All of lore.kernel.org
 help / color / mirror / Atom feed
From: Milosz Tanski <milosz@adfin.com>
To: Jeff Moyer <jmoyer@redhat.com>
Cc: lsf-pc@lists.linux-foundation.org,
	"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
	linux-mm@kvack.org, Christoph Hellwig <hch@infradead.org>
Subject: Re: [LSF/MM TOPIC] async buffered diskio read for userspace apps
Date: Fri, 16 Jan 2015 10:40:23 -0500	[thread overview]
Message-ID: <CANP1eJF0m-48--2ysYgymgBXES1nCefU-06SxZOv0hKzia8AUg@mail.gmail.com> (raw)
In-Reply-To: <x49zj9jaocy.fsf@segfault.boston.devel.redhat.com>

On Thu, Jan 15, 2015 at 1:30 PM, Jeff Moyer <jmoyer@redhat.com> wrote:
> Milosz Tanski <milosz@adfin.com> writes:
>
>> I would like to talk about enhancing the user interfaces for doing
>> async buffered disk IO for userspace applications. There's a whole
>> class of distributed web applications (most new applications today)
>> that would benefit from such an API. Most of them today rely on
>> cobbling one together in user space using a threadpool.
>>
>> The current in kernel AIO interfaces that only support DIRECTIO, they
>> were generally designed by and for big database vendors. The consensus
>> is that the current AIO interfaces usually lead to decreased
>> performance for those app.
>>
>> I've been developing a new read syscall that allows non-blocking
>> diskio read (provided that data is in the page cache). It's analogous
>> to what exists today in the network world with recvmsg with MSG_NOWAIT
>> flag. The work has been previously described by LWN here:
>> https://lwn.net/Articles/612483/
>>
>> Previous attempts (over the last 12+ years) at non-blocking buffered
>> diskio has stalled due to their complexity. I would like to talk about
>> the problem, my solution, and get feedback on the course of action.
>
> This email seems to conflate async I/O and non-blocking I/O.  Could you
> please be more specific about what you're proposing to talk about?  Is
> it just the non-blocking read support?
>
> Cheers,
> Jeff

Jeff, I'm sorry if I wasn't clear, let me restate why we should care
and why it matters.

The current applications that power the lower levels of the web stacks
as generally process streams of network data. Many of them (and the
frameworks for building them) are structured as a large async
processing loop. Disk IO a big pain point; the way are structured
(threadpools for diskio) introduces additional latency. sendfile() is
only helpful if you need to do additional processing (say SSL).

Non-blocking diskio can help us lower the response latency in those
webapps applications in the common cases (cached data, sequential
scan).

-- 
Milosz Tanski
CTO
16 East 34th Street, 15th floor
New York, NY 10016

p: 646-253-9055
e: milosz@adfin.com

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  reply	other threads:[~2015-01-16 15:40 UTC|newest]

Thread overview: 59+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-01-15 17:43 [LSF/MM TOPIC] async buffered diskio read for userspace apps Milosz Tanski
2015-01-15 18:30 ` Jeff Moyer
2015-01-15 18:30   ` Jeff Moyer
2015-01-16 15:40   ` Milosz Tanski [this message]
2015-01-15 22:31 ` [Lsf-pc] " Jan Kara
2015-01-16 15:44   ` Milosz Tanski
2015-01-16 16:55     ` Jeremy Allison
2015-01-16 16:55       ` Jeremy Allison
2015-01-19  3:49       ` Milosz Tanski
2015-01-19  7:12         ` Jeremy Allison
2015-01-19  7:12           ` Jeremy Allison
2015-01-19  7:34           ` James Bottomley
2015-01-19  7:34             ` James Bottomley
2015-01-19 14:18             ` Milosz Tanski
2015-01-19 15:31               ` Jens Axboe
2015-01-19 15:31                 ` Jens Axboe
2015-01-19 15:49                 ` James Bottomley
2015-01-19 15:49                   ` James Bottomley
2015-01-19 16:48                   ` Jeremy Allison
2015-01-19 16:48                     ` Jeremy Allison
2015-01-19 17:26                     ` James Bottomley
2015-01-19 17:26                       ` James Bottomley
2015-01-19 17:32                       ` Jeremy Allison
2015-01-19 17:36                       ` Richard Sharpe
2015-01-19 20:00                     ` Jens Axboe
2015-01-19 20:00                       ` Jens Axboe
2015-01-19 16:10                 ` Volker Lendecke
2015-01-19 16:10                   ` Volker Lendecke
2015-01-19 16:20                   ` Milosz Tanski
2015-01-19 16:20                     ` Milosz Tanski
2015-01-19 16:50                     ` Jeremy Allison
2015-01-19 16:50                       ` Jeremy Allison
2015-01-23 23:15                     ` Steve French
2015-01-23 23:15                       ` Steve French
2015-01-24  2:17                       ` Jens Axboe
2015-01-24  2:17                         ` Jens Axboe
2015-01-24  5:53                         ` Jeremy Allison
2015-01-24  5:53                           ` Jeremy Allison
2015-01-24 23:00                           ` Jens Axboe
2015-01-24 23:00                             ` Jens Axboe
2015-01-20 19:33                 ` Jens Axboe
2015-01-20 19:33                   ` Jens Axboe
2015-01-20 23:07                   ` Milosz Tanski
2015-01-20 23:07                     ` Milosz Tanski
2015-01-20 23:22                     ` Milosz Tanski
2015-01-20 23:26                       ` Jens Axboe
2015-01-20 23:26                         ` Jens Axboe
2015-01-20 23:30                         ` Jens Axboe
2015-01-20 23:30                           ` Jens Axboe
2015-01-20 23:53                           ` Milosz Tanski
2015-01-20 23:53                             ` Milosz Tanski
2015-01-21 21:17                             ` Jens Axboe
2015-01-23  0:03                               ` Milosz Tanski
2015-01-23  0:10                               ` Milosz Tanski
2015-01-19 17:04               ` Jeremy Allison
2015-01-19 17:04                 ` Jeremy Allison
2015-01-19 17:18                 ` Richard Sharpe
2015-01-19 17:20                   ` Jeremy Allison
2015-01-19 17:20                     ` Jeremy Allison

Reply instructions:

You may reply publicly 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=CANP1eJF0m-48--2ysYgymgBXES1nCefU-06SxZOv0hKzia8AUg@mail.gmail.com \
    --to=milosz@adfin.com \
    --cc=hch@infradead.org \
    --cc=jmoyer@redhat.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=lsf-pc@lists.linux-foundation.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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.