Bruce,
Currently the copy is between vols.
These are mounted:
# mount -p | grep brs-data wesson:/vol/brs2/argus/brs-data2 - /export/argus/brs-data2 nfs - no smith:/vol/brs1/argus/brs-data1 - /export/argus/brs-data1 nfs - no #
And files like this one get copied from 1 to 2 or vice versa daily:
# ls -l inv0.db -rw-rw-rw- 1 siteadm netsite 439784576 Jan 8 06:50 inv0.db #
We've played with nfs a little bit on this box:
# uname -a ; nfsstat -m | head SunOS load1 5.6 Generic_105181-16 sun4u sparc SUNW,Ultra-4 /export/argus/brs-data2 from wesson:/vol/brs2/argus/brs-data2 Flags: vers=3,proto=tcp,sec=sys,hard,intr,link,symlink,rsize=32768,wsize=32768,retrans=5
/export/argus/brs-data1 from smith:/vol/brs1/argus/brs-data1 Flags: vers=3,proto=tcp,sec=sys,hard,intr,link,symlink,rsize=32768,wsize=32768,retrans=5
There may be simultaneous multiple copies happening across these two mount points; I don't have any good numbers yet for sustained cp rates.
The developers are asking for gigabit ethernet between the filers and the box that does the copies. Taking them out back and shooting 'em isn't currently an option, but we're looking into it for future releases.
Dave
Bruce Sterling Woodcock wrote:
I've been looking for a way around nfs bandwidth problems --one of the database update steps involves copying 1 to 2 G files from one vol mountpoint to another.
To another mountpoint in the same vol, or another vol? If the former, just mount from the root of the volume and do the mv from there.
Bruce