From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hans Reiser Subject: Re: [PATCH] various allocator optimizations Date: Wed, 12 Mar 2003 14:12:53 +0300 Message-ID: <3E6F1635.2010305@namesys.com> References: <1047400482.8215.312.camel@tiny.suse.com> <20030311194205.A4493@namesys.com> <1047403968.8219.337.camel@tiny.suse.com> <3E6E584D.4080809@namesys.com> <1047421551.8219.448.camel@tiny.suse.com> <3E6E674E.4040305@namesys.com> <1047433739.8215.487.camel@tiny.suse.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: list-help: list-unsubscribe: list-post: Errors-To: flx@namesys.com In-Reply-To: <1047433739.8215.487.camel@tiny.suse.com> List-Id: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: Chris Mason Cc: Oleg Drokin , reiserfs-list@namesys.com Chris Mason wrote: >On Tue, 2003-03-11 at 17:46, Hans Reiser wrote: > > >>Chris Mason wrote: >> >> >> >>>On Tue, 2003-03-11 at 16:42, Hans Reiser wrote: >>> >>> >>> >>> >>> >>>>Chris, don't you think the right answer would be to take zam's resizer >>>>and make a defragmenter out of it? >>>> >>>> >>>> >>>> >>>Yes and no, for a defrag program to fix things we'd have to agree on an >>>optimal layout ;-) Also it assumes the machine has idle time when a >>>defragment cycle is possible. >>> >>> >>> >>No, it assumes that 80% of files don't move during the course of a week, >>so if defrag takes a week, it still adds value. >> >> >> >I ran a number of systems where this wasn't true. Spool files are >constantly appended to or deleted by pop/imap servers. Idle time is >important, between normal operations and backups it is almost impossible >to find a time when big servers have disk bandwidth to spare. > >I like the idea of a dynamic in-kernel fragmentation tool though, you >mark a file as being in need of reallocation, and it happens before io >or something (hand waving is fun). > You mean you like the allocate on flush we do in reiser4? Perhaps I should say only that for 80% of machines 80% of the files never move during the course of a week.;-) allocate on flush will improve a lot of the cases, and an online repacker will improve a lot of the ones that allocate on flush does not improve. -- Hans