If we had a "good" cache tiering implementation, rbd and cephfs clients could read and write via a fast, smaller replicated pool, with writeback etc. to the backing EC pool. Our existing cache tiering implementation didn't work well and has been deprecated. (Do we have a summary of the learnings there and plans for tiering v2 ?) IIRC, the main issue was that misses are particularly expensive: a cache miss partial read or write required the promotion of the entire object from the underlying EC pool before the small IO could proceed. A better approach could be to read around misses, using the new faster EC partial read to respond to the client quickly, and then asynchronously promote the whole object into the cache pool. Partial writes are more tricky, probably best handled all via a WAL. I also wanted to bring up this related work: https://github.com/ceph/ceph/pull/57631 That PR is a response to: * real time compression of small writes is too slow. (a small write turns into a larger read, modify, write..) * small writes fragment larger objects over time, hurting performance eventually. So the common theme is: when we are synchronous with a client, do whatever to be as fast as possible, and then optimize things in the background. (in this case -- recompress/defrag during deep scrub). -- Dan van der Ster CTO Clyso GmbH p: +49 89 215252722 | a: Vancouver, Canada w: https://clyso.com | e: dan.vanderster@xxxxxxxxx We are hiring: https://www.clyso.com/jobs/ On Mon, Jul 1, 2024 at 1:09â?¯AM Bill Scales <bill_scales@xxxxxxxxxx> wrote: > > Hi, > > We are planning to enhance the performance of erasure coding, in particular for use with block and file. We've got a design document https://github.com/bill-scales/CephErasureCodingDesign that sets out what we are hoping to achieve. We welcome your feedback, either posting your comments in Slack on #ceph-devel , raising issues in github or getting in contact with myself > > > > Cheers, > > Bill. > bill_scales@xxxxxxxxxx > IBM Distinguished Engineer, IBM Storage > > > > Unless otherwise stated above: > > IBM United Kingdom Limited > Registered in England and Wales with number 741598 > Registered office: PO Box 41, North Harbour, Portsmouth, Hants. PO6 3AU > _______________________________________________ > Dev mailing list -- dev@xxxxxxx > To unsubscribe send an email to dev-leave@xxxxxxx _______________________________________________ Dev mailing list -- dev@xxxxxxx To unsubscribe send an email to dev-leave@xxxxxxx
