Re: really slow RMAN backups

Okay...

So, your NFS parameters look good. You're using 32KB (maximum) rsize and wsize. And NFS v3 over TCP/IP.

Aside from using 1500 Byte MTU, you're doing fine. (Increasing MTU to the max -- 9000Bytes -- will reduce the number of packets sent and received, the amount of packet assembly/disassembly, and thus the overhead on your CPU. But this may not be an option if you're not using a dedicated network.)

So legato is not using the Netapp. You're still pretty close to my worst-case scenario for network crossings and NetApp I/O. Except 2 networks involved. (We hope. Are you sure both NICs don't plug in to the same switch?)

Finding your bottleneck *could* be a challenge.

Recognise that the bottleneck *could* be the lack of Asynch I/O...

What does 'sar' tell you about performance on your server while backups are running? 'dd' and 'ping' tests might help you assess the speed of your networks. (Try these while backups are running, and while they are not.)

Now, here's a (possible) biggie: Are you sure you have exclusive access to the Legato server? What else is being backed up at the same time?

--

Cheers,
-- Mark Brinsmead

   Staff DBA,
   The Pythian Group
   http://www.pythian.com/blogs

--
http://www.freelists.org/webpage/oracle-l