I am curious if anyone else is taking a close look at their Logical Disk Performance/Disk Bottlenecks in their environment. One metric we track as a vital statistic on windows is Logical Disk Transfers/sec, with the general idea that consistent high water marks are indications of SAN layout or design problems. We ran into a problem today where we were asked to troubleshoot a database issue, Logical Transfers were off the charts but spikey. Then we got to thinking, Orion is polling the data, and the data is already spikey in perform. So if Orion happens to poll it when its high it will report high, if it polls it when its low then it will be low. In other words its an unreliable metric due to the nature of polling.
Now everything I have read says to use this perfmon value, but there is another perfmon value called Average Disk Transfers/sec which appears to be a running *duh* average of the other metric. But in several websites I cant find anyone saying you should use the average.
Is anyone proactively monitoring disk performance? What metrics are you using?
