How Metric Collections Works
The Virtana Infrastructure Observability InfiniBox translator aggregates together live performance data (1 seconds data) and some point-in-time values (capacity etc) to populate all of the Virtana Infrastructure Observability metrics. The live performance data requires an InfiniBox collector be set up on the InfiniBox array. These InfiniBox collectors track approximately the last 30 seconds of data at 1 second granularity. The Virtana Infrastructure Observability InfiniBox translator therefore must connect to the array every 15 seconds (+/-) to pull the latest data before it is discarded. This live data is stored internally in the translator. Every 2 minutes the live data is batched into 30 second buckets and sent into Virtana Infrastructure Observability for processing. This results in data arriving every 2 minutes with a granularity of 30 seconds.
During collection if the Virtana Infrastructure Observability InfiniBox translator detects a problem with the InfiniBox collectors it will attempt to recreate the translators in order to maintain consistent data collection.
Notice
Operations per second/Bytes per second will not be collected if there are no associate bytes transferred for a given initiator/host. For example, if you have an NFS agent performing requests which result in no payload (0 bytes transferred over entire window) at all then there will be no data collected for that initiator/host. This is because InifiniBox does not return data for entries which have 0 throughput and IO collects statistics keyed by throughput.
The complete list of metrics for this integration is available in a separate guide.