Thanks Ben, yes coalescing is true if you are *not* running with direct io, are you ? We could easily figure out client side coalescing by analyzing network traffic I suppose. << I'm looking forward to hearing other people's results with Cephfs as well Within SanDisk, we started benchmarking CephFS on all flash and at some point we should be able to share result with community. Regards Somnath -----Original Message----- From: Ben England [mailto:bengland@xxxxxxxxxx] Sent: Wednesday, August 03, 2016 12:34 PM To: Somnath Roy Cc: ceph-devel Subject: Re: CephFs performance data Thanks Somnath, here are the Cephfs slides from this morning. https://s3.amazonaws.com/ben.england/Cephfs+performance+results+June+2016.pdf Let me know if you have trouble reading them. As for sequential 4-KB I/O performance, I believe kernel Cephfs client is aggregating reads and writes - otherwise you are quite right that the IOPS level could not be sustained. This is common practice in other distributed filesystems as well, such as NFS and Gluster. For RANDOM I/O, coalescing of I/O requests is far less likely. I have some performance data on these runs that should be useful in confirming the result, will try to look that up for you. I'm looking forward to hearing other people's results with Cephfs as well. -ben ----- Original Message ----- > From: "Somnath Roy" <Somnath.Roy@xxxxxxxxxxx> > To: "Ben England (bengland@xxxxxxxxxx)" <bengland@xxxxxxxxxx> > Cc: "ceph-devel" <ceph-devel@xxxxxxxxxxxxxxx> > Sent: Wednesday, August 3, 2016 12:23:21 PM > Subject: CephFs performance data > > Ben, > Thanks for the first cut data on CephFS , it was really helpful. > Do you mind to send out the slides ? > Also, continuing the discussion on 4K SW throughput is ~1GB/S , if you > are running test with direct IO, client Linux system can't coalesce > the writes and I didn't see much coalescing at the OSD servers as > well. Unless CephFS has some coalescing logic it will be good if we > can find out how we are getting this bigger throughput for 4K. Let me > know if I am missing something.. > > Thanks & Regards > Somnath PLEASE NOTE: The information contained in this electronic mail message is intended only for the use of the designated recipient(s) named above. If the reader of this message is not the intended recipient, you are hereby notified that you have received this message in error and that any review, dissemination, distribution, or copying of this message is strictly prohibited. If you have received this communication in error, please notify the sender by telephone or e-mail (as shown above) immediately and destroy any and all copies of this message in your possession (whether hard copies or electronically stored copies). ��.n��������+%������w��{.n����z��u���ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
