[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [PATCH] Disable fsync at transaction commit

From: Benjamin Pflugmann <benjamin-svn-dev_at_pflugmann.de>
Date: 2003-02-24 14:09:18 CET

On Mon 2003-02-24 at 04:23:15 +0100, Branko ??ibej wrote:
> Benjamin Pflugmann wrote:
[...]
> >Very well possible. Dependend on implementation, RAID5 is slower than a
> >plain disk at writes, but never better.
              ^^^^^^^^^

> Uh, no, it will be better on reads, because it can parallelize them
> (just like RAID0) -- but that doesn't help the fsync, of course.

Yeah, what I said... I already limited my argument to writes.

> Nope, it's a H/W IDE raid controller (the 3ware Escalade, if you're
> interested about such things).

Ummm. That leaves the possibility that the raid controller disables
write caching for the IDE disks, which would be a sensible thing to
do.

[...]
> >which is more in the range of your results.
> >
> Nope, mine was 56 to 34 minutes. :-)

:-)

    Benjamin.

  • application/pgp-signature attachment: stored
Received on Mon Feb 24 14:10:10 2003

This is an archived mail posted to the Subversion Dev mailing list.