywara wrote:
> Dario,
> 
Hi Andre!

> sorry for joining the discussion so late, but I was busy with other 
> things and saw the project in good hands.
>
Well, thanks. It's being a big deal (and we're just at the beginning!),
but I'm enjoying working on it so I won't give up easily! :-P

> Finally I managed to get some testing on these patches.
> 
That's very cool, thanks.

> I took my 8-node machine, alternately equipped with 16GB and 8GB per 
> node. Each node has 8 pCPUs.
> As a special(i)ty I removed the DIMMs from node 2 and 3 to test Andrew's 
> memory-less node patches, leading to this configuration:
> node:    memsize    memfree    distances
>     0:     17280       4518      10,16,16,22,16,22,16,22
>     1:      8192       3639      16,10,16,22,22,16,22,16
>     2:         0          0      16,16,10,16,16,16,16,22
>     3:         0          0      22,22,16,10,16,16,22,16
>     4:     16384       4766      16,22,16,16,10,16,16,16
>     5:      8192       2882      22,16,16,16,16,10,22,22
>     6:     16384       4866      16,22,16,22,16,22,10,16
>     7:      8176       2799      22,16,22,16,16,22,16,10
> 
Interesting. That's really the kind of testing we need in order to
fine-tune the details. Thanks for doing this.

> Then I started 32 guests, each 4 vCPUs and 1 GB of RAM.
> Now since the code prefers free memory so much over free CPUs, the 
> placement was the following:
> node0: guests 2,5,8,11,14,17,20,25,30
> node1: guests 21,27
> node2: none
> node3: none
> node4: guests 1,4,7,10,13,16,19,23,29
> node5: guests 24,31
> node6: guests 3,6,9,12,15,18,22,28
> node7: guests 26,32
> 
> As you can see, the nodes with more memory are _way_ overloaded, while 
> the lower memory ones are underutilized. In fact the first 20 guests 
> didn't use the other nodes at all.
> I don't care so much about the two memory-less nodes, but I'd like to 
> know how you came to the magic "3" in the formula:
> 
> > +
> > +    return sign(3*freememkb_diff + nrdomains_diff);
> > +}
> 
Ok. The idea behind the current implementation of that heuristics is to
prefer nodes with more free memory. In fact, this leaves larger "holes",
maximizing the probability of being able to put more domain there. Of
course that means more domains exploiting local accesses, but introduces
the risk of overloading large (from a memory POV) nodes with a lot of
domains (which seems right what's happening to you! :-P).

Therefore, I wanted to balance that by putting something related to the
current load on a node into the equation. Unfortunately, I really am not
sure yet what a reasonable estimation of the actual "load on a node"
could be. Even worse, Xen does not report anything even close to that,
at least not right now. That's why I went for a quite dumb count of the
number of domains for now, waiting to find the time to implement
something more clever.

So, that is basically why I thought it could be a good idea to
overweight the differences in free memory wrt the differences in number
of assigned domain. The first implementation was only considering the
number of assigned domain to decide which was the best candidate between
two that were less 10% different in their amount of free memory.
However, that didn't produce a good comparison function, and thus I
rewrote it like above, with the magic 3 selected via trial and error to
mimic something similar to the old 10% rule.

That all being said, this is the first time the patchset had the chance
to run on such a big system, so I'm definitely open to suggestion on how
to make that formula better in reflecting what we think it's The Right
Thing!

> I haven't done any measurements on this, but I guess scheduling 36 vCPUs 
> on 8 pCPUs has a much bigger performance penalty than any remote NUMA 
> access, which nowadays is much better than a few years ago, with big L3 
> caches, better predictors and faster interconnects.
> 
Definitely. Consider that the guests are being pinned because that is
what XenD does and because there has been no time to properly refine the
implementation of a more NUMA-aware scheduler for 4.2. In future, as
soon as I'll have it ready, the vcpus from the overloaded nodes would
get some runtime on the otherwise idle ones, even if they're remote.

Nevertheless, this is what Xen 4.2 will have, and I really think initial
placement is a very important step, and we must get the most out of
being able to do it well (as opposed to other technologies, where
something like that has to happen in the kernel/hypervisor, which
entails a lot of limitations we don't have!), and am therefore happy
about trying to do so as hard as I can.

> I will now put in the memory again and also try to play a bit with the 
> amount of memory and VCPUs per guest.
> 
Great, keep me super-posted about these things and feel free to ask
anything that comes to your mind! :-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Andre Przywara]

On 07/19/2012 04:22 PM, Dario Faggioli wrote:
> On Thu, 2012-07-19 at 14:21 +0200, Andre Przywara wrote:
>> Dario,

thanks for the warm welcome.

 >> ...
>> As you can see, the nodes with more memory are _way_ overloaded, while
>> the lower memory ones are underutilized. In fact the first 20 guests
>> didn't use the other nodes at all.
>> I don't care so much about the two memory-less nodes, but I'd like to
>> know how you came to the magic "3" in the formula:
>>
>>> +
>>> +    return sign(3*freememkb_diff + nrdomains_diff);
>>> +}
>>
> Ok. The idea behind the current implementation of that heuristics is to
> prefer nodes with more free memory. In fact, this leaves larger "holes",
> maximizing the probability of being able to put more domain there. Of
> course that means more domains exploiting local accesses, but introduces
> the risk of overloading large (from a memory POV) nodes with a lot of
> domains (which seems right what's happening to you! :-P).

I always assumed the vast majority of actual users/customers use 
comparably small domains, something like 1-4 VCPUs and like 4 GB of RAM. 
So these domains are much smaller than a usual node. I'd consider a node 
size of 16 GB the lower boundary, with up to 128GB as the common 
scenarios. Sure there are bigger or smaller machines, but I'd consider 
this the sweet spot.

> Therefore, I wanted to balance that by putting something related to the
> current load on a node into the equation. Unfortunately, I really am not
> sure yet what a reasonable estimation of the actual "load on a node"
> could be. Even worse, Xen does not report anything even close to that,
> at least not right now. That's why I went for a quite dumb count of the
> number of domains for now, waiting to find the time to implement
> something more clever.

Right. So we just use the number of already pinned vCPUs as the metric. 
Let me look if I can change the code to really use number of vCPUs 
instead of number of domains. A domain could be UP or 8-way SMP, which 
really makes much difference wrt to load on a node.

In the long run we need something like a per-node (or per-pCPU) load 
average. We cannot foresee the future, but we just assume that the past 
is a good indicator for it. xl top generates such numbers on demand 
already. But that surely is something for 4.3, just wanted to mention it.

> So, that is basically why I thought it could be a good idea to
> overweight the differences in free memory wrt the differences in number
> of assigned domain. The first implementation was only considering the
> number of assigned domain to decide which was the best candidate between
> two that were less 10% different in their amount of free memory.
> However, that didn't produce a good comparison function, and thus I
> rewrote it like above, with the magic 3 selected via trial and error to
> mimic something similar to the old 10% rule.

OK, I see. I thought about this a bit more and agree a single heuristic 
formula isn't easy to find. After reading the code I consider this a bit 
over-engineered, but I cannot possibly complain about this after having 
remained silent for such a long time.
So lets see what we can make out of this code, just firing up some 
ideas, feel free to just ignore them in case they are dumb ;-)

So if you agree to the small-domain assumption, then domains easily 
fitting into a node are the rule, not the exception. We should handle it 
that way. Maybe we can also solve the complexity problem by only 
generating single node candidates in the first place and only if these 
don't fit look at alternatives?

I really admire that lazy comb_next generation function, so why we don't 
use it in a really lazy way? I think there was already a discussion 
about this, just don't remember what it's outcome was.
Some debugging code showed that on the above (empty) machine a 2 
VCPUs/2GB domain generated already 255 candidates. That really looks 
like overkill, especially if we actually should focus on the 8 
single-node ones.

Maybe we can use a two-step approach? First use a simple heuristic 
similar to the xend one:
We only consider domains with enough free memory. Then we look for the 
least utilized ones: Simply calculate the difference between the number 
of currently pinned vCPUs and the number pCPUs. So any node with free 
(aka non-overcommited) CPUs should really be considered first.
After all we don't need to care about memory latency if the domains 
starve for compute time and only get a fraction of each pCPU.
If you don't want to believe this, I can run some benchmarks to prove this.

If we somehow determine that this approach doesn't work (no nodes with 
enough free memory or more vCPUs than CPUs-per-node) we should use the 
sophisticated algorithm.

Also consider the following: With really big machines or with odd 
configurations people will probably do their pinning/placement 
themselves (or by external mgmt applications).
What this automatic placement algorithm is good for is more the 
What-is-this-NUMA-thingie-anyways people.

>
> That all being said, this is the first time the patchset had the chance
> to run on such a big system, so I'm definitely open to suggestion on how
> to make that formula better in reflecting what we think it's The Right
> Thing!
>
>> I haven't done any measurements on this, but I guess scheduling 36 vCPUs
>> on 8 pCPUs has a much bigger performance penalty than any remote NUMA
>> access, which nowadays is much better than a few years ago, with big L3
>> caches, better predictors and faster interconnects.
>>
> Definitely. Consider that the guests are being pinned because that is
> what XenD does and because there has been no time to properly refine the
> implementation of a more NUMA-aware scheduler for 4.2. In future, as
> soon as I'll have it ready, the vcpus from the overloaded nodes would
> get some runtime on the otherwise idle ones, even if they're remote.

Right, that sounds good. If you have any good (read: meaningful to 
customers) benchmarks I can do some experiments on my machine to 
fine-tune this.

> Nevertheless, this is what Xen 4.2 will have, and I really think initial
> placement is a very important step, and we must get the most out of
> being able to do it well (as opposed to other technologies, where
> something like that has to happen in the kernel/hypervisor, which
> entails a lot of limitations we don't have!), and am therefore happy
> about trying to do so as hard as I can.

Right. We definitely need some placement for 4.2. Lets push this in if 
anyhow possible.

>> I will now put in the memory again and also try to play a bit with the
>> amount of memory and VCPUs per guest.
>>
> Great, keep me super-posted about these things and feel free to ask
> anything that comes to your mind! :-)

So changing the VCPUs and memory config didn't make any real difference. 
I think that is because the number of domains is considered, not the 
number of vCPUS. This should be fixed.

Second I inserted the memory again. I only have 24 DIMMs for 32 sockets 
(we have easy access to boards and CPUs, but memory we have to buy like 
everyone else ;-), so I have to go with this alternating setup, having 
four 16GB nodes and four 8 GB nodes.

This didn't change much, so 16 guests ended up with:
4-1-0-0-4-0-4-3 setup (domains per node). The 0's or 1's where the 8GB 
nodes. The guests were 2P/2GB ones.

So far.

Regards,
Andre.

-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany
Tel: +49 351 448-3567-12

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 5602 bytes --]

On Fri, 2012-07-20 at 10:19 +0200, Andre Przywara wrote:
> thanks for the warm welcome.
> 
:-)

> I always assumed the vast majority of actual users/customers use 
> comparably small domains, something like 1-4 VCPUs and like 4 GB of RAM. 
> So these domains are much smaller than a usual node. I'd consider a node 
> size of 16 GB the lower boundary, with up to 128GB as the common 
> scenarios. Sure there are bigger or smaller machines, but I'd consider 
> this the sweet spot.
> 
Yep, I agree this should be our target, at least for the default
behaviour.

> Right. So we just use the number of already pinned vCPUs as the metric. 
> Let me look if I can change the code to really use number of vCPUs 
> instead of number of domains. A domain could be UP or 8-way SMP, which 
> really makes much difference wrt to load on a node.
> 
As we both are saying in the other e-mail, this is probably something
good to do. I'm working on a new version of the patchset that I'm going
to release later today (addressing the comments and the outcome of the
long discussion the last round generated). If you can go as far as
having a patch, that would be great, and I guess it could make it in
even in early -rc days.

> In the long run we need something like a per-node (or per-pCPU) load 
> average. We cannot foresee the future, but we just assume that the past 
> is a good indicator for it. xl top generates such numbers on demand 
> already. But that surely is something for 4.3, just wanted to mention it.
>
Definitely.

> After reading the code I consider this a bit 
> over-engineered, but I cannot possibly complain about this after having 
> remained silent for such a long time.
>
:-)

> So lets see what we can make out of this code, just firing up some 
> ideas, feel free to just ignore them in case they are dumb ;-)
> 
I bet they are not!

> So if you agree to the small-domain assumption, then domains easily 
> fitting into a node are the rule, not the exception. We should handle it 
> that way. Maybe we can also solve the complexity problem by only 
> generating single node candidates in the first place and only if these 
> don't fit look at alternatives?
> 
That's exactly what I'm doing right now. You'll see the code later but,
basically, at the i-eth step, I now compare all the candidates with i
nodes and, if I find at least one, I quit the whole thing and avoid
proceeding at step i+1. Of course, if I find more than one, I return the
one that is best according to the heuristics.

Given that the very first step is very likely going to be looking at
candidates with 1 node in them, here you are exactly what you're talking
about.

> I really admire that lazy comb_next generation function, so why we don't 
> use it in a really lazy way? I think there was already a discussion 
> about this, just don't remember what it's outcome was.
> Some debugging code showed that on the above (empty) machine a 2 
> VCPUs/2GB domain generated already 255 candidates. That really looks 
> like overkill, especially if we actually should focus on the 8 
> single-node ones.
> 
Yep, and in fact 255 is what it takes on 8 nodes. As said above, I
sort-of read your mind and started implementing what you write here
yesterday. :-P

> Maybe we can use a two-step approach? First use a simple heuristic 
> similar to the xend one:
> We only consider domains with enough free memory. Then we look for the 
> least utilized ones: Simply calculate the difference between the number 
> of currently pinned vCPUs and the number pCPUs. So any node with free 
> (aka non-overcommited) CPUs should really be considered first.
>
Again, with the change above, this thing you're saying here can be
achieved just removing the memfree_diff from the comparison function
(which is no longer used during a proper sort, rather is is being called
on-line as soon as a new candidate is found, to compare it with the
current cached best). And yes, of course turning the domain count into a
vcpu count, as said above.

> If we somehow determine that this approach doesn't work (no nodes with 
> enough free memory or more vCPUs than CPUs-per-node) we should use the 
> sophisticated algorithm.
> 
And again, you'll see the new code and will tell me what you think
later, but I really think I turned it into something like that. The only
thing 

> Right, that sounds good. If you have any good (read: meaningful to 
> customers) benchmarks I can do some experiments on my machine to 
> fine-tune this.
> 
Nothing that goes that far. I tried to run specjbb2005 concurrently on
some VM with and without placement, but I only have a very small
testbox. :-(

> > Nevertheless, this is what Xen 4.2 will have, and I really think initial
> > placement is a very important step, and we must get the most out of
> > being able to do it well (as opposed to other technologies, where
> > something like that has to happen in the kernel/hypervisor, which
> > entails a lot of limitations we don't have!), and am therefore happy
> > about trying to do so as hard as I can.
> 
> Right. We definitely need some placement for 4.2. Lets push this in if 
> anyhow possible.
> 
That's what I'm working quite hard for. :-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1492 bytes --]

On Fri, 2012-07-20 at 11:39 +0200, Dario Faggioli wrote:
> > If we somehow determine that this approach doesn't work (no nodes with 
> > enough free memory or more vCPUs than CPUs-per-node) we should use the 
> > sophisticated algorithm.
> > 
> And again, you'll see the new code and will tell me what you think
> later, but I really think I turned it into something like that. The only
> thing 
> 
Some text got cut here.

What I wanted to ask is what you think we should do with guests that
have more vcpus than pcpus of a node. Right now, as I do for memory, I'm
using it as a reason to consider candidates with more than one node.
XenD did something similar, as it adds nodes (although it does that
afterwords) until the above stop being true.

Should we ignore that and stick to the number of nodes that gives us
enough memory? Should we do that but only up to a certain extent (e.g.,
by asking for a candidate to have at least half pcpus as the domain has
vcpus)? After all, even if we give the domain enough pcpus, he's not
going to have exclusive access to them anyway...

Again, just random thought seeking for a confirmation that is very hard
to find! :-)

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 2107 bytes --]

On Thu, 2012-07-19 at 16:22 +0200, Dario Faggioli wrote:
> Interesting. That's really the kind of testing we need in order to
> fine-tune the details. Thanks for doing this.
> 
> > Then I started 32 guests, each 4 vCPUs and 1 GB of RAM.
> > Now since the code prefers free memory so much over free CPUs, the 
> > placement was the following:
> > node0: guests 2,5,8,11,14,17,20,25,30
> > node1: guests 21,27
> > node2: none
> > node3: none
> > node4: guests 1,4,7,10,13,16,19,23,29
> > node5: guests 24,31
> > node6: guests 3,6,9,12,15,18,22,28
> > node7: guests 26,32
> > 
> > As you can see, the nodes with more memory are _way_ overloaded, while 
> > the lower memory ones are underutilized. In fact the first 20 guests 
> > didn't use the other nodes at all.
> > I don't care so much about the two memory-less nodes, but I'd like to 
> > know how you came to the magic "3" in the formula:
> > 
> > > +
> > > +    return sign(3*freememkb_diff + nrdomains_diff);
> > > +}
> > 
>
> That all being said, this is the first time the patchset had the chance
> to run on such a big system, so I'm definitely open to suggestion on how
> to make that formula better in reflecting what we think it's The Right
> Thing!
> 
Thinking more about this, I realize that I was implicitly assuming some
symmetry in the amount of memory each nodes comes with, which is
probably something I shouldn't have done...

I really am not sure what to do here, perhaps treating the two metrics
more evenly? Or maybe even reverse the logic and give nr_domains more
weight?

I was also thinking whether it could be worthwhile to consider the total
number of vcpus on a node instead than the number of domain, but again,
that's not guaranteed to be any more meaningful (suppose there are a lot
of idle vcpus)...

Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Andre Przywara]

On 07/20/2012 10:20 AM, Dario Faggioli wrote:
> On Thu, 2012-07-19 at 16:22 +0200, Dario Faggioli wrote:
>> Interesting. That's really the kind of testing we need in order to
>> fine-tune the details. Thanks for doing this.
>>
>>> Then I started 32 guests, each 4 vCPUs and 1 GB of RAM.
>>> Now since the code prefers free memory so much over free CPUs, the
>>> placement was the following:
>>> node0: guests 2,5,8,11,14,17,20,25,30
>>> node1: guests 21,27
>>> node2: none
>>> node3: none
>>> node4: guests 1,4,7,10,13,16,19,23,29
>>> node5: guests 24,31
>>> node6: guests 3,6,9,12,15,18,22,28
>>> node7: guests 26,32
>>>
>>> As you can see, the nodes with more memory are _way_ overloaded, while
>>> the lower memory ones are underutilized. In fact the first 20 guests
>>> didn't use the other nodes at all.
>>> I don't care so much about the two memory-less nodes, but I'd like to
>>> know how you came to the magic "3" in the formula:
>>>
>>>> +
>>>> +    return sign(3*freememkb_diff + nrdomains_diff);
>>>> +}
>>>
>>
>> That all being said, this is the first time the patchset had the chance
>> to run on such a big system, so I'm definitely open to suggestion on how
>> to make that formula better in reflecting what we think it's The Right
>> Thing!
>>
> Thinking more about this, I realize that I was implicitly assuming some
> symmetry in the amount of memory each nodes comes with, which is
> probably something I shouldn't have done...
>
> I really am not sure what to do here, perhaps treating the two metrics
> more evenly? Or maybe even reverse the logic and give nr_domains more
> weight?

I replaced the 3 with 1 already, that didn't change so much. I think we 
should kind of reverse the importance of node load, since starving for 
CPU time is much worse than bad memory latency. I will do some 
experiments...

> I was also thinking whether it could be worthwhile to consider the total
> number of vcpus on a node instead than the number of domain, but again,
> that's not guaranteed to be any more meaningful (suppose there are a lot
> of idle vcpus)...

Right, that was my thinking on the ride to work also ;-)
What about this: 1P and 2P guests really use their vCPUs, but for bigger 
guests we assume only a fractional usage?

Andre

-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany
Tel: +49 351 448-3567-12

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Juergen Gross]

Am 20.07.2012 10:26, schrieb Andre Przywara:
> On 07/20/2012 10:20 AM, Dario Faggioli wrote:
>> On Thu, 2012-07-19 at 16:22 +0200, Dario Faggioli wrote:
>>> Interesting. That's really the kind of testing we need in order to
>>> fine-tune the details. Thanks for doing this.
>>>
>>>> Then I started 32 guests, each 4 vCPUs and 1 GB of RAM.
>>>> Now since the code prefers free memory so much over free CPUs, the
>>>> placement was the following:
>>>> node0: guests 2,5,8,11,14,17,20,25,30
>>>> node1: guests 21,27
>>>> node2: none
>>>> node3: none
>>>> node4: guests 1,4,7,10,13,16,19,23,29
>>>> node5: guests 24,31
>>>> node6: guests 3,6,9,12,15,18,22,28
>>>> node7: guests 26,32
>>>>
>>>> As you can see, the nodes with more memory are _way_ overloaded, while
>>>> the lower memory ones are underutilized. In fact the first 20 guests
>>>> didn't use the other nodes at all.
>>>> I don't care so much about the two memory-less nodes, but I'd like to
>>>> know how you came to the magic "3" in the formula:
>>>>
>>>>> +
>>>>> + return sign(3*freememkb_diff + nrdomains_diff);
>>>>> +}
>>>>
>>>
>>> That all being said, this is the first time the patchset had the chance
>>> to run on such a big system, so I'm definitely open to suggestion on how
>>> to make that formula better in reflecting what we think it's The Right
>>> Thing!
>>>
>> Thinking more about this, I realize that I was implicitly assuming some
>> symmetry in the amount of memory each nodes comes with, which is
>> probably something I shouldn't have done...
>>
>> I really am not sure what to do here, perhaps treating the two metrics
>> more evenly? Or maybe even reverse the logic and give nr_domains more
>> weight?
>
> I replaced the 3 with 1 already, that didn't change so much. I think we
> should kind of reverse the importance of node load, since starving for
> CPU time is much worse than bad memory latency. I will do some
> experiments...
>
>> I was also thinking whether it could be worthwhile to consider the total
>> number of vcpus on a node instead than the number of domain, but again,
>> that's not guaranteed to be any more meaningful (suppose there are a lot
>> of idle vcpus)...
>
> Right, that was my thinking on the ride to work also ;-)
> What about this: 1P and 2P guests really use their vCPUs, but for bigger
> guests we assume only a fractional usage?

Hmm. I wouldn't be sure about this. I would guess there is a reason a guest has
more than 1 vcpu. Normally this is because the guest needs more power.

Having only 1 vcpu means this is enough. It probably would need only 0.1 vcpus.

A guest with many vcpus suffers much more from a heavy loaded node, as lock
contention in the guest is increasing with number of vcpus and waiting for a
vcpu holding a lock but being preempted.


Juergen

-- 
Juergen Gross                 Principal Developer Operating Systems
PDG ES&S SWE OS6                       Telephone: +49 (0) 89 3222 2967
Fujitsu Technology Solutions              e-mail: juergen.gross@ts.fujitsu.com
Domagkstr. 28                           Internet: ts.fujitsu.com
D-80807 Muenchen                 Company details: ts.fujitsu.com/imprint.html

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 984 bytes --]

On Fri, 2012-07-20 at 10:38 +0200, Juergen Gross wrote:
> > Right, that was my thinking on the ride to work also ;-)
> > What about this: 1P and 2P guests really use their vCPUs, but for bigger
> > guests we assume only a fractional usage?
> 
> Hmm. I wouldn't be sure about this. I would guess there is a reason a guest has
> more than 1 vcpu. Normally this is because the guest needs more power.
> 
Hehe, I agree with both! :-P

Maybe it's better to avoid this kind of mangling for now, and just
consider all the vcpus.

When we'll be tracking the actual load better, we'll see if it could be
worthwhile to do something along that line either.

What do you think?

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Juergen Gross]

Am 20.07.2012 11:52, schrieb Dario Faggioli:
> On Fri, 2012-07-20 at 10:38 +0200, Juergen Gross wrote:
>>> Right, that was my thinking on the ride to work also ;-)
>>> What about this: 1P and 2P guests really use their vCPUs, but for bigger
>>> guests we assume only a fractional usage?
>>
>> Hmm. I wouldn't be sure about this. I would guess there is a reason a guest has
>> more than 1 vcpu. Normally this is because the guest needs more power.
>>
> Hehe, I agree with both! :-P
>
> Maybe it's better to avoid this kind of mangling for now, and just
> consider all the vcpus.
>
> When we'll be tracking the actual load better, we'll see if it could be
> worthwhile to do something along that line either.
>
> What do you think?

Yep.


Juergen

-- 
Juergen Gross                 Principal Developer Operating Systems
PDG ES&S SWE OS6                       Telephone: +49 (0) 89 3222 2967
Fujitsu Technology Solutions              e-mail: juergen.gross@ts.fujitsu.com
Domagkstr. 28                           Internet: ts.fujitsu.com
D-80807 Muenchen                 Company details: ts.fujitsu.com/imprint.html

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1076 bytes --]

On Fri, 2012-07-20 at 10:26 +0200, Andre Przywara wrote:
> > I really am not sure what to do here, perhaps treating the two metrics
> > more evenly? Or maybe even reverse the logic and give nr_domains more
> > weight?
> 
> I replaced the 3 with 1 already, that didn't change so much. I think we 
> should kind of reverse the importance of node load, since starving for 
> CPU time is much worse than bad memory latency. I will do some 
> experiments...
> 
That would be nice. If you happen to have time to put something like
"3*nrdomains_diff+memfree_diff" and see how it goes, I'll be happy to
include at least that change, even in next posting.

After that, let's see what it takes to turn the domain counting into
vcpu counting (it shouldn't be too hard).

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Andre Przywara]

On 07/20/2012 11:44 AM, Dario Faggioli wrote:
> On Fri, 2012-07-20 at 10:26 +0200, Andre Przywara wrote:
>>> I really am not sure what to do here, perhaps treating the two metrics
>>> more evenly? Or maybe even reverse the logic and give nr_domains more
>>> weight?
>>
>> I replaced the 3 with 1 already, that didn't change so much. I think we
>> should kind of reverse the importance of node load, since starving for
>> CPU time is much worse than bad memory latency. I will do some
>> experiments...
>>
> That would be nice. If you happen to have time to put something like
> "3*nrdomains_diff+memfree_diff" and see how it goes, I'll be happy to
> include at least that change, even in next posting.

I did that. Guests are 2 VCPUs/2GB RAM. The results looked much better. 
After 16 guests I get:
# xl vcpu-list | sed -e 1d | sort -n -k 7 | tr -s \  | cut -d\  -f7 | 
uniq -c
      16 any       (Dom0 had max_vcpus=16)
       4 0-7
       4 8-15
       4 16-23
       4 24-31
       4 32-39
       4 40-47
       4 48-55
       4 56-63
This is number of VCPUs per node. Also distributes equally.
Memory looked like this:
    0:     17280       9969
    1:      8192       2268
    2:      8192       1690
    3:      8192       1754
    4:     16384      10049
    5:      8192       1879
    6:     16384      10947
    7:     16368       9267

After another 8 guests:
      16 any
       8 0-7
       4 8-15
       4 16-23
       4 24-31
       8 32-39
       4 40-47
       8 48-55
       8 56-63
still not over-commited.
    0:     17280       5846
    1:      8192       2266
    2:      8192       1690
    3:      8192       1752
    4:     16384       5925
    5:      8192       1877
    6:     16384       6824
    7:     16368       5142

Finally with all 32 guests:
      12 0-7
       6 8-15
       4 16-23
       4 24-31
      12 32-39
       4 40-47
      12 48-55
      10 56-63
The bigger nodes are overcommited, while the others have free pCPUs (8 
cores per node). But memory dictates this:
    0:     17280       3130
    1:      8192          0
    2:      8192        485
    3:      8192       1747
    4:     16384       1803
    5:      8192       1876
    6:     16384       2701
    7:     16368       3081
And the last domain already took memory from multiple nodes:
(XEN) Domain 105 (total: 523255):
(XEN)     Node 0: 162740
(XEN)     Node 1: 52616
(XEN)     Node 2: 307899
(XEN)     Node 3: 0
(XEN)     Node 4: 0
(XEN)     Node 5: 0
(XEN)     Node 6: 0
(XEN)     Node 7: 0

All other domains had all their 523255 pages from a single node.

# xl vcpu-list 105
Name                        ID  VCPU   CPU State   Time(s) CPU Affinity
Guest32                    105     0    5   -b-       3.1  0-7
Guest32                    105     1    2   -b-       0.7  0-7

So without any deeper thinking this looks much better than the original 
version and possibly good enough for Xen 4.2.
Maybe the automated testing finds some leftovers.


Regards,
Andre.

-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1395 bytes --]

On Fri, 2012-07-20 at 13:47 +0200, Andre Przywara wrote:
> > That would be nice. If you happen to have time to put something like
> > "3*nrdomains_diff+memfree_diff" and see how it goes, I'll be happy to
> > include at least that change, even in next posting.
> 
> I did that. Guests are 2 VCPUs/2GB RAM. 
>
Thanks!

> The results looked much better. 
>
That's very good to hear. :-)

>
> [snip]
>
> So without any deeper thinking this looks much better than the original 
> version and possibly good enough for Xen 4.2.
> Maybe the automated testing finds some leftovers.
> 
So, here's what we can do:

 1) I can resend the series with this change in it (i.e., number of 
   domains weighting 3 times more than free memory);

 2) I can resend the series like in 1) _but_ while also converting the 
   domain counting in vcpu counting (I checked that and it's trivial);

 3) I can resend the series like in 1) and also send a separate patch 
   turning the domain counting in vcpu counting, and we can apply that 
   after it get some testing;

Which one you like better?

Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Andre Przywara]

On 07/20/2012 02:54 PM, Dario Faggioli wrote:
> On Fri, 2012-07-20 at 13:47 +0200, Andre Przywara wrote:
>>> That would be nice. If you happen to have time to put something like
>>> "3*nrdomains_diff+memfree_diff" and see how it goes, I'll be happy to
>>> include at least that change, even in next posting.
>>
>> I did that. Guests are 2 VCPUs/2GB RAM.
>>
> Thanks!
>
>> The results looked much better.
>>
> That's very good to hear. :-)
>
>>
>> [snip]
>>
>> So without any deeper thinking this looks much better than the original
>> version and possibly good enough for Xen 4.2.
>> Maybe the automated testing finds some leftovers.
>>
> So, here's what we can do:
>
>   1) I can resend the series with this change in it (i.e., number of
>     domains weighting 3 times more than free memory);
>
>   2) I can resend the series like in 1) _but_ while also converting the
>     domain counting in vcpu counting (I checked that and it's trivial);
>
>   3) I can resend the series like in 1) and also send a separate patch
>     turning the domain counting in vcpu counting, and we can apply that
>     after it get some testing;
>
> Which one you like better?

I'd prefer 2). To me it makes more sense and I can test it over the 
weekend and can send my acked-by on it.

Regards,
Andre.

-- 
Andre Przywara
AMD-OSRC (Dresden)
Tel: x29712

Re: [PATCH 1 of 3 v5/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1077 bytes --]

On Fri, 2012-07-20 at 15:07 +0200, Andre Przywara wrote:
> > So, here's what we can do:
> >
> >   1) I can resend the series with this change in it (i.e., number of
> >     domains weighting 3 times more than free memory);
> >
> >   2) I can resend the series like in 1) _but_ while also converting the
> >     domain counting in vcpu counting (I checked that and it's trivial);
> >
> >   3) I can resend the series like in 1) and also send a separate patch
> >     turning the domain counting in vcpu counting, and we can apply that
> >     after it get some testing;
> >
> > Which one you like better?
> 
> I'd prefer 2). To me it makes more sense and I can test it over the 
> weekend and can send my acked-by on it.
> 
Ok, done. Sorry if I am a little bit late.

Regards,
Dario


-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

[PATCH 2 of 3 v5/leftover] libxl: have NUMA placement deal with cpupools

[Dario Faggioli]

In such a way that only the cpus belonging to the cpupool of the
domain being placed are considered for the placement itself.

This happens by filtering out all the nodes in which the cpupool has
not any cpu from the placement candidates. After that -- as a cpu pooling
not necessarily happens at NUMA nodes boundaries -- we also make sure
only the actual cpus that are part of the pool are considered when
counting how much processors a placement candidate provides.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

---
Changes from v4:
 * fixed a bug in the cpupool filtering logic (also the number of nodes
   should be affected by that)

Changes from v3:
 * fixed a double free on non-NUMA systems (namely, when just 1 node exists).
 * Changed the filtering logic: now only the meaningful combinations and
   nodemap are being considered.

Changes from v2:
 * fixed typos in comments.

diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -132,25 +132,29 @@ static int numa_cmpf(const void *v1, con
 }
 
 /* The actual automatic NUMA placement routine */
-static int numa_place_domain(libxl__gc *gc, libxl_domain_build_info *info)
+static int numa_place_domain(libxl__gc *gc, uint32_t domid,
+                             libxl_domain_build_info *info)
 {
     int nr_candidates = 0;
     libxl__numa_candidate *candidates = NULL;
     libxl_bitmap candidate_nodemap;
-    libxl_cpupoolinfo *pinfo;
-    int nr_pools, rc = 0;
+    libxl_cpupoolinfo cpupool_info;
+    int i, cpupool, rc = 0;
     uint32_t memkb;
 
     libxl_bitmap_init(&candidate_nodemap);
 
-    /* First of all, if cpupools are in use, better not to mess with them */
-    pinfo = libxl_list_cpupool(CTX, &nr_pools);
-    if (!pinfo)
-        return ERROR_FAIL;
-    if (nr_pools > 1) {
-        LOG(NOTICE, "skipping NUMA placement as cpupools are in use");
-        goto out;
-    }
+    /*
+     * Extract the cpumap from the cpupool the domain belong to. In fact,
+     * it only makes sense to consider the cpus/nodes that are in there
+     * for placement.
+     */
+    rc = cpupool = libxl__domain_cpupool(gc, domid);
+    if (rc < 0)
+        return rc;
+    rc = libxl_cpupool_info(CTX, &cpupool_info, cpupool);
+    if (rc)
+        return rc;
 
     rc = libxl_domain_need_memory(CTX, info, &memkb);
     if (rc)
@@ -162,7 +166,8 @@ static int numa_place_domain(libxl__gc *
 
     /* Find all the candidates with enough free memory and at least
      * as much pcpus as the domain has vcpus.  */
-    rc = libxl__get_numa_candidates(gc, memkb, info->max_vcpus, 0, 0,
+    rc = libxl__get_numa_candidates(gc, memkb, info->max_vcpus,
+                                    0, 0, &cpupool_info.cpumap,
                                     &candidates, &nr_candidates);
     if (rc)
         goto out;
@@ -188,13 +193,20 @@ static int numa_place_domain(libxl__gc *
     if (rc)
         goto out;
 
+    /* Avoid trying to set the affinity to cpus that might be in the
+     * nodemap but not in our cpupool. */
+    libxl_for_each_set_bit(i, info->cpumap) {
+        if (!libxl_bitmap_test(&cpupool_info.cpumap, i))
+            libxl_bitmap_reset(&info->cpumap, i);
+    }
+
     LOG(DETAIL, "NUMA placement candidate with %d nodes, %d cpus and "
                 "%"PRIu32" KB free selected", candidates[0].nr_nodes,
                 candidates[0].nr_cpus, candidates[0].free_memkb / 1024);
 
  out:
     libxl_bitmap_dispose(&candidate_nodemap);
-    libxl_cpupoolinfo_list_free(pinfo, nr_pools);
+    libxl_cpupoolinfo_dispose(&cpupool_info);
     return rc;
 }
 
@@ -217,7 +229,7 @@ int libxl__build_pre(libxl__gc *gc, uint
      * whatever that turns out to be.
      */
     if (libxl_bitmap_is_full(&info->cpumap)) {
-        int rc = numa_place_domain(gc, info);
+        int rc = numa_place_domain(gc, domid, info);
         if (rc)
             return rc;
     }
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -2289,14 +2289,17 @@ typedef struct {
  * least that amount of free memory and that number of cpus, respectively. If
  * min_free_memkb and/or min_cpus are 0, the candidates' free memory and number
  * of cpus won't be checked at all, which means a candidate will always be
- * considered suitable wrt the specific constraint.  cndts is where the list of
- * exactly nr_cndts candidates is returned. Note that, in case no candidates
- * are found at all, the function returns successfully, but with nr_cndts equal
- * to zero.
+ * considered suitable wrt the specific constraint. suitable_cpumap is useful
+ * for specifying we want only the cpus in that mask to be considered while
+ * generating placement candidates (for example because of cpupools). cndts is
+ * where the list of exactly nr_cndts candidates is returned. Note that, in
+ * case no candidates are found at all, the function returns successfully, but
+ * with nr_cndts equal to zero.
  */
 _hidden int libxl__get_numa_candidates(libxl__gc *gc,
                                 uint32_t min_free_memkb, int min_cpus,
                                 int min_nodes, int max_nodes,
+                                const libxl_bitmap *suitable_cpumap,
                                 libxl__numa_candidate *cndts[], int *nr_cndts);
 
 /* Initialization, allocation and deallocation for placement candidates */
diff --git a/tools/libxl/libxl_numa.c b/tools/libxl/libxl_numa.c
--- a/tools/libxl/libxl_numa.c
+++ b/tools/libxl/libxl_numa.c
@@ -107,30 +107,48 @@ static int comb_next(comb_iter_t it, int
 /* NUMA automatic placement (see libxl_internal.h for details) */
 
 /*
- * This function turns a k-combination iterator into a node map.
- * This means the bits in the node map corresponding to the indexes
- * of the given combination are the ones that will be set.
- * For example, if the iterator represents the combination { 0, 2, 4},
- * the node map will have bits #0, #2 and #4 set.
+ * This function turns a k-combination iterator into a node map,
+ * given another map, telling us which nodes should be considered.
+ *
+ * This means the bits that are set in suitable_nodemap and that
+ * corresponds to the indexes of the given combination are the ones
+ * that will be set in nodemap.
+ *
+ * For example, given a fully set suitable_nodemap, if the iterator
+ * represents the combination { 0, 2, 4}, nodmeap will have bits #0,
+ * #2 and #4 set.
+ * On the other hand, if, say,  suitable_nodemap=01011011, the same
+ * iterator will cause bits #1, #4 and #7 of nodemap to be set.
  */
-static void comb_get_nodemap(comb_iter_t it, libxl_bitmap *nodemap, int k)
+static void comb_get_nodemap(comb_iter_t it, libxl_bitmap *suitable_nodemap,
+                             libxl_bitmap *nodemap, int k)
 {
-    int i;
+    int i, m = 0, n = 0;
 
     libxl_bitmap_set_none(nodemap);
-    for (i = 0; i < k; i++)
-        libxl_bitmap_set(nodemap, it[i]);
+    libxl_for_each_set_bit(i, *suitable_nodemap) {
+        /* Check wether the n-th set bit of suitable_nodemap
+         * matches with the m-th element of the iterator (and,
+         * only if it does, advance to the next one) */
+        if (m < k && n == it[m]) {
+            libxl_bitmap_set(nodemap, i);
+            m++;
+        }
+        n++;
+    }
 }
 
 /* Retrieve the number of cpus that the nodes that are part of the nodemap
- * span. */
+ * span and are also set in suitable_cpumap. */
 static int nodemap_to_nr_cpus(libxl_cputopology *tinfo, int nr_cpus,
+                              const libxl_bitmap *suitable_cpumap,
                               const libxl_bitmap *nodemap)
 {
     int i, nodes_cpus = 0;
 
     for (i = 0; i < nr_cpus; i++) {
-        if (libxl_bitmap_test(nodemap, tinfo[i].node))
+        if (libxl_bitmap_test(suitable_cpumap, i) &&
+            libxl_bitmap_test(nodemap, tinfo[i].node))
             nodes_cpus++;
     }
     return nodes_cpus;
@@ -236,15 +254,17 @@ static int cpus_per_node_count(libxl_cpu
 int libxl__get_numa_candidates(libxl__gc *gc,
                                uint32_t min_free_memkb, int min_cpus,
                                int min_nodes, int max_nodes,
+                               const libxl_bitmap *suitable_cpumap,
                                libxl__numa_candidate *cndts[], int *nr_cndts)
 {
     libxl__numa_candidate *new_cndts = NULL;
     libxl_cputopology *tinfo = NULL;
     libxl_numainfo *ninfo = NULL;
-    int nr_nodes = 0, nr_cpus = 0;
-    libxl_bitmap nodemap;
+    int nr_nodes = 0, nr_suit_nodes, nr_cpus = 0;
+    libxl_bitmap suitable_nodemap, nodemap;
     int array_size, rc;
 
+    libxl_bitmap_init(&suitable_nodemap);
     libxl_bitmap_init(&nodemap);
 
     /* Get platform info and prepare the map for testing the combinations */
@@ -262,6 +282,15 @@ int libxl__get_numa_candidates(libxl__gc
     if (rc)
         goto out;
 
+    /* Allocate and prepare the map of the node that can be utilized for
+     * placement, basing on the map of suitable cpus. */
+    rc = libxl_node_bitmap_alloc(CTX, &suitable_nodemap, 0);
+    if (rc)
+        goto out;
+    rc = libxl_cpumap_to_nodemap(CTX, suitable_cpumap, &suitable_nodemap);
+    if (rc)
+        goto out;
+
     /*
      * If the minimum number of NUMA nodes is not explicitly specified
      * (i.e., min_nodes == 0), we try to figure out a sensible number of nodes
@@ -279,10 +308,14 @@ int libxl__get_numa_candidates(libxl__gc
         else
             min_nodes = (min_cpus + cpus_per_node - 1) / cpus_per_node;
     }
-    if (min_nodes > nr_nodes)
-        min_nodes = nr_nodes;
-    if (!max_nodes || max_nodes > nr_nodes)
-        max_nodes = nr_nodes;
+    /* We also need to be sure we do not exceed the number of
+     * nodes we are allowed to use. */
+    nr_suit_nodes = libxl_bitmap_count_set(&suitable_nodemap);
+
+    if (min_nodes > nr_suit_nodes)
+        min_nodes = nr_suit_nodes;
+    if (!max_nodes || max_nodes > nr_suit_nodes)
+        max_nodes = nr_suit_nodes;
     if (min_nodes > max_nodes) {
         rc = ERROR_INVAL;
         goto out;
@@ -290,7 +323,7 @@ int libxl__get_numa_candidates(libxl__gc
 
     /* Initialize the local storage for the combinations */
     *nr_cndts = 0;
-    array_size = nr_nodes;
+    array_size = nr_suit_nodes;
     GCNEW_ARRAY(new_cndts, array_size);
 
     /* Generate all the combinations of any size from min_nodes to
@@ -306,12 +339,16 @@ int libxl__get_numa_candidates(libxl__gc
 	 * amount of free memory and number of cpus) and it becomes an actual
 	 * placement candidate iff it passes the check.
          */
-        for (comb_ok = comb_init(gc, &comb_iter, nr_nodes, min_nodes); comb_ok;
-             comb_ok = comb_next(comb_iter, nr_nodes, min_nodes)) {
+        for (comb_ok = comb_init(gc, &comb_iter, nr_nodes, min_nodes);
+             comb_ok;
+             comb_ok = comb_next(comb_iter, nr_suit_nodes, min_nodes)) {
             uint32_t nodes_free_memkb;
             int nodes_cpus;
 
-            comb_get_nodemap(comb_iter, &nodemap, min_nodes);
+            /* Get the nodemap for the combination, only considering
+             * suitable nodes. */
+            comb_get_nodemap(comb_iter, &suitable_nodemap,
+                             &nodemap, min_nodes);
 
             /* If there is not enough memory in this combination, skip it
              * and go generating the next one... */
@@ -320,7 +357,8 @@ int libxl__get_numa_candidates(libxl__gc
                 continue;
 
             /* And the same applies if this combination is short in cpus */
-            nodes_cpus = nodemap_to_nr_cpus(tinfo, nr_cpus, &nodemap);
+            nodes_cpus = nodemap_to_nr_cpus(tinfo, nr_cpus, suitable_cpumap,
+                                            &nodemap);
             if (min_cpus && nodes_cpus < min_cpus)
                 continue;
 
@@ -336,7 +374,7 @@ int libxl__get_numa_candidates(libxl__gc
              * new candidate).
              */
             if (*nr_cndts == array_size)
-                array_size += nr_nodes;
+                array_size += nr_suit_nodes;
             GCREALLOC_ARRAY(new_cndts, array_size);
 
             libxl__numa_candidate_alloc(gc, &new_cndts[*nr_cndts]);
@@ -345,12 +383,13 @@ int libxl__get_numa_candidates(libxl__gc
             new_cndts[*nr_cndts].nr_domains =
                                     nodemap_to_nr_domains(gc, tinfo, &nodemap);
             new_cndts[*nr_cndts].free_memkb = nodes_free_memkb;
-            new_cndts[*nr_cndts].nr_nodes = min_nodes;
+            new_cndts[*nr_cndts].nr_nodes = libxl_bitmap_count_set(&nodemap);
             new_cndts[*nr_cndts].nr_cpus = nodes_cpus;
 
             LOG(DEBUG, "NUMA placement candidate #%d found: nr_nodes=%d, "
                        "nr_cpus=%d, free_memkb=%"PRIu32"", *nr_cndts,
-                       min_nodes, new_cndts[*nr_cndts].nr_cpus,
+                       new_cndts[*nr_cndts].nr_nodes,
+                       new_cndts[*nr_cndts].nr_cpus,
                        new_cndts[*nr_cndts].free_memkb / 1024);
 
             (*nr_cndts)++;
@@ -360,6 +399,7 @@ int libxl__get_numa_candidates(libxl__gc
 
     *cndts = new_cndts;
  out:
+    libxl_bitmap_dispose(&suitable_nodemap);
     libxl_bitmap_dispose(&nodemap);
     libxl_cputopology_list_free(tinfo, nr_cpus);
     libxl_numainfo_list_free(ninfo, nr_nodes);

[PATCH 3 of 3 v5/leftover] Some automatic NUMA placement documentation

[Dario Faggioli]

About rationale, usage and (some small bits of) API.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

---
Changes from v3:
 * typos and rewording of some sentences, as suggested during review.

Changes from v1:
 * API documentation moved close to the actual functions.

diff --git a/docs/misc/xl-numa-placement.markdown b/docs/misc/xl-numa-placement.markdown
new file mode 100644
--- /dev/null
+++ b/docs/misc/xl-numa-placement.markdown
@@ -0,0 +1,89 @@
+# Guest Automatic NUMA Placement in libxl and xl #
+
+## Rationale ##
+
+NUMA means the memory accessing times of a program running on a CPU depends on
+the relative distance between that CPU and that memory. In fact, most of the
+NUMA systems are built in such a way that each processor has its local memory,
+on which it can operate very fast. On the other hand, getting and storing data
+from and on remote memory (that is, memory local to some other processor) is
+quite more complex and slow. On these machines, a NUMA node is usually defined
+as a set of processor cores (typically a physical CPU package) and the memory
+directly attached to the set of cores.
+
+The Xen hypervisor deals with Non-Uniform Memory Access (NUMA]) machines by
+assigning to its domain a "node affinity", i.e., a set of NUMA nodes of the
+host from which it gets its memory allocated.
+
+NUMA awareness becomes very important as soon as many domains start running
+memory-intensive workloads on a shared host. In fact, the cost of accessing non
+node-local memory locations is very high, and the performance degradation is
+likely to be noticeable.
+
+## Guest Placement in xl ##
+
+If using xl for creating and managing guests, it is very easy to ask for both
+manual or automatic placement of them across the host's NUMA nodes.
+
+Note that xm/xend does the very same thing, the only differences residing in
+the details of the heuristics adopted for the placement (see below).
+
+### Manual Guest Placement with xl ###
+
+Thanks to the "cpus=" option, it is possible to specify where a domain should
+be created and scheduled on, directly in its config file. This affects NUMA
+placement and memory accesses as the hypervisor constructs the node affinity of
+a VM basing right on its CPU affinity when it is created.
+
+This is very simple and effective, but requires the user/system administrator
+to explicitly specify affinities for each and every domain, or Xen won't be
+able to guarantee the locality for their memory accesses.
+
+It is also possible to deal with NUMA by partitioning the system using cpupools
+(available in the upcoming release of Xen, 4.2). Again, this could be "The
+Right Answer" for many needs and occasions, but  has to to be carefully
+considered and manually setup by hand.
+
+### Automatic Guest Placement with xl ###
+
+If no "cpus=" option is specified in the config file, libxl tries to figure out
+on its own on which node(s) the domain could fit best.  It is worthwhile noting
+that optimally fitting a set of VMs on the NUMA nodes of an host is an
+incarnation of the Bin Packing Problem. In fact, the various VMs with different
+memory sizes are the items to be packed, and the host nodes are the bins. That
+is known to be NP-hard, thus. We will therefore be using some heuristics.
+
+The first thing to do is finding  a node, or even a set of nodes, that have
+enough free memory and enough physical CPUs for accommodating the one new
+domain. The idea is to find a spot for the domain with at least as much free
+memory as it has configured, and as much pCPUs as it has vCPUs.  After that,
+the actual decision on which solution to go for happens accordingly to the
+following heuristics:
+
+  *  candidates involving fewer nodes come first. In case two (or more)
+     candidates span the same number of nodes,
+  *  the amount of free memory and the number of domains assigned to the
+     candidates are considered. In doing that, candidates with greater amount
+     of free memory and fewer assigned domains are preferred, with free memory
+     "weighting" three times as much as number of domains.
+
+Giving preference to candidates with fewer nodes ensures better performance for
+the guest, as it avoid spreading its memory among different nodes.  Favouring
+the nodes that have the largest amounts of free memory helps keeping the memory
+fragmentation small, from a system wide perspective.  However, if more
+candidates fulfil these criteria by roughly the same extent, having the number
+of domains the candidates are "hosting" helps balancing the load on the various
+nodes.
+
+## Guest Placement within libxl ##
+
+xl achieves automatic NUMA just because libxl does it interrnally.
+No API is provided (yet) for interacting with this feature and modify
+the library behaviour regarding automatic placement, it just happens
+by default if no affinity is specified (as it is with xm/xend).
+
+For actually looking and maybe tweaking the mechanism and the algorithms it
+uses, all is implemented as a set of libxl internal interfaces and facilities.
+Look at the comment "Automatic NUMA placement" in libxl\_internal.h.
+
+Note this may change in future versions of Xen/libxl.

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[David Vrabel]

On 16/07/12 18:13, Dario Faggioli wrote:
> Hello again,
> 
> This is a new version (fixing a small bug) of this series:
> <patchbomb.1341932613@Solace>, which in turn was a repost of what remained
> un-committed of my NUMA placement series.

Whilst I don't think this should prevent the merging of this series now,
I think there needs to be some sort of unit tests for the placement
algorithm before the 4.2 release.

I think the tests should test a representative sample of common system
configurations, available memory and VM memory requirements.  I'd
suggest you'd be looking at 100s of test cases here for reasonable coverage.

One method would be to start with various 'empty' systems and pile as
many differently sized VMs as will fit.  You may want both fixed test of
reproducible tests and random ones.

Some means of automatically verifying the quality of the result at each
stage would be essential.

Without a comprehensive set of tests, tweaking the algorithm/heuristics
has a high risk of introducing regressions.

Having the ability to specify specific setups so you can (e.g.) run
tests when someone reports sub-optimal performance with their particular
setup would also be useful.

David

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Andre Przywara]

On 07/20/2012 01:07 PM, David Vrabel wrote:
> On 16/07/12 18:13, Dario Faggioli wrote:
>> Hello again,
>>
>> This is a new version (fixing a small bug) of this series:
>> <patchbomb.1341932613@Solace>, which in turn was a repost of what remained
>> un-committed of my NUMA placement series.
>
> Whilst I don't think this should prevent the merging of this series now,
> I think there needs to be some sort of unit tests for the placement
> algorithm before the 4.2 release.
>
> I think the tests should test a representative sample of common system
> configurations, available memory and VM memory requirements.  I'd
> suggest you'd be looking at 100s of test cases here for reasonable coverage.
>
> One method would be to start with various 'empty' systems and pile as
> many differently sized VMs as will fit.  You may want both fixed test of
> reproducible tests and random ones.

That sounds good. Though I don't have much automated testing experience 
with Xen I could chime in with two things:

1. If we focus on placement only, I have good experience with 
ttylinux.iso. Those live distros can be killed easily at any time and 
you just need one instance of the .iso file on the disk.
2. # xl vcpu-list | sed -e 1d | sort -n -k 7 | tr -s \  | cut -d\  -f7 | 
uniq -c
This gives the number of VCPUs per node (sort of ;-)

> Some means of automatically verifying the quality of the result at each
> stage would be essential.

This "automatically verifying the quality of the result" doesn't sound 
trivial. If we knew this exactly, we could just code this into the 
algorithm, right?
Also it seems to depend on the setup. Maybe we just collect some test 
output first and then try to come up with quality algorithms?

Regards,
Andre.


-- 
Andre Przywara
AMD-Operating System Research Center (OSRC), Dresden, Germany

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Ian Campbell]

On Fri, 2012-07-20 at 12:43 +0100, Andre Przywara wrote:
> On 07/20/2012 01:07 PM, David Vrabel wrote:
> > On 16/07/12 18:13, Dario Faggioli wrote:
> >> Hello again,
> >>
> >> This is a new version (fixing a small bug) of this series:
> >> <patchbomb.1341932613@Solace>, which in turn was a repost of what remained
> >> un-committed of my NUMA placement series.
> >
> > Whilst I don't think this should prevent the merging of this series now,
> > I think there needs to be some sort of unit tests for the placement
> > algorithm before the 4.2 release.
> >
> > I think the tests should test a representative sample of common system
> > configurations, available memory and VM memory requirements.  I'd
> > suggest you'd be looking at 100s of test cases here for reasonable coverage.
> >
> > One method would be to start with various 'empty' systems and pile as
> > many differently sized VMs as will fit.  You may want both fixed test of
> > reproducible tests and random ones.
> 
> That sounds good. Though I don't have much automated testing experience 
> with Xen I could chime in with two things:
> 
> 1. If we focus on placement only, I have good experience with 
> ttylinux.iso. Those live distros can be killed easily at any time and 
> you just need one instance of the .iso file on the disk.
> 2. # xl vcpu-list | sed -e 1d | sort -n -k 7 | tr -s \  | cut -d\  -f7 | 
> uniq -c
> This gives the number of VCPUs per node (sort of ;-)

Ideally you wouldn't need a xen system at all for this, you just want a
database of input configurations (host NUMA setup, existing guest
layout) and hypothetical new guests and their mapping to the expected
output. You can then feed these offline to the algorithmn and validate
that the output is the expected one.

> > Some means of automatically verifying the quality of the result at each
> > stage would be essential.
> 
> This "automatically verifying the quality of the result" doesn't sound 
> trivial. If we knew this exactly, we could just code this into the 
> algorithm, right?

Well, just knowing something has changed is useful, even you can't
decide in an automated way if it is the before or after case which is
the good/best one.

Even if we just blindly store the current result today as the expected
one then when someone makes a tweak we can manually inspect the
differences in the output and say "yes, that seems sane" or "no, that's
mad". The need to do this will be encoded in the diff which makes the
change (since you'd have to patch the test suite...)

> Also it seems to depend on the setup. Maybe we just collect some test 
> output first and then try to come up with quality algorithms?
> 
> Regards,
> Andre.
> 
> 

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Ian Campbell]

On Fri, 2012-07-20 at 13:00 +0100, Ian Campbell wrote:
> Even if we just blindly store the current result today as the expected
> one then when someone makes a tweak we can manually inspect the
> differences in the output and say "yes, that seems sane" or "no, that's
> mad". The need to do this will be encoded in the diff which makes the
> change (since you'd have to patch the test suite...)

BTW, this is mostly what we do for
tools/libxl/check-xl-{disk,vif}-parse.

and what we should do for tools/pygrub/examples/* but so far we only
collect examples...

Ian.

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1793 bytes --]

On Fri, 2012-07-20 at 13:08 +0100, Ian Campbell wrote:
> On Fri, 2012-07-20 at 13:00 +0100, Ian Campbell wrote:
> > Even if we just blindly store the current result today as the expected
> > one then when someone makes a tweak we can manually inspect the
> > differences in the output and say "yes, that seems sane" or "no, that's
> > mad". The need to do this will be encoded in the diff which makes the
> > change (since you'd have to patch the test suite...)
> 
> BTW, this is mostly what we do for
> tools/libxl/check-xl-{disk,vif}-parse.
> 
That is very nice, although our case here is a bit different. I can put
together a similar script, and I can also think about some test cases
and methods to check how things are going.

The tricky part is I _can't_ run xl just like we do in that file, as
there's not a specific command for running just the placement, and I
can't invoke `cl create', can I? :-P

So, it seems I should write a sort of C program from which to trigger
the libxl domain creation, after taking care of intercepting and
properly replacing quite a bit of functions to either nops or (e.g., in
case of libxl_get_cpu_topology(), etc) proper reads of the test
database. Also, this involves quite a bit of libxl internal function.
Will I be able to intercept them too? Not sure, I need to check...

How does that sound? I'll think about and try to implement this in next
days, any thoughts that may come to your mind, do feel free to
share. :-D

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Ian Campbell]

On Mon, 2012-07-23 at 11:38 +0100, Dario Faggioli wrote:
> On Fri, 2012-07-20 at 13:08 +0100, Ian Campbell wrote:
> > On Fri, 2012-07-20 at 13:00 +0100, Ian Campbell wrote:
> > > Even if we just blindly store the current result today as the expected
> > > one then when someone makes a tweak we can manually inspect the
> > > differences in the output and say "yes, that seems sane" or "no, that's
> > > mad". The need to do this will be encoded in the diff which makes the
> > > change (since you'd have to patch the test suite...)
> > 
> > BTW, this is mostly what we do for
> > tools/libxl/check-xl-{disk,vif}-parse.
> > 
> That is very nice, although our case here is a bit different. I can put
> together a similar script, and I can also think about some test cases
> and methods to check how things are going.
> 
> The tricky part is I _can't_ run xl just like we do in that file, as
> there's not a specific command for running just the placement, and I
> can't invoke `cl create', can I? :-P
> 
> So, it seems I should write a sort of C program from which to trigger
> the libxl domain creation, after taking care of intercepting and
> properly replacing quite a bit of functions to either nops or (e.g., in
> case of libxl_get_cpu_topology(), etc) proper reads of the test
> database. Also, this involves quite a bit of libxl internal function.
> Will I be able to intercept them too? Not sure, I need to check...
> 
> How does that sound? I'll think about and try to implement this in next
> days, any thoughts that may come to your mind, do feel free to
> share. :-D

I think it would be fine for a test program such as this to make use of
libxl internals and "look behind the curtain" as it were and call the
core of the placement algorithm directly with falsified inputs.

However I wouldn't prioritise this over any outstanding repost of the
series.

Ian.
-- 
Ian Campbell
Current Noise: Judas Priest - Painkiller

  After they got rid of capital punishment, they had to hang twice
  as many people as before.

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1055 bytes --]

On Mon, 2012-07-23 at 11:42 +0100, Ian Campbell wrote: 
> > How does that sound? I'll think about and try to implement this in next
> > days, any thoughts that may come to your mind, do feel free to
> > share. :-D
> 
> I think it would be fine for a test program such as this to make use of
> libxl internals and "look behind the curtain" as it were and call the
> core of the placement algorithm directly with falsified inputs.
> 
Ok, so, IIUC, I'll do something like "#include <libxl_internal.h" and
call my functions directly. Sounds reasonable.

> However I wouldn't prioritise this over any outstanding repost of the
> series.
> 
Me neither, and I won't do that... I was just hoping for that not to be
necessary, I guess! :-P :-P

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1760 bytes --]

On Fri, 2012-07-20 at 13:00 +0100, Ian Campbell wrote:
> On Fri, 2012-07-20 at 12:43 +0100, Andre Przywara wrote:
> > On 07/20/2012 01:07 PM, David Vrabel wrote:
> > > On 16/07/12 18:13, Dario Faggioli wrote:
> > >> Hello again,
> > >>
> > > I think the tests should test a representative sample of common system
> > > configurations, available memory and VM memory requirements.  I'd
> > > suggest you'd be looking at 100s of test cases here for reasonable coverage.
> > >
> > > One method would be to start with various 'empty' systems and pile as
> > > many differently sized VMs as will fit.  You may want both fixed test of
> > > reproducible tests and random ones.
> > 
> > 1. If we focus on placement only, I have good experience with 
> > ttylinux.iso. Those live distros can be killed easily at any time and 
> > you just need one instance of the .iso file on the disk.
> > 2. # xl vcpu-list | sed -e 1d | sort -n -k 7 | tr -s \  | cut -d\  -f7 | 
> > uniq -c
> > This gives the number of VCPUs per node (sort of ;-)
> 
> Ideally you wouldn't need a xen system at all for this, you just want a
> database of input configurations (host NUMA setup, existing guest
> layout) and hypothetical new guests and their mapping to the expected
> output. You can then feed these offline to the algorithmn and validate
> that the output is the expected one.
> 
Ok, I see. I like this and I think it should be done. I'll look into it.

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH 0 of 3 v5/leftover] Automatic NUMA placement for xl

[David Vrabel]

On 20/07/12 12:43, Andre Przywara wrote:
> On 07/20/2012 01:07 PM, David Vrabel wrote:
>> On 16/07/12 18:13, Dario Faggioli wrote:
>>> Hello again,
>>>
>>> This is a new version (fixing a small bug) of this series:
>>> <patchbomb.1341932613@Solace>, which in turn was a repost of what
>>> remained
>>> un-committed of my NUMA placement series.
>>
>> Whilst I don't think this should prevent the merging of this series now,
>> I think there needs to be some sort of unit tests for the placement
>> algorithm before the 4.2 release.
>>
>> I think the tests should test a representative sample of common system
>> configurations, available memory and VM memory requirements.  I'd
>> suggest you'd be looking at 100s of test cases here for reasonable
>> coverage.
>>
>> One method would be to start with various 'empty' systems and pile as
>> many differently sized VMs as will fit.  You may want both fixed test of
>> reproducible tests and random ones.
> 
> That sounds good. Though I don't have much automated testing experience
> with Xen I could chime in with two things:
> 
> 1. If we focus on placement only, I have good experience with
> ttylinux.iso. Those live distros can be killed easily at any time and
> you just need one instance of the .iso file on the disk.
> 2. # xl vcpu-list | sed -e 1d | sort -n -k 7 | tr -s \  | cut -d\  -f7 |
> uniq -c
> This gives the number of VCPUs per node (sort of ;-)

I'm talking about unit tests of the algorithm not system tests on real
hardware.  i.e., some sort of wrapper around the C functions that call
then with various sets of input data.

>> Some means of automatically verifying the quality of the result at each
>> stage would be essential.
> 
> This "automatically verifying the quality of the result" doesn't sound
> trivial. If we knew this exactly, we could just code this into the
> algorithm, right?

I think it is much easier to verify the result than generate the
solution.  The quality score is basically the percentage of memory that
ended up on the expected number of nodes -- this is easy to calculate.

If the algorithm is tweaked and the resulting quality score takes a nose
dive this would give a very quick indication that the tweak may be
broken.  Conversely, if the quality goes up, the tweak is more likely to
be good.

David

Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Ian Campbell]

On Tue, 2012-07-17 at 16:55 +0100, Ian Jackson wrote:

> > +/* Subtract two values and translate the result in [0, 1] */
> > +static double normalized_diff(double a, double b)
> > +{
> > +#define max(a, b) (a > b ? a : b)
> > +    if (!a && a == b)
> > +        return 0.0;
> > +    return (a - b) / max(a, b);
> > +}
> 
> 1. This macro max() should be in libxl_internal.h.
> 2. It should be MAX so people are warned it's a macro
> 3. It should have all the necessary ()s for macro precedence safety

You probably also want to do the necessary tricks to avoid multiple
evaluation of the arguments?

Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Ian Jackson]

Ian Campbell writes ("Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
> On Tue, 2012-07-17 at 16:55 +0100, Ian Jackson wrote:
> > 1. This macro max() should be in libxl_internal.h.
> > 2. It should be MAX so people are warned it's a macro
> > 3. It should have all the necessary ()s for macro precedence safety
> 
> You probably also want to do the necessary tricks to avoid multiple
> evaluation of the arguments?

Those tricks are a pain.  I think in general macros are `allowed' to
do this provided their names are SHOUTY.  That's why it should be
called `MAX'.

Ian.

Re: [PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 5699 bytes --]

On Tue, 2012-07-17 at 16:55 +0100, Ian Jackson wrote: 
> Dario Faggioli writes ("[PATCH 1 of 3 v4/leftover] libxl: enable automatic placement of guests on NUMA nodes"):
> > If a domain does not have a VCPU affinity, try to pin it automatically to some
> > PCPUs. This is done taking into account the NUMA characteristics of the host.
> > In fact, we look for a combination of host's NUMA nodes with enough free memory
> > and number of PCPUs for the new domain, and pin it to the VCPUs of those nodes.
> 
> Thanks for this admirably clear patch.
> 
Thanks to you for looking at it.

> Can you please rewrap your commit messages to around 70 lines ?  Many
> VCSs indent them in the log in some situations, and as you see here
> mail programs indent them when quoting too.
> 
That should have happened in the first place. As I'm reposting, I'll
take extra attention to that, sorry.

> > +/* Subtract two values and translate the result in [0, 1] */
> > +static double normalized_diff(double a, double b)
> > +{
> > +#define max(a, b) (a > b ? a : b)
> > +    if (!a && a == b)
> > +        return 0.0;
> > +    return (a - b) / max(a, b);
> > +}
> 
> 1. This macro max() should be in libxl_internal.h.
> 2. It should be MAX so people are warned it's a macro
> 3. It should have all the necessary ()s for macro precedence safety
> 
Ok, will do that.

> > +    double freememkb_diff = normalized_diff(c2->free_memkb, c1->free_memkb);
> > +    double nrdomains_diff = normalized_diff(c1->nr_domains, c2->nr_domains);
> > +
> > +    if (c1->nr_nodes != c2->nr_nodes)
> > +        return c1->nr_nodes - c2->nr_nodes;
> > +
> > +    return sign(3*freememkb_diff + nrdomains_diff);
> 
> The reason you need what sign() does is that you need to convert from
> double to int, I guess.
> 
Mostly for that and to make what's happening even more clear.

> > +
> > +    /*
> > +     * Check if the domain has any CPU affinity. If not, try to build up one.
> > +     * In case numa_place_domain() find at least a suitable candidate, it will
> > +     * affect info->cpumap accordingly; if it does not, it just leaves it
> > +     * as it is. This means (unless some weird error manifests) the subsequent
> > +     * call to libxl_set_vcpuaffinity_all() will do the actual placement,
> > +     * whatever that turns out to be.
> > +     */
> > +    if (libxl_bitmap_is_full(&info->cpumap)) {
> > +        int rc = numa_place_domain(gc, info);
> > +        if (rc)
> > +            return rc;
> > +    }
> 
> I guess it would be preferable to do this only if the bitmap was full
> by default, so that setting the bitmap explicitly to all cpus still
> works.
> 
> I'm not sure that that's essential to have, though.
> 
I was thinking about this right in this days, and I think it should be
exactly as you say, as one need a mechanism for disabling this thing as
a whole. I really don't think it should take to much to put something
together, even as a separate, future, patch, if these get checked-in.
Thanks.

> > +            /*
> > +             * Conditions are met, we can add this combination to the
> > +             * NUMA placement candidates list. We first make sure there
> > +             * is enough space in there, and then we initialize the new
> > +             * candidate element with the node map corresponding to the
> > +             * combination we are dealing with. Memory allocation for
> > +             * expanding the array that hosts the list happens in chunks
> > +             * equal to the number of NUMA nodes in the system (to
> > +             * avoid allocating memory each and every time we find a
> > +             * new candidate).
> > +             */
> > +            if (*nr_cndts == array_size)
> > +                array_size += nr_nodes;
> > +            GCREALLOC_ARRAY(new_cndts, array_size);
> 
> This part of the algorithm is quadratic in the number of combinations
> divided by the number of nodes.  So the algorithm is
>    O( (  C( nr_nodes, min_nodes ) / min_nodes  )^2 )
> which is quite bad really.
> 
I might well be wrong, but I was thinking to it as something like this:

O( C(nr_nodes,(nr_nodes/2)) * nr_nodes )

That's because the external while() is repeated, at most, nr_nodes times
(if min_nodes=1 and max_nodes=nr_nodes). Each of these steps hosts a
for() which visits all the combinations, the maximum number of which
ISTR to be C(nr_nodes,(nr_nodes/2)).

I'm not sure what you meant when putting min_nodes up there in your
formula (min_nodes is likely to be 1 most of the cases...), so I can't
get numbers and compare them, but it looked (looks?) a bit less bad to
me... Or did I make some obvious mistake I'm not seeing right now? :-(

> At the very least this needs to be an exponential allocation, eg
>   +                array_size += nr_nodes + array_size;
> 
> But if you didn't insist on creating the whole list and sorting it,
> you would avoid this allocation entirely, wouldn't you ?
> 
I'll kill the separation between candidate identification and sorting:
that's easy, quick, and will make George happy as well. :-)

> Should we bve worried that this algorithm will be too slow even if it
> involves just
>   O( C(nr_nodes,min_nodes) )
> iterations ?
> 
I'm commenting about this in the other thread you opened...

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)



[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

[PATCH 2 of 3 v4/leftover] libxl: have NUMA placement deal with cpupools

[Dario Faggioli]

In such a way that only the cpus belonging to the cpupool of the
domain being placed are considered for the placement itself.

This happens by filtering out all the nodes in which the cpupool has
not any cpu from the placement candidates. After that -- as a cpu pooling
not necessarily happens at NUMA nodes boundaries -- we also make sure
only the actual cpus that are part of the pool are considered when
counting how much processors a placement candidate provides.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

---
Changes from v3:
 * fixed a double free on non-NUMA systems (namely, when just 1 node exists).
 * Changed the filtering logic: now only the meaningful combinations and
   nodemap are being considered.

Changes from v2:
 * fixed typos in comments.

diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -132,25 +132,29 @@ static int numa_cmpf(const void *v1, con
 }
 
 /* The actual automatic NUMA placement routine */
-static int numa_place_domain(libxl__gc *gc, libxl_domain_build_info *info)
+static int numa_place_domain(libxl__gc *gc, uint32_t domid,
+                             libxl_domain_build_info *info)
 {
     int nr_candidates = 0;
     libxl__numa_candidate *candidates = NULL;
     libxl_bitmap candidate_nodemap;
-    libxl_cpupoolinfo *pinfo;
-    int nr_pools, rc = 0;
+    libxl_cpupoolinfo cpupool_info;
+    int i, cpupool, rc = 0;
     uint32_t memkb;
 
     libxl_bitmap_init(&candidate_nodemap);
 
-    /* First of all, if cpupools are in use, better not to mess with them */
-    pinfo = libxl_list_cpupool(CTX, &nr_pools);
-    if (!pinfo)
-        return ERROR_FAIL;
-    if (nr_pools > 1) {
-        LOG(NOTICE, "skipping NUMA placement as cpupools are in use");
-        goto out;
-    }
+    /*
+     * Extract the cpumap from the cpupool the domain belong to. In fact,
+     * it only makes sense to consider the cpus/nodes that are in there
+     * for placement.
+     */
+    rc = cpupool = libxl__domain_cpupool(gc, domid);
+    if (rc < 0)
+        return rc;
+    rc = libxl_cpupool_info(CTX, &cpupool_info, cpupool);
+    if (rc)
+        return rc;
 
     rc = libxl_domain_need_memory(CTX, info, &memkb);
     if (rc)
@@ -162,7 +166,8 @@ static int numa_place_domain(libxl__gc *
 
     /* Find all the candidates with enough free memory and at least
      * as much pcpus as the domain has vcpus.  */
-    rc = libxl__get_numa_candidates(gc, memkb, info->max_vcpus, 0, 0,
+    rc = libxl__get_numa_candidates(gc, memkb, info->max_vcpus,
+                                    0, 0, &cpupool_info.cpumap,
                                     &candidates, &nr_candidates);
     if (rc)
         goto out;
@@ -188,13 +193,20 @@ static int numa_place_domain(libxl__gc *
     if (rc)
         goto out;
 
+    /* Avoid trying to set the affinity to cpus that might be in the
+     * nodemap but not in our cpupool. */
+    libxl_for_each_set_bit(i, info->cpumap) {
+        if (!libxl_bitmap_test(&cpupool_info.cpumap, i))
+            libxl_bitmap_reset(&info->cpumap, i);
+    }
+
     LOG(DETAIL, "NUMA placement candidate with %d nodes, %d cpus and "
                 "%"PRIu32" KB free selected", candidates[0].nr_nodes,
                 candidates[0].nr_cpus, candidates[0].free_memkb / 1024);
 
  out:
     libxl_bitmap_dispose(&candidate_nodemap);
-    libxl_cpupoolinfo_list_free(pinfo, nr_pools);
+    libxl_cpupoolinfo_dispose(&cpupool_info);
     return rc;
 }
 
@@ -217,7 +229,7 @@ int libxl__build_pre(libxl__gc *gc, uint
      * whatever that turns out to be.
      */
     if (libxl_bitmap_is_full(&info->cpumap)) {
-        int rc = numa_place_domain(gc, info);
+        int rc = numa_place_domain(gc, domid, info);
         if (rc)
             return rc;
     }
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -2289,14 +2289,17 @@ typedef struct {
  * least that amount of free memory and that number of cpus, respectively. If
  * min_free_memkb and/or min_cpus are 0, the candidates' free memory and number
  * of cpus won't be checked at all, which means a candidate will always be
- * considered suitable wrt the specific constraint.  cndts is where the list of
- * exactly nr_cndts candidates is returned. Note that, in case no candidates
- * are found at all, the function returns successfully, but with nr_cndts equal
- * to zero.
+ * considered suitable wrt the specific constraint. suitable_cpumap is useful
+ * for specifying we want only the cpus in that mask to be considered while
+ * generating placement candidates (for example because of cpupools). cndts is
+ * where the list of exactly nr_cndts candidates is returned. Note that, in
+ * case no candidates are found at all, the function returns successfully, but
+ * with nr_cndts equal to zero.
  */
 _hidden int libxl__get_numa_candidates(libxl__gc *gc,
                                 uint32_t min_free_memkb, int min_cpus,
                                 int min_nodes, int max_nodes,
+                                const libxl_bitmap *suitable_cpumap,
                                 libxl__numa_candidate *cndts[], int *nr_cndts);
 
 /* Initialization, allocation and deallocation for placement candidates */
diff --git a/tools/libxl/libxl_numa.c b/tools/libxl/libxl_numa.c
--- a/tools/libxl/libxl_numa.c
+++ b/tools/libxl/libxl_numa.c
@@ -107,30 +107,48 @@ static int comb_next(comb_iter_t it, int
 /* NUMA automatic placement (see libxl_internal.h for details) */
 
 /*
- * This function turns a k-combination iterator into a node map.
- * This means the bits in the node map corresponding to the indexes
- * of the given combination are the ones that will be set.
- * For example, if the iterator represents the combination { 0, 2, 4},
- * the node map will have bits #0, #2 and #4 set.
+ * This function turns a k-combination iterator into a node map,
+ * given another map, telling us which nodes should be considered.
+ *
+ * This means the bits that are set in suitable_nodemap and that
+ * corresponds to the indexes of the given combination are the ones
+ * that will be set in nodemap.
+ *
+ * For example, given a fully set suitable_nodemap, if the iterator
+ * represents the combination { 0, 2, 4}, nodmeap will have bits #0,
+ * #2 and #4 set.
+ * On the other hand, if, say,  suitable_nodemap=01011011, the same
+ * iterator will cause bits #1, #4 and #7 of nodemap to be set.
  */
-static void comb_get_nodemap(comb_iter_t it, libxl_bitmap *nodemap, int k)
+static void comb_get_nodemap(comb_iter_t it, libxl_bitmap *suitable_nodemap,
+                             libxl_bitmap *nodemap, int k)
 {
-    int i;
+    int i, m = 0, n = 0;
 
     libxl_bitmap_set_none(nodemap);
-    for (i = 0; i < k; i++)
-        libxl_bitmap_set(nodemap, it[i]);
+    libxl_for_each_set_bit(i, *suitable_nodemap) {
+        /* Check wether the n-th set bit of suitable_nodemap
+         * matches with the m-th element of the iterator (and,
+         * only if it does, advance to the next one) */
+        if (m < k && n == it[m]) {
+            libxl_bitmap_set(nodemap, i);
+            m++;
+        }
+        n++;
+    }
 }
 
 /* Retrieve the number of cpus that the nodes that are part of the nodemap
- * span. */
+ * span and are also set in suitable_cpumap. */
 static int nodemap_to_nr_cpus(libxl_cputopology *tinfo, int nr_cpus,
+                              const libxl_bitmap *suitable_cpumap,
                               const libxl_bitmap *nodemap)
 {
     int i, nodes_cpus = 0;
 
     for (i = 0; i < nr_cpus; i++) {
-        if (libxl_bitmap_test(nodemap, tinfo[i].node))
+        if (libxl_bitmap_test(suitable_cpumap, i) &&
+            libxl_bitmap_test(nodemap, tinfo[i].node))
             nodes_cpus++;
     }
     return nodes_cpus;
@@ -236,15 +254,17 @@ static int cpus_per_node_count(libxl_cpu
 int libxl__get_numa_candidates(libxl__gc *gc,
                                uint32_t min_free_memkb, int min_cpus,
                                int min_nodes, int max_nodes,
+                               const libxl_bitmap *suitable_cpumap,
                                libxl__numa_candidate *cndts[], int *nr_cndts)
 {
     libxl__numa_candidate *new_cndts = NULL;
     libxl_cputopology *tinfo = NULL;
     libxl_numainfo *ninfo = NULL;
     int nr_nodes = 0, nr_cpus = 0;
-    libxl_bitmap nodemap;
+    libxl_bitmap suitable_nodemap, nodemap;
     int array_size, rc;
 
+    libxl_bitmap_init(&suitable_nodemap);
     libxl_bitmap_init(&nodemap);
 
     /* Get platform info and prepare the map for testing the combinations */
@@ -262,6 +282,15 @@ int libxl__get_numa_candidates(libxl__gc
     if (rc)
         goto out;
 
+    /* Allocate and prepare the map of the node that can be utilized for
+     * placement, basing on the map of suitable cpus. */
+    rc = libxl_node_bitmap_alloc(CTX, &suitable_nodemap, 0);
+    if (rc)
+        goto out;
+    rc = libxl_cpumap_to_nodemap(CTX, suitable_cpumap, &suitable_nodemap);
+    if (rc)
+        goto out;
+
     /*
      * If the minimum number of NUMA nodes is not explicitly specified
      * (i.e., min_nodes == 0), we try to figure out a sensible number of nodes
@@ -287,6 +316,13 @@ int libxl__get_numa_candidates(libxl__gc
         rc = ERROR_INVAL;
         goto out;
     }
+    /* We also need to be sure we do not exceed the number of
+     * nodes we are allowed to use. */
+    if (libxl_bitmap_count_set(&suitable_nodemap) < max_nodes) {
+        max_nodes = libxl_bitmap_count_set(&suitable_nodemap);
+        if (min_nodes > max_nodes)
+            min_nodes = max_nodes;
+    }
 
     /* Initialize the local storage for the combinations */
     *nr_cndts = 0;
@@ -311,7 +347,10 @@ int libxl__get_numa_candidates(libxl__gc
             uint32_t nodes_free_memkb;
             int nodes_cpus;
 
-            comb_get_nodemap(comb_iter, &nodemap, min_nodes);
+            /* Get the nodemap for the combination, only considering
+             * suitable nodes. */
+            comb_get_nodemap(comb_iter, &suitable_nodemap,
+                             &nodemap, min_nodes);
 
             /* If there is not enough memory in this combination, skip it
              * and go generating the next one... */
@@ -320,7 +359,8 @@ int libxl__get_numa_candidates(libxl__gc
                 continue;
 
             /* And the same applies if this combination is short in cpus */
-            nodes_cpus = nodemap_to_nr_cpus(tinfo, nr_cpus, &nodemap);
+            nodes_cpus = nodemap_to_nr_cpus(tinfo, nr_cpus, suitable_cpumap,
+                                            &nodemap);
             if (min_cpus && nodes_cpus < min_cpus)
                 continue;
 
@@ -345,12 +385,13 @@ int libxl__get_numa_candidates(libxl__gc
             new_cndts[*nr_cndts].nr_domains =
                                     nodemap_to_nr_domains(gc, tinfo, &nodemap);
             new_cndts[*nr_cndts].free_memkb = nodes_free_memkb;
-            new_cndts[*nr_cndts].nr_nodes = min_nodes;
+            new_cndts[*nr_cndts].nr_nodes = libxl_bitmap_count_set(&nodemap);
             new_cndts[*nr_cndts].nr_cpus = nodes_cpus;
 
             LOG(DEBUG, "NUMA placement candidate #%d found: nr_nodes=%d, "
                        "nr_cpus=%d, free_memkb=%"PRIu32"", *nr_cndts,
-                       min_nodes, new_cndts[*nr_cndts].nr_cpus,
+                       new_cndts[*nr_cndts].nr_nodes,
+                       new_cndts[*nr_cndts].nr_cpus,
                        new_cndts[*nr_cndts].free_memkb / 1024);
 
             (*nr_cndts)++;
@@ -360,6 +401,7 @@ int libxl__get_numa_candidates(libxl__gc
 
     *cndts = new_cndts;
  out:
+    libxl_bitmap_dispose(&suitable_nodemap);
     libxl_bitmap_dispose(&nodemap);
     libxl_cputopology_list_free(tinfo, nr_cpus);
     libxl_numainfo_list_free(ninfo, nr_nodes);

[PATCH 3 of 3 v4/leftover] Some automatic NUMA placement documentation

[Dario Faggioli]

About rationale, usage and (some small bits of) API.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

---
Changes from v3:
 * typos and rewording of some sentences, as suggested during review.

Changes from v1:
 * API documentation moved close to the actual functions.

diff --git a/docs/misc/xl-numa-placement.markdown b/docs/misc/xl-numa-placement.markdown
new file mode 100644
--- /dev/null
+++ b/docs/misc/xl-numa-placement.markdown
@@ -0,0 +1,89 @@
+# Guest Automatic NUMA Placement in libxl and xl #
+
+## Rationale ##
+
+NUMA means the memory accessing times of a program running on a CPU depends on
+the relative distance between that CPU and that memory. In fact, most of the
+NUMA systems are built in such a way that each processor has its local memory,
+on which it can operate very fast. On the other hand, getting and storing data
+from and on remote memory (that is, memory local to some other processor) is
+quite more complex and slow. On these machines, a NUMA node is usually defined
+as a set of processor cores (typically a physical CPU package) and the memory
+directly attached to the set of cores.
+
+The Xen hypervisor deals with Non-Uniform Memory Access (NUMA]) machines by
+assigning to its domain a "node affinity", i.e., a set of NUMA nodes of the
+host from which it gets its memory allocated.
+
+NUMA awareness becomes very important as soon as many domains start running
+memory-intensive workloads on a shared host. In fact, the cost of accessing non
+node-local memory locations is very high, and the performance degradation is
+likely to be noticeable.
+
+## Guest Placement in xl ##
+
+If using xl for creating and managing guests, it is very easy to ask for both
+manual or automatic placement of them across the host's NUMA nodes.
+
+Note that xm/xend does the very same thing, the only differences residing in
+the details of the heuristics adopted for the placement (see below).
+
+### Manual Guest Placement with xl ###
+
+Thanks to the "cpus=" option, it is possible to specify where a domain should
+be created and scheduled on, directly in its config file. This affects NUMA
+placement and memory accesses as the hypervisor constructs the node affinity of
+a VM basing right on its CPU affinity when it is created.
+
+This is very simple and effective, but requires the user/system administrator
+to explicitly specify affinities for each and every domain, or Xen won't be
+able to guarantee the locality for their memory accesses.
+
+It is also possible to deal with NUMA by partitioning the system using cpupools
+(available in the upcoming release of Xen, 4.2). Again, this could be "The
+Right Answer" for many needs and occasions, but  has to to be carefully
+considered and manually setup by hand.
+
+### Automatic Guest Placement with xl ###
+
+If no "cpus=" option is specified in the config file, libxl tries to figure out
+on its own on which node(s) the domain could fit best.  It is worthwhile noting
+that optimally fitting a set of VMs on the NUMA nodes of an host is an
+incarnation of the Bin Packing Problem. In fact, the various VMs with different
+memory sizes are the items to be packed, and the host nodes are the bins. That
+is known to be NP-hard, thus. We will therefore be using some heuristics.
+
+The first thing to do is finding  a node, or even a set of nodes, that have
+enough free memory and enough physical CPUs for accommodating the one new
+domain. The idea is to find a spot for the domain with at least as much free
+memory as it has configured, and as much pCPUs as it has vCPUs.  After that,
+the actual decision on which solution to go for happens accordingly to the
+following heuristics:
+
+  *  candidates involving fewer nodes come first. In case two (or more)
+     candidates span the same number of nodes,
+  *  the amount of free memory and the number of domains assigned to the
+     candidates are considered. In doing that, candidates with greater amount
+     of free memory and fewer assigned domains are preferred, with free memory
+     "weighting" three times as much as number of domains.
+
+Giving preference to candidates with fewer nodes ensures better performance for
+the guest, as it avoid spreading its memory among different nodes.  Favouring
+the nodes that have the largest amounts of free memory helps keeping the memory
+fragmentation small, from a system wide perspective.  However, if more
+candidates fulfil these criteria by roughly the same extent, having the number
+of domains the candidates are "hosting" helps balancing the load on the various
+nodes.
+
+## Guest Placement within libxl ##
+
+xl achieves automatic NUMA just because libxl does it interrnally.
+No API is provided (yet) for interacting with this feature and modify
+the library behaviour regarding automatic placement, it just happens
+by default if no affinity is specified (as it is with xm/xend).
+
+For actually looking and maybe tweaking the mechanism and the algorithms it
+uses, all is implemented as a set of libxl internal interfaces and facilities.
+Look at the comment "Automatic NUMA placement" in libxl\_internal.h.
+
+Note this may change in future versions of Xen/libxl.

Re: [PATCH 0 of 3 v4/leftover] Automatic NUMA placement for xl

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 687 bytes --]

On Tue, 2012-07-10 at 17:03 +0200, Dario Faggioli wrote:
> Hello again,
> 
Hi,

This seems to not have been considered yet... If that is so, I'd like to
post a new version, in which I fix a small bug that is presenti in this
version (patch 2/3).

Please, consider that one repost for review or inclusion (subject will
be "[... v5/leftover] Automatic NUMA placement for xl").

Thanks and Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel