'How does Azure Autoscaling-rule aggregate samples?
My understanding is that the time grain statistic-setting controls how the metric is aggregated within the time grain sample duration, and the time aggregation-setting controls how the metric is aggregated within the timeWindow (Duration).
Following that, I can set a time grain statistic to Maximum, which will take the maximum value from the metric within the time grain (e.g. 1 minute).
If the timeWindow is set to 10 minutes, and time aggregation is set to Average, I expect the average of 10 maximum values of the metric to be used to determine whether the rule threshold has been crossed.
This is different from what the following quote from the learning path explains:
From the learning path 3/7:
"The aggregation calculation for the Duration can be different for that of the time grain. For example, if the time aggregation is Average and the statistic gathered is CPU Percentage across a one-minute time grain, each minute the average CPU percentage utilization across all instances for that minute will be calculated. If the time grain statistic is set to Maximum, and the Duration of the rule is set to 10 minutes, the maximum of the 10 average values for the CPU percentage utilization will be used to determine whether the rule threshold has been crossed."
I admit that I can be mistaken in my understanding, but in any case, the example - or this section - should be revised.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|