| Message ID | 20200611171143.21589-1-den@openvz.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | block: seriously improve savevm performance | expand |
* Denis V. Lunev (den@openvz.org) wrote: > This series do standard basic things: > - it creates intermediate buffer for all writes from QEMU migration code > to QCOW2 image, > - this buffer is sent to disk asynchronously, allowing several writes to > run in parallel. > > In general, migration code is fantastically inefficent (by observation), > buffers are not aligned and sent with arbitrary pieces, a lot of time > less than 100 bytes at a chunk, which results in read-modify-write > operations with non-cached operations. It should also be noted that all > operations are performed into unallocated image blocks, which also suffer > due to partial writes to such new clusters. It surprises me a little that you're not benefiting from the buffer internal to qemu-file.c Dave > This patch series is an implementation of idea discussed in the RFC > posted by Denis Plotnikov > https://lists.gnu.org/archive/html/qemu-devel/2020-04/msg01925.html > Results with this series over NVME are better than original code > original rfc this > cached: 1.79s 2.38s 1.27s > non-cached: 3.29s 1.31s 0.81s > > Changes from v2: > - code moved from QCOW2 level to generic block level > - created bdrv_flush_vmstate helper to fix 022, 029 tests > - added recursive for bs->file in bdrv_co_flush_vmstate (fix 267) > - fixed blk_save_vmstate helper > - fixed coroutine wait as Vladimir suggested with waiting fixes from me > > Changes from v1: > - patchew warning fixed > - fixed validation that only 1 waiter is allowed in patch 1 > > Signed-off-by: Denis V. Lunev <den@openvz.org> > CC: Kevin Wolf <kwolf@redhat.com> > CC: Max Reitz <mreitz@redhat.com> > CC: Stefan Hajnoczi <stefanha@redhat.com> > CC: Fam Zheng <fam@euphon.net> > CC: Juan Quintela <quintela@redhat.com> > CC: "Dr. David Alan Gilbert" <dgilbert@redhat.com> > CC: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> > CC: Denis Plotnikov <dplotnikov@virtuozzo.com> > > -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
On 6/15/20 3:17 PM, Dr. David Alan Gilbert wrote: > * Denis V. Lunev (den@openvz.org) wrote: >> This series do standard basic things: >> - it creates intermediate buffer for all writes from QEMU migration code >> to QCOW2 image, >> - this buffer is sent to disk asynchronously, allowing several writes to >> run in parallel. >> >> In general, migration code is fantastically inefficent (by observation), >> buffers are not aligned and sent with arbitrary pieces, a lot of time >> less than 100 bytes at a chunk, which results in read-modify-write >> operations with non-cached operations. It should also be noted that all >> operations are performed into unallocated image blocks, which also suffer >> due to partial writes to such new clusters. > It surprises me a little that you're not benefiting from the buffer > internal to qemu-file.c > > Dave There are a lot of problems with this buffer: Flushes to block driver state are performed in the abstract places, pushing a) small IO b) non-aligned IO both to 1) page size 2) cluster size It should also be noted that buffer in QEMU file is quite small and all IO operations with it are synchronous. IO, like ethernet, wants good queues. The difference is on the table. Den
* Denis V. Lunev (den@openvz.org) wrote: > On 6/15/20 3:17 PM, Dr. David Alan Gilbert wrote: > > * Denis V. Lunev (den@openvz.org) wrote: > >> This series do standard basic things: > >> - it creates intermediate buffer for all writes from QEMU migration code > >> to QCOW2 image, > >> - this buffer is sent to disk asynchronously, allowing several writes to > >> run in parallel. > >> > >> In general, migration code is fantastically inefficent (by observation), > >> buffers are not aligned and sent with arbitrary pieces, a lot of time > >> less than 100 bytes at a chunk, which results in read-modify-write > >> operations with non-cached operations. It should also be noted that all > >> operations are performed into unallocated image blocks, which also suffer > >> due to partial writes to such new clusters. > > It surprises me a little that you're not benefiting from the buffer > > internal to qemu-file.c > > > > Dave > There are a lot of problems with this buffer: > > Flushes to block driver state are performed in the abstract places, > pushing > Â a) small IO > Â b) non-aligned IO both to > Â Â Â Â Â Â 1) page size > Â Â Â Â Â Â 2) cluster size > It should also be noted that buffer in QEMU file is quite small and > all IO operations with it are synchronous. IO, like ethernet, wants > good queues. Yeh, for ethernet we immediately get the kernels buffer so it's not too bad; and I guess the async page writes are easier as well. Dave > The difference is on the table. > > Den > -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
This series do standard basic things: - it creates intermediate buffer for all writes from QEMU migration code to QCOW2 image, - this buffer is sent to disk asynchronously, allowing several writes to run in parallel. In general, migration code is fantastically inefficent (by observation), buffers are not aligned and sent with arbitrary pieces, a lot of time less than 100 bytes at a chunk, which results in read-modify-write operations with non-cached operations. It should also be noted that all operations are performed into unallocated image blocks, which also suffer due to partial writes to such new clusters. This patch series is an implementation of idea discussed in the RFC posted by Denis Plotnikov https://lists.gnu.org/archive/html/qemu-devel/2020-04/msg01925.html Results with this series over NVME are better than original code original rfc this cached: 1.79s 2.38s 1.27s non-cached: 3.29s 1.31s 0.81s Changes from v2: - code moved from QCOW2 level to generic block level - created bdrv_flush_vmstate helper to fix 022, 029 tests - added recursive for bs->file in bdrv_co_flush_vmstate (fix 267) - fixed blk_save_vmstate helper - fixed coroutine wait as Vladimir suggested with waiting fixes from me Changes from v1: - patchew warning fixed - fixed validation that only 1 waiter is allowed in patch 1 Signed-off-by: Denis V. Lunev <den@openvz.org> CC: Kevin Wolf <kwolf@redhat.com> CC: Max Reitz <mreitz@redhat.com> CC: Stefan Hajnoczi <stefanha@redhat.com> CC: Fam Zheng <fam@euphon.net> CC: Juan Quintela <quintela@redhat.com> CC: "Dr. David Alan Gilbert" <dgilbert@redhat.com> CC: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com> CC: Denis Plotnikov <dplotnikov@virtuozzo.com>