Back again with my NFS testings.
I give a try to kernel 2.2.9 today to see how much improvement it brings.
The results against a 2.2.6-ac1 (or 2.2.9) knfsd 1.3 server are quite
bad (using iozone) :
> /mnt/tmp > ~/iozone/iozone-2.2.9 200 8192 zz
>
> IOZONE: Performance Test of Sequential File I/O -- V2.01 (10/21/94)
> By Bill Norcott
>
> Operating System: POSIX 1003.1-1990
>
> Send comments to: b_norcott@xway.com
>
> IOZONE writes a 200 Megabyte sequential file consisting of
> 25600 records which are each 8192 bytes in length.
> It then reads the file. It prints the bytes-per-second
> rate at which the computer can read and write files.
>
>
> Writing the 200 Megabyte file, 'zz'...33.510000 seconds
> Reading the file...203.620000 seconds
>
> IOZONE performance measurements:
> 6258287 bytes/second for writing the file
> 1029934 bytes/second for reading the file (2.2.9 client)
4298323 bytes/second for reading the file (2.2.6 client)
> /mnt/tmp >
With 2.2.9 client, the load of the server climbs up to 3 during reading
(we have the same with many 2.2.6[-ac?] readers). It used to have a much more
reasonnable load with one 2.2.6 client (around 0,5) during reading.
One more thing : server 2.2.9 seems to give a boost to FreeBSD clients :
> /mnt/tmp > ~/iozone/iozone.freebsd 200 8192 zz
>
> IOZONE: Performance Test of Sequential File I/O -- V2.01 (10/21/94)
> By Bill Norcott
>
> Operating System: POSIX 1003.1-1990
>
> Send comments to: b_norcott@xway.com
>
> IOZONE writes a 200 Megabyte sequential file consisting of
> 25600 records which are each 8192 bytes in length.
> It then reads the file. It prints the bytes-per-second
> rate at which the computer can read and write files.
>
>
> Writing the 200 Megabyte file, 'zz'...24.476562 seconds
> Reading the file...36.835938 seconds
>
> IOZONE performance measurements:
> 8568000 bytes/second for writing the file
> 5693222 bytes/second for reading the file
> /mnt/tmp >
I think it can be easily reproduced.
A+,
-- Thierry Danis Poste : 53 53 danis@spmo.sagem.fr- 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/