Re: NFS vs. md raid
"Christopher McCrory" <[email protected]>
| Newsgroups | gmane.linux.redhat.release.taroon.general |
|---|---|
| Message-ID | <[email protected]> |
On Thu, June 8, 2006 10:06 am, Stephen John Smoogen wrote: > On 6/8/06, Christopher McCrory <[email protected]> wrote: >> Hello... >> >> speaking of NFS... >> >> I've continually had slow write performance ( specifically file >> creation) with RHEL3 when the NFS server filesystem is a software raid >> device. THe same thing happens with RHEL4. When using a single drive >> as an export, performance is fast. So far the only option I have found >> that helps is exporting with the 'async' option. All clients are also >> RHEL > > > Well a lot of questions can affect this: > > 1) What kind of disk controller LSI scsi > 2) What kind of disk drives? U320 15k 300G same with U160 and other sizes > 3) What kind of RAID? raid 1 and raid 5 (maybe raid0 also) > 4) What kind of NFS? standard RHEL3 NFSv3 > >> >> example: >> time tar -xzf lots.of.files.tar.gz >> >> local drive: >> real 0m6.053s >> user 0m2.210s >> sys 0m1.180s >> >> nfs with async (insane) >> real 0m6.492s >> user 0m2.150s >> sys 0m1.140s >> >> nfs with sync (sane) >> real 1m52.756s >> user 0m2.430s >> sys 0m1.290s >> >> >> 6 seconds vs. 120 seconds. Ouch >> >> Any ideas? >> <snip> -- Christopher McCrory "The guy that keeps the servers running" [email protected] http://www.pricegrabber.com Let's face it, there's no Hollow Earth, no robots, and no 'mute rays.' And even if there were, waxed paper is no defense. I tried it. Only tinfoil works. -- Taroon-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/taroon-list