> > Hi, all. I've just tried measuring Linux NFS client performance on a
> > dual PPro running 2.3.4. Block size is 8 kiB, network is 100 Mb/s
> > switched full duplex. The server is running Solaris 2.5.1 on a quad
> > UltraSparc (250 MHz).
> >
> > Why is NFS write performance so bad? I apologise for not having paid
> > closer attention to this, if this is a well-worn topic.
>
> 2.3.x doesn't have the -ac NFS improvements in (for one Linus I think wants
> them a bit cleaner than they are now). That kills you. Really we need to get
> NFSv3 in the main code for 2.3.x too
The latter is the real solution to the problem of Solaris
servers. They seem very touchy about the block sizes used, and as long
as we stick with NFSv2, we're limited to 8k. The Solaris server
performance seems, in general, to drop significantly for block sizes
other than 32k.
Cheers,
Trond
-
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/