Rollup Aggregation in Charts
A chart can show only so many points. Over a long time range, a time series has far more samples than the chart has room for, so Kloudfuse plots one point per time interval, called the step, and aggregates all the samples in each interval into that one point. This aggregation is a rollup.
There are two types of rollups, implicit and explicit.
Implicit rollup
When a chart’s time range holds more data points than the chart can display, Kloudfuse rolls them up automatically. It picks the step from the time range (see Rollup Aggregation Step Size) and aggregates each step with avg.
Explicit rollup
To choose the interval or the aggregation yourself:
-
Set the step in the Metrics Explorer’s Options (Step (s)); see Chart options.
-
Add a function from the Rollup category in the query builder’s operations to pick the aggregation; see Operations: aggregations and functions.
Rollup aggregation methods
Kloudfuse supports the following rollup aggregation methods: avg, sum, min, max, and count.
Chart rollups are separate from the stored 5-minute and 30-minute rollups described in How Metrics Rollup Works: the stored rollups make long queries cheaper to run, and a chart rollup decides how the result is plotted.
|
Additional Resources
|