Originally posted by: stephen2615
We have Exchange 2003 running in a HSC configuration and every once in a while, the write performance goes downhill and this seems to compound problems with Exchange. Any apparent glitch with writing with Sync True Copy takes what seems forever to rectify. People then start getting upset and its always the storage's fault.
I monitor the TC performance and it seems that update response and update transfer are closely aligned. A spike with the response rate more often than not comes at the same time as the amount of data increases but our infrastructure should be able to cope with almost 8 times the performance. We have huge pipes between the sites for TC and they don't have any issues with traffic throughput. Both the USP's don't work all that hard for the parts that are used by Exchange.
Its a mystery why we get these spikes. I have a sneaking suspicion that Exhange is the problem as it does its writes at some randon times. It seems to wait for a quite time (if thats possible) and then blasts out hundreds or thousands of small I/O's which are the Microsoft standard of 4 kb's. The queue depths on the host goes ballistic and everyone blames the storage.
Am I the only person on this planet that has grief with Exchange 2003 performance in a sync True Copy situation? I have read all the suggested HDS material on Exchange so please don't offer that as a start.