From mboxrd@z Thu Jan 1 00:00:00 1970 From: Suresh Jayaraman Subject: Re: [PATCH 00/09] cifs: local caching support using FS-Cache Date: Fri, 23 Jul 2010 21:46:31 +0530 Message-ID: <4C49C05F.8000003@suse.de> References: <4C480F51.8070204@suse.de> <1278333663-30464-1-git-send-email-sjayaraman@suse.de> <4C3DF6BF.3070001@gmail.com> <4C3F35F7.8060408@suse.de> <1892.1279820400@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: Andreas Dilger , David Howells , Steve French , Scott Lovenberg , linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-cachefs-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org To: Stef Bon Return-path: In-Reply-To: Sender: linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: On 07/23/2010 02:05 PM, Stef Bon wrote: > In my opinion there should be article published about this, describing > fs-cache generally, and these kinds of benchmarks! FS-Cache is nicely documented on Documentation/filesystems/caching/fscache.txt present in the kernel source. IIRC, LWN - http://lwn.net/ had published a couple of articles about FS-Cache. I agree that there is little information available about the use-cases and benchmarks. What do you mean exactly by article? More blog entries or article in magazines or whitepaper sort? Which one do you think has a good chance of reaching the target audience? Thanks, > Using fs-cache for network filesystem is an important issue, and > should get "exposure". > > Stef Bon > > 2010/7/23 Andreas Dilger : >> On 2010-07-22, at 11:40, David Howells wrote: >>> Suresh Jayaraman wrote: >>>> As it can been seen, the performance while reading when data is cache >>>> hot (disk) is not great as the network link is a Gigabit ethernet (with >>>> server having working set in memory) which is mostly expected. >>> >>> That's what I see with NFS and AFS too. >>> >>>> (I could not get access to a slower network (say 100 Mb/s) where the real >>>> performance boost could be evident). >> >> More interesting than a slow network is testing with more clients. 10 clients should be able to get 10x the read performance from the client-local cache, probably more than the server's peak disk/network bandwidth. >> -- Suresh Jayaraman