When gathering Windows host performance data to model a solution in a business continuity/storage design tool, the architect must first scope the exercise by using Disk Administrator to identify the specific LUNs that require replication, since this determines which disks need to be instrumented. Perfmon is then set up to collect disk statistics for those identified LUNs. Next, perfmon itself is configured for the data-collection run — this includes setting up the log file and resetting the counters so that the capture begins from a clean state rather than including stale accumulated values. The collection interval and schedule are then configured to control the sampling frequency and start/stop timing of the log. Finally, the data collection is executed over a three- to five-day period that spans peak business activity, since a shorter window risks missing the peak I/O load that the sizing model needs to capture.
Community Discussion