Messages in this thread | | | Date | Fri, 14 Jan 2000 09:21:14 +0100 | From | Steffen Grunewald <> | Subject | Re: NFS question |
| |
On Thu 2000-01-13 (08:38), Kay Nettle wrote: > We have around 100 Linux boxes that get home directory service from a > Sun ultra enterprise 4000. We've started doing RAID5 on the Sun and > have noticed a x4 increase in NFS writes. The problem seems to be > specific to RAID5, we have a couple of RAID0+1 disks and we do see a > small performance hit, it is nothing like RAID5. > > Are there any NFS patches that might help with this problem, or are > we just screwed?
I'd guess a hardware NFS accelerator would be the best solution, since NFS waits for the actual blocks being written to disk physically and does the ACK then. RAID5 would require matching block sizes, too.
There have been PrestoServe solutions some years ago. I'm still using one in an old SS10 machine, works great. Improves data integrity, too (stores meta data and saves them after a crash or power outage)
Ask SunOnline whether they still exist.
Steffen -- Steffen Grunewald = steffen(at)gfz-potsdam.de = fax +49-331-288-1266 + steffen.grunewald(at)gmx.de, (without subject) 9311356(at)scall.de My opinions are mine. You can rent them. Just ask me for prices :-) "Art is a lie which makes us realize the truth" - Pablo Picasso
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |