Contributors netapp-manishc

From the System Fitness Dashboard, you can click the Performance icon image to view the performance history of your system. These charts provide up to 60 days of historical performance data, which is useful for performance trend and pattern analysis. The hourly averages used to prepare these charts are reported in a daily performance AutoSupport data summary.

System interruptions, such as reboots and service disablements, can cause gaps in the chart. These performance charts are intended for trending analysis, and they should not be used for detailed performance monitoring or diagnostics. You should use onsite products such as OnCommand suite of products for such use cases.

There are several viewable performance charts including Peak Performance (Headroom), CPU and Disk Utilization, IOPS, Latency and Throughput. Users can check one or more of these charts for selective viewing of performance charts. Charts are downloadable in PDF, SVG, and PNG formats. You can also export all the counter information into a CSV from the menu.

Peak performance zone is the area which is equal to or below the peak performance line. In simple terms, it specifies the limit of good operating behavior for the given storage resource. When a resource’s utilization rises above this line, the client latencies increases rapidly.

Headroom is the difference between peak performance line and current utilization line. Monitor the performance graphs periodically to identify the nodes that may run out of headroom. If the current resources utilization is above this peak performance line for an extended time, a performance remediation plan might be appropriate. A performance remediation plan might include setting QoS workload limits, moving volumes or LUNs to another storage controller, or expanding the storage cluster.

The confidence factor is used to determine the accuracy of the peak performance line that is used in CPU and aggregate headroom graphs. The confidence factor counter indicates how good the range of utilizations and latencies were observed for a resource in the system. The higher the confidence factor, the more accurate the peak performance line will be. Confidence factors range from 1 (low) to 3 (high).

There are cluster aggregated performance charts in cluster performance dashboard and can view node level graphs.

Response Time by Protocol and Concurrency graphs are not available for cDOT systems.