qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
To: qemu-block@nongnu.org
Cc: kwolf@redhat.com, vsementsov@virtuozzo.com, armbru@redhat.com,
	qemu-devel@nongnu.org, mreitz@redhat.com, den@openvz.org
Subject: [PATCH 0/7] qcow2: compressed write cache
Date: Fri, 29 Jan 2021 19:50:23 +0300	[thread overview]
Message-ID: <20210129165030.640169-1-vsementsov@virtuozzo.com> (raw)

Hi all!

I know, I have several series waiting for a resend, but I had to switch
to another task spawned from our customer's bug.

Original problem: we use O_DIRECT for all vm images in our product, it's
the policy. The only exclusion is backup target qcow2 image for
compressed backup, because compressed backup is extremely slow with
O_DIRECT (due to unaligned writes). Customer complains that backup
produces a lot of pagecache.

So we can either implement some internal cache or use fadvise somehow.
Backup has several async workes, which writes simultaneously, so in both
ways we have to track host cluster filling (before dropping the cache
corresponding to the cluster).  So, if we have to track anyway, let's
try to implement the cache.

Idea is simple: cache small unaligned write and flush the cluster when
filled.

Performance result is very good (results in a table is time of
compressed backup of 1000M disk filled with ones in seconds):

---------------  -----------  -----------
                 backup(old)  backup(new)
ssd:hdd(direct)  3e+02        4.4
                                -99%
ssd:hdd(cached)  5.7          5.4
                                -5%
---------------  -----------  -----------

So, we have benefit even for cached mode! And the fastest thing is
O_DIRECT with new implemented cache. So, I suggest to enable the new
cache by default (which is done by the series).

Command to generate performance comparison table:

 ./scripts/simplebench/bench-backup.py --compressed --target-cache both --dir ssd:/ssd --dir hdd:/work --test ssd:hdd --env old:/work/src/qemu/master/build/qemu-system-x86_64 new:/work/src/qemu/up/qcow2-compressed-write-cache/build/qemu-system-x86_6

Vladimir Sementsov-Ogievskiy (7):
  qemu/queue: add some useful QLIST_ and QTAILQ_ macros
  block/qcow2: introduce cache for compressed writes
  block/qcow2: use compressed write cache
  simplebench: bench_one(): add slow_limit argument
  simplebench: bench_one(): support count=1
  simplebench/bench-backup: add --compressed option
  simplebench/bench-backup: add target-cache argument

 qapi/block-core.json                   |   8 +-
 block/qcow2.h                          |  33 ++
 include/qemu/queue.h                   |  14 +
 block/qcow2-compressed-write-cache.c   | 770 +++++++++++++++++++++++++
 block/qcow2-refcount.c                 |  13 +
 block/qcow2.c                          |  87 ++-
 block/meson.build                      |   1 +
 scripts/simplebench/bench-backup.py    |  74 ++-
 scripts/simplebench/bench_block_job.py |  33 +-
 scripts/simplebench/simplebench.py     |  34 +-
 10 files changed, 1039 insertions(+), 28 deletions(-)
 create mode 100644 block/qcow2-compressed-write-cache.c

-- 
2.29.2



             reply	other threads:[~2021-01-29 17:06 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-29 16:50 Vladimir Sementsov-Ogievskiy [this message]
2021-01-29 16:50 ` [PATCH 1/7] qemu/queue: add some useful QLIST_ and QTAILQ_ macros Vladimir Sementsov-Ogievskiy
2021-02-01  8:29   ` Markus Armbruster
2021-02-01  8:34     ` Vladimir Sementsov-Ogievskiy
2021-02-10 17:07   ` Max Reitz
2021-01-29 16:50 ` [PATCH 2/7] block/qcow2: introduce cache for compressed writes Vladimir Sementsov-Ogievskiy
2021-02-10 17:07   ` Max Reitz
2021-02-11 12:49     ` Vladimir Sementsov-Ogievskiy
2021-02-18 15:04       ` Max Reitz
2021-01-29 16:50 ` [PATCH 3/7] block/qcow2: use compressed write cache Vladimir Sementsov-Ogievskiy
2021-02-10 17:11   ` Max Reitz
2021-02-11 12:53     ` Vladimir Sementsov-Ogievskiy
2021-02-18 16:02       ` Max Reitz
2021-01-29 16:50 ` [PATCH 4/7] simplebench: bench_one(): add slow_limit argument Vladimir Sementsov-Ogievskiy
2021-01-29 16:50 ` [PATCH 5/7] simplebench: bench_one(): support count=1 Vladimir Sementsov-Ogievskiy
2021-01-29 16:50 ` [PATCH 6/7] simplebench/bench-backup: add --compressed option Vladimir Sementsov-Ogievskiy
2021-01-29 16:50 ` [PATCH 7/7] simplebench/bench-backup: add target-cache argument Vladimir Sementsov-Ogievskiy
2021-01-29 17:30 ` [PATCH 0/7] qcow2: compressed write cache no-reply
2021-02-01  8:24   ` Vladimir Sementsov-Ogievskiy
2021-02-09 13:25 ` Max Reitz
2021-02-09 14:10   ` Vladimir Sementsov-Ogievskiy
2021-02-09 14:47     ` Max Reitz
2021-02-09 16:39       ` Vladimir Sementsov-Ogievskiy
2021-02-09 18:36         ` Vladimir Sementsov-Ogievskiy
2021-02-09 18:41           ` Denis V. Lunev
2021-02-09 18:51             ` Vladimir Sementsov-Ogievskiy
2021-02-10 10:00               ` Max Reitz
2021-02-10 10:10                 ` Vladimir Sementsov-Ogievskiy
2021-02-09 16:52       ` Denis V. Lunev
2021-02-10 10:00         ` Max Reitz
2021-02-10 12:35 ` Kevin Wolf
2021-02-10 14:35   ` Vladimir Sementsov-Ogievskiy

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=20210129165030.640169-1-vsementsov@virtuozzo.com \
    --to=vsementsov@virtuozzo.com \
    --cc=armbru@redhat.com \
    --cc=den@openvz.org \
    --cc=kwolf@redhat.com \
    --cc=mreitz@redhat.com \
    --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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).