The effect of granularity and polling period on aggregation
The effective granularity for an aggregation is the granularity period that is used in the aggregation. In some cases, the granularity that is specified for an aggregation is not the same as the effective granularity.
The effective granularity of an aggregation depends upon the relationship
between the specified granularity and the polling period, as follows:
- Granularity = polling period
- The effective granularity is the same as the polling period, for
example:
- If the polling period is 1 minute and the specified granularity is 1 minute, the effective granularity is also 1 minute.
- If the polling period is 5 minutes and the specified granularity is 5 minutes, the effective granularity is also 5 minutes. In both cases, if all resources have metrics for each polling period, one data point is collected for each resource during the effective granularity period.
- Granularity < polling period
- The effective granularity is the same as the polling period. For example, given a specified granularity of 1 minute and a polling period of 1 minutes, the effective granularity is 1 minutes. If all resources have metrics for each polling period, a data point is produced every fifth aggregation (every 5 minutes) because no data has been sampled for four of the 5 minutes. If the group has 10 resources, the aggregation is based on a sample of 10 metrics during the 5 minute effective granularity period.
- Granularity > polling period
- The effective granularity is the same as the specified granularity. For example, if the polling period is 5 minutes and the specified granularity is 10 minutes, the effective granularity is 10 minutes. If all resources have metrics for each polling period, the number of data points included in the aggregation are equal to twice the number of polled resources. If the group has three resources, the aggregation is based on a sample of six metrics (two for each resource) during the effective aggregation period.