Steve I sure hope that is not the problem. I had gone down that line of thought but I keep coming back to the snapmirror transfer rates. I don't know the details of a snapmirror transfer but I would expect to see issues there also if latency was the problem.
Mike Miller General Dynamics Information Technology Michael.Miller.ctr@ustranscom.mil Michael.Miller@gdit.com Phone: 618-229-1185
-----Original Message----- From: Stephen C. Losen [mailto:scl@sasha.acc.Virginia.EDU] Sent: Wednesday, January 31, 2007 1:43 PM To: Miller, Michael CTR USTRANSCOM J2 Subject: Re: NFS speed
Perhaps your problem is network latency rather than bandwidth. Perhaps you can send a large burst of data to the netapp (high bandwidth) but it takes a while for that burst to travel all the way across the network (high latency).
In this situation a process on the NFS client sends a request and has to wait for the request to arrive at the netapp and then wait for the response to before sending the next request. During these wait periods, more data could be sent or received, so the network just sits idle for most of the time.
If the load on the netapp is really low (which is probably the case) then I suggest testing this theory by firing up a bunch of parallel processes on the NFS client that each read and/or write a large file on the netapp. If the throughput goes up, then that probably indicates latency.
Steve Losen scl@virginia.edu phone: 434-924-0640
University of Virginia ITC Unix Support