Do a sysstat while copying the data and watch the CPU usage. We've found that when copying large amounts of data within a volume (or between volumes for that matter) that the filer becomes CPU bound and just can't move it any faster.
J-
--------------------------------------------------- "Allow me to extol the virtues of the Net Fairy, and of the fantastic dorks that make the nice packets go from here to there. Amen." TB (www.penny-arcade.com) ---------------------------------------------------
|--------+----------------------------> | | Dale Gass | | | <dale@novamediasw.| | | com> | | | Sent by: | | | owner-toasters@mat| | | hworks.com | | | | | | | | | 01/21/2002 12:53 | | | PM | | | | |--------+---------------------------->
---------------------------------------------------------------------------------------------------------------------------------------------|
| | | To: Rick Gardner rgardner@westernelectronics.com | | cc: "'toasters@mathworks.com'" toasters@mathworks.com | | Subject: Re: Performance Issues |
---------------------------------------------------------------------------------------------------------------------------------------------|
Rick Gardner wrote:
Anyone seen this problem?
I have performance problems when I copy large amounts of data between directories within the same volume on the NetApp. It is about 10 times faster to copy it back to my Sun box / tmp and then back to the filer.. I have a Cluster 820 with Gig NICs and the Sun box is a SunFire 280R, Solaris 8, with Gig NICs. No options with the mount commands.
I have similar experiences with netapps. I often found it faster to copy over the network to a second machine (Linux, in my case), then copy it directly between two places in the same volume.
Seems rather strange and disappointing...
-dale
No CRC errors on the switch, NetApps, or Suns...
Any ideas?
Thanks!