Ceph Luminous backup improvement?

[jeffwadsworth]

I am using the Proxmox 5 beta with ceph luminous configured in a 3 node cluster. Works very well. I noticed that backups of the VM's (housed in the RBD) seem to work much faster now. Am I dreaming or did something get fixed?

INFO: status: 73% (3962830848/5368709120), sparse 73% (3942830080), duration 33, 122/0 MB/s
INFO: status: 80% (4326490112/5368709120), sparse 80% (4306481152), duration 36, 121/0 MB/s
INFO: status: 87% (4706336768/5368709120), sparse 87% (4686327808), duration 39, 126/0 MB/s
INFO: status: 94% (5085200384/5368709120), sparse 94% (5065191424), duration 42, 126/0 MB/s
INFO: status: 100% (5368709120/5368709120), sparse 99% (5348700160), duration 45, 94/0 MB/s
INFO: transferred 5368 MB in 45 seconds (119 MB/s)
INFO: archive file size: 17MB
INFO: Finished Backup of VM 100 (00:00:47)
INFO: Backup job finished successfully
TASK OK

 

[e100]

Is this with or without KRBD?

 

[spirit]

qemu 2.9 + librbd have an improvement, about memory copy inside librbd. (this need luminious cluster + liminous librbd)

http://tracker.ceph.com/issues/13025

http://git.qemu.org/?p=qemu.git;a=commit;h=1d393bdeae22fde2cb83c1ea719675747c85c40e


don't have tested it yet, but it should help for small iops.

 

[spirit]

check also /etc/pve/ceph.conf, some tuning has been done on debug features

 

[jeffwadsworth]

Is this with or without KRBD?

This is without KRBD.