Turn cluster performance history into an explicit PowerShell comparison

Source facts

Microsoft shows performance-history output being piped from Get-ClusterPerf into Measure-Object to calculate statistics from the Value property. Its CPU example selects a named series and timeframe, then reports minimum, maximum, average, and time above chosen utilization thresholds. A separate example identifies drives whose hourly average latency exceeds the population average by more than three standard deviations. That is the example’s statistical selection rule. Microsoft documentation.

Applicability

Define the operational question, measured object set, metric, time window, and aggregation before using a sample. Review whether the selected interval contains the activity being investigated.

DSE recommendation

Write the script’s comparison rule in plain language beside its parameters. Keep statistical outlier selection separate from a diagnosis of failed hardware. Have another administrator review units, object membership, and time selection, and preserve the raw output before formatting a report.

Verification

Run the script against a small, understood object set and independently check one calculated value. Compare the results with the corresponding activity timeline. Record missing samples and excluded objects, and investigate an outlier using additional workload evidence before recommending a replacement or configuration change.

Official references

Microsoft Learn: Scripting with Storage Spaces Direct performance history. Source reviewed September 8, 2026.