-------Sequential Output-------- ---Sequential Input--
--Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block---
--Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec
%CPU
Local 100 5437 83.0 10734 15.4 3268 9.6 6166 86.1 7364 8.8 91.7
2.3
NFSv3-1 100 3023 52.2 3406 8.7 1275 3.7 2267 32.2 2517 2.2 200.7
4.1
NFSv3-2 100 3138 58.2 3699 11.8 1602 5.6 3010 42.9 2677 2.7 214.3
4.2
NFSv2-1 100 454 6.3 572 1.6 706 2.9 4986 74.1 8970 13.9 226.5
5.8
Local Ext2fs Local Disk 450Mhz PII
NFSv3-1 Version 3 NFS, rsize=32768, wsize=32768
NFSv3-2 Version 3 NFS, rsize=32768, wsize=8192
NFSv2-1 Version 2 NFS, rsize=8192, wsize=8192
You'll notice that NFSv2 outperforms v3 with reads. I had sent
Trond a tcpdump of of the V3 vs V2 reads and he thought he knew
what the problem was, I just don't know if he was planning on
trying to fix it. It seems to be a readahead mismatch between
linux's NFS and Sun's disks.
Andrew
> What happens when u specify a higher wsize? can you get the best of both
> worlds? Just curious?
> jpd
>
> At 06:40 PM 4/20/99 -0700, Thomas Davis wrote:
> >Thomas Davis wrote:
> >>
> >> Trond Myklebust wrote:
> >> >
> >> > Doesn't the solaris server normally select a 32k rsize/wsize if you
> >> > don't force it to 8k? If it does, that should give better performance.
> >> >
> >>
> >> I have no ideas; the values are forced in via a autofs mount table
> >> (there are about another 40 linux machines that need the larger values).
> >>
> >> I can do a seperate mount and see what happens..
> >>
> >
> >Ok, here's the new results:
> >
> >mount command used:
> >
> >mount starsu00:/data01 /test -ohard,intr
> >
> >[tdavis@pdsflx16 bonnie]$ ./bonnie -d /test/b -s 1024
> >File '/test/b/Bonnie.11787', size: 1073741824
> >Writing with putc()...done
> >Rewriting...done
> >Writing intelligently...done
> >Reading with getc()...done
> >Reading intelligently...done
> >Seeker 3...Seeker 1...Seeker 2...start 'em...done...done...done...
> > -------Sequential Output-------- ---Sequential Input--
> >--Random--
> > -Per Char- --Block--- -Rewrite-- -Per Char- --Block---
> >--Seeks---
> >Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU
> >/sec %CPU
> >32768 1024 5180 75.6 5814 12.1 2770 7.3 6144 92.0 10308 7.9
> >290.3 4.7
> >8192 1024 5298 81.1 7138 17.2 4270 17.8 4826 70.1 9980 13.3
> >389.1 6.2
> >
> >from /proc/mounts (mount command doesn't give you the rsize/wsize)
> >
> >starsu00:/data01 /test nfs
> >rw,v3,rsize=32768,wsize=32768,intr,addr=starsu00 0 0
> >
> >so, 32k wsize slows you down some. 32k rsize speeds it up alot though..
> >
> >--
> >------------------------+--------------------------------------------------
> >Thomas Davis | PDSF Project Leader
> >tadavis@lbl.gov |
> >(510) 486-4524 | "Only a petabyte of data this year?"
> >
> >-
> >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/
> >
>
> -
> 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/
>
Andrew Schretter
Systems Programmer, Duke University
Dept. of Mathematics (919) 660-2866
-
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/