We're trying to improve the performance of our NFS server. It's a Sun
X4100M2 running b63 (upgrade from b62) connected via a 4GB Fibre
Channel QLogic HBA to a StorageTek 6140 array (2GB Fibre Channel).
1. rhel# mount -o noatime,vers=3,proto=tcp \
file-server:/opt/fibre-disk /mnt
rhel# time tar cf - gcc343 | (cd /mnt; tar xpf - )
...
419721216 bytes in 56:87 => 7380362.51 bytes/sec
2. rhel# time tar cf - gcc343 | ssh -oForwardX11=no file-server \
'cd /opt/fibre-disk; tar xpf -'
...
419721216 bytes in 35:89 => 11694656.34 bytes/sec
/opt/fibre-disk is one drive on the 6140 array configured as a RAID-0
array with a 128k stripe.
We have the following in /etc/system on the file server:
set maxphys = 0x800000
set ssd:ssd_max_throttle = 64
set zfs:zfs_nocacheflush = 1
set nfssrv:rfs_write_async = 262144
Why is tar/SSH performing so much better than tar/NFS? There's a
100MB/s network between the two systems.
--
albert chin (china at thewrittenword.com)
