|From:||Theodore Tso <tytso-AT-mit.edu>|
|To:||Linus Torvalds <torvalds-AT-linux-foundation.org>|
|Subject:||Re: [PATCH 0/8][RFC] IO latency/throughput fixes|
|Date:||Mon, 6 Apr 2009 14:31:57 -0400|
|Cc:||Jens Axboe <jens.axboe-AT-oracle.com>, Linux Kernel Mailing List <linux-kernel-AT-vger.kernel.org>|
On Mon, Apr 06, 2009 at 08:45:09AM -0700, Linus Torvalds wrote: > > With that request size limiting, my fsync pauses tend to be in the 250ms > range, with a one 0.75s outlier in the last few minutes. I can definitely > feel that quarter-second thing, and the .75s pause was a real stutter, but > boy what a difference. I don't get the uncontrollable urge to kill that > nasty background writer any more. Cool! Maybe we can make things even better, but given where we are at, this brings up an question: given Jens block latency patches (thanks, Jens!), and the patches which add the flush on replace-via-rename/replace-via-truncate for ext3 data=writeback, should we make ext3 data=writeback the default for 2.6.30? - Ted
Copyright © 2009, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds