My FAS2650 2-node cluster is running out of gas. I'm exceeding the amount of data that this number of spindles can deliver. I need to micromanage performance to keep us up and running while we wait for additional resources to arrive. The filer is running 9.3P2.
I'm noticing that when I back up volumes with NDMP (using NetVault Backup), the backup traffic can consume a significant portion of the available performance capacity. I'm looking for a way to throttle NDMP so it doesn't impact the overall performance of the filer.
I've looked at storage QoS, but I can't find a way to manage requests by protocol. In some cases, I want a volume to give high priority to NFS requests while giving low priority to NDMP requests to that same volume. I haven't found a way to do that.
Any suggestions are appreciated.