QEMU-Devel Archive on lore.kernel.org
 help / color / Atom feed
From: "Richard W.M. Jones" <rjones@redhat.com>
To: Eric Blake <eblake@redhat.com>
Cc: libguestfs@redhat.com, QEMU <qemu-devel@nongnu.org>,
	"qemu-block@nongnu.org" <qemu-block@nongnu.org>,
	nbd@other.debian.org
Subject: Re: [Qemu-devel] [Libguestfs] cross-project patches: Add NBD Fast Zero support
Date: Tue, 27 Aug 2019 13:14:49 +0100
Message-ID: <20190827121449.GX7304@redhat.com> (raw)
In-Reply-To: <25ead363-4f37-5450-b985-1876374e314d@redhat.com>

On Fri, Aug 23, 2019 at 09:30:36AM -0500, Eric Blake wrote:
> I've run several tests to demonstrate why this is useful, as well as
> prove that because I have multiple interoperable projects, it is worth
> including in the NBD standard.  The original proposal was here:
> https://lists.debian.org/nbd/2019/03/msg00004.html
> where I stated:
> 
> > I will not push this without both:
> > - a positive review (for example, we may decide that burning another
> > NBD_FLAG_* is undesirable, and that we should instead have some sort
> > of NBD_OPT_ handshake for determining when the server supports
> > NBD_CMF_FLAG_FAST_ZERO)
> > - a reference client and server implementation (probably both via qemu,
> > since it was qemu that raised the problem in the first place)

Is the plan to wait until NBD_CMF_FLAG_FAST_ZERO gets into the NBD
protocol doc before doing the rest?  Also I would like to release both
libnbd 1.0 and nbdkit 1.14 before we introduce any large new features.
Both should be released this week, in fact maybe even today or
tomorrow.

[...]
> First, I had to create a scenario where falling back to writes is
> noticeably slower than performing a zero operation, and where
> pre-zeroing also shows an effect.  My choice: let's test 'qemu-img
> convert' on an image that is half-sparse (every other megabyte is a
> hole) to an in-memory nbd destination.  Then I use a series of nbdkit
> filters to force the destination to behave in various manners:
>  log logfile=>(sed ...|uniq -c) (track how many normal/fast zero
> requests the client makes)
>  nozero $params (fine-tune how zero requests behave - the parameters
> zeromode and fastzeromode are the real drivers of my various tests)
>  blocksize maxdata=256k (allows large zero requests, but forces large
> writes into smaller chunks, to magnify the effects of write delays and
> allow testing to provide obvious results with a smaller image)
>  delay delay-write=20ms delay-zero=5ms (also to magnify the effects on a
> smaller image, with writes penalized more than zeroing)
>  stats statsfile=/dev/stderr (to track overall time and a decent summary
> of how much I/O occurred).
>  noextents (forces the entire image to report that it is allocated,
> which eliminates any testing variability based on whether qemu-img uses
> that to bypass a zeroing operation [1])

I can't help thinking that a sh plugin might have been simpler ...

> I hope you enjoyed reading this far, and agree with my interpretation of
> the numbers about why this feature is useful!

Yes it seems reasonable.

The only thought I had is whether the qemu block layer does or should
combine requests in flight so that a write-zero (offset) followed by a
write-data (same offset) would erase the earlier request.  In some
circumstances that might provide a performance improvement without
needing any changes to protocols.

> - NBD should have a way to advertise (probably via NBD_INFO_ during
> NBD_OPT_GO) if the initial image is known to begin life with all zeroes
> (if that is the case, qemu-img can skip the extents calls and
> pre-zeroing pass altogether)

Yes, I really think we should do this one as well.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-top is 'top' for virtual machines.  Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top


  parent reply index

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-23 14:30 [Qemu-devel] " Eric Blake
2019-08-23 14:34 ` [Qemu-devel] [PATCH 0/1] NBD protocol change to add fast zero support Eric Blake
2019-08-23 14:34   ` [Qemu-devel] [PATCH 1/1] protocol: Add NBD_CMD_FLAG_FAST_ZERO Eric Blake
2019-08-23 18:48     ` Wouter Verhelst
2019-08-23 18:58       ` Eric Blake
2019-08-24  6:44         ` Wouter Verhelst
2019-08-28  9:57     ` Vladimir Sementsov-Ogievskiy
2019-08-28 13:04       ` Eric Blake
2019-08-28 13:45         ` Vladimir Sementsov-Ogievskiy
2019-09-03 20:53   ` [Qemu-devel] [Libguestfs] [PATCH 0/1] NBD protocol change to add fast zero support Eric Blake
2019-08-23 14:37 ` [Qemu-devel] [PATCH 0/5] Add NBD fast zero support to qemu client and server Eric Blake
2019-08-23 14:37   ` [Qemu-devel] [PATCH 1/5] nbd: Improve per-export flag handling in server Eric Blake
2019-08-30 18:00     ` Vladimir Sementsov-Ogievskiy
2019-08-30 23:10       ` Eric Blake
2019-08-30 23:32         ` Eric Blake
2019-09-03 16:39           ` Eric Blake
2019-09-04 17:08     ` Vladimir Sementsov-Ogievskiy
2019-08-23 14:37   ` [Qemu-devel] [PATCH 2/5] nbd: Prepare for NBD_CMD_FLAG_FAST_ZERO Eric Blake
2019-08-30 18:07     ` Vladimir Sementsov-Ogievskiy
2019-08-30 23:37       ` Eric Blake
2019-08-31  8:11         ` Vladimir Sementsov-Ogievskiy
2019-09-03 18:49       ` Eric Blake
2019-08-31  8:20     ` Vladimir Sementsov-Ogievskiy
2019-08-23 14:37   ` [Qemu-devel] [PATCH 3/5] nbd: Implement client use of NBD FAST_ZERO Eric Blake
2019-08-30 18:11     ` Vladimir Sementsov-Ogievskiy
2019-08-23 14:37   ` [Qemu-devel] [PATCH 4/5] nbd: Implement server " Eric Blake
2019-08-30 18:40     ` Vladimir Sementsov-Ogievskiy
2019-08-23 14:37   ` [Qemu-devel] [PATCH 5/5] nbd: Tolerate more errors to structured reply request Eric Blake
2019-08-23 16:41     ` Eric Blake
2019-08-28 13:55   ` [Qemu-devel] [PATCH 0/5] Add NBD fast zero support to qemu client and server Vladimir Sementsov-Ogievskiy
2019-08-28 14:05     ` Eric Blake
2019-08-23 14:38 ` [Qemu-devel] [libnbd PATCH 0/1] libnbd support for new fast zero Eric Blake
2019-08-23 14:38   ` [Qemu-devel] [libnbd PATCH 1/1] api: Add support for FAST_ZERO flag Eric Blake
2019-08-27 12:25     ` [Qemu-devel] [Libguestfs] " Richard W.M. Jones
2019-08-23 14:40 ` [Qemu-devel] [nbdkit PATCH 0/3] nbdkit support for new NBD fast zero Eric Blake
2019-08-23 14:40   ` [Qemu-devel] [nbdkit PATCH 1/3] server: Add internal support for NBDKIT_FLAG_FAST_ZERO Eric Blake
2019-08-23 14:40   ` [Qemu-devel] [nbdkit PATCH 2/3] filters: Add .can_fast_zero hook Eric Blake
2019-08-23 14:40   ` [Qemu-devel] [nbdkit PATCH 3/3] plugins: " Eric Blake
2019-08-23 21:16     ` [Qemu-devel] [Libguestfs] " Eric Blake
2019-08-27 15:43     ` Richard W.M. Jones
2019-08-23 15:05 ` [Qemu-devel] cross-project patches: Add NBD Fast Zero support Vladimir Sementsov-Ogievskiy
2019-08-27 12:14 ` Richard W.M. Jones [this message]
2019-08-27 13:23   ` [Qemu-devel] [Libguestfs] " Eric Blake

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190827121449.GX7304@redhat.com \
    --to=rjones@redhat.com \
    --cc=eblake@redhat.com \
    --cc=libguestfs@redhat.com \
    --cc=nbd@other.debian.org \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

QEMU-Devel Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/qemu-devel/0 qemu-devel/git/0.git
	git clone --mirror https://lore.kernel.org/qemu-devel/1 qemu-devel/git/1.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 qemu-devel qemu-devel/ https://lore.kernel.org/qemu-devel \
		qemu-devel@nongnu.org qemu-devel@archiver.kernel.org
	public-inbox-index qemu-devel

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.nongnu.qemu-devel


AGPL code for this site: git clone https://public-inbox.org/ public-inbox