How Metric Collections Works

Metrics are collected via the PowerStore REST API each polling interval. The polling interval can be set to five minutes for small environments or up to one hour for larger environments. Network latency and/or bandwidth could affect metrics collection: a high latency/low bandwidth connection between IO and PowerStore will result in slower data collection, requiring a longer duration polling interval. The granularity of the data-- the number of data points per poll interval--is set by PowerStore: usually twenty seconds for space-related metrics and five minutes for performance-related metrics.

Unlike discovery, the PowerStore API only supports non-bulk metric data collection, i.e., one entity at a time. Thus, for large numbers of entities (e.g., volumes) it may take a significant amount of time to collect data. The polling interval should be set accordingly.

Metric data collection is elastic. For the first collection or for preparing for the most recent querying time windows, data for up to ten minutes in the past will be queried and the timestamp will be cached. For the next interval and starting from the cached timestamp, data up to the current timestamp will be collected. In case of downtime or a system restart, up to sixty minutes of data will be queried. If the latest stored timestamp is greater than sixty minutes in the past, only data for the most recent querying time window will be collected.

The complete list of metrics for this integration is available in a separate guide.