The magic bullet may be a filer option. I found at one point when the filers were used and an ATM cloud was in the mix, toggling the option
Ip.fastpath.enable
Usually (not always though) did the trick.
The improvement was immediate, if I recall.
Toggle and let me know.
--tmac
Sent from my Verizon Wireless BlackBerry
-----Original Message----- From: "Miller, Michael CTR USTRANSCOM J2" michael.miller.ctr@ustranscom.mil Date: Wed, 31 Jan 2007 12:29:20 To:toasters@mathworks.com Subject: NFS speed
I'm trying to increase the NFS ops of an R200 that sits at the end of a WAN link and I can't find the magic bullet if there is one.
Some background for you: The destination end is an 18MB section of an OC3 ATM pipe with little extra traffic. The source end is a 100MB pipe into an ATM cloud. The filer is a snapmirror destination and using sysstat I see Net in readings of 1300+ kB/s while a snap operation is in progress. Snapmirror reports show one transfer of 4.20GB taking 2:28 hours and another of 1.26GB taking 53 minutes.
My problem comes into play with NFS operations. I use the filer as a disk backup destination using Legato Networker and only get less than 80KB/s transfers. This shows up in sysstat as 6-8 NFS ops/s with less than 75 kB/s showing for input traffic.
I've tried various changes to the mount options and played with settings on the Solaris host and the filer. Had no joy anywhere.
Anyone have any thoughts on how to speed this up? I have thought about pulling the backup target in-house and using the Legato clone mechanism to get the data back to the filer but I doubt that would help much.
Thanks Mike Miller General Dynamics Information Technology Michael.Miller.ctr@ustranscom.mil Michael.Miller@gdit.com Phone: 618-229-1185