Ask Your Question

Glance performances when uploading snapshots to a ceph backend

asked 2015-07-06 08:34:45 -0500

jpmethot gravatar image

I noticed some performance irregularities regarding when Glance uploads a snapshot to a ceph backend. Basically, the upload process will throttle itself at about 30% of one core's cpu usage and will not upload faster than 80 MB/sec no matter the actual link speed. Additionally, if I start a second simultaneous snapshot upload, the new process will also throttle itself at the same values (which means I reach a total speed of 160 MB/sec). The same happens with additional snapshots, until the network link speed is reached.

This is definitely caused by either inefficient code or a throttling mechanism inside glance itself. If this is a throttling mechanism, is there a way to modify it so that the upload speed and resource usage may increase per glance process?

I also understand that my question will become moot once copy-on-write cloning is available for snapshots, but that's not even close to be ready yet.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2015-08-26 07:58:55 -0500

dbayle gravatar image

Any chance to get an answer on this ?

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2015-07-06 08:34:45 -0500

Seen: 383 times

Last updated: Jul 06 '15