Aggregations
Statful allows you to send metrics from any period in time.
However, as we’ve seen before, going for extended time frames might lead to an automatic adjustment of the granularity displayed on your monitoring graph.
In this context, we can say data points are aggregated, an operation that is performed via averages, and rounded up to one-second resolution.
N.B.: Aggregations are optional, so you can choose to send and store raw data points instead.
Aggregation Functions
Statful also enables you to aggregate metric data to produce statistical summaries at different data granularities.
When specifying an aggregation, Statful takes the metrics it receives and aggregates them based on the mathematical function and time interval of your choice.
It proves itself useful, e.g., when you have to collect a lot of data over a minute.
The following aggregation functions are supported:
- avg - computes the average of the ingested numeric values.
- count - counts the number of ingested values.
- sum - sums up the ingested numeric values.
- first - returns the first received value.
- last - returns the last received value.
- min - returns the minimum of the ingested numeric values.
- max - returns the maximum of the ingested numeric values.
Percentiles are particularly useful to understand the distribution of your metric data, especially when large variances take place.
For example, you can decide to examine data that scored below the Xth percentile of the observed values, where x is the percentile value that you want to apply:
- p90 - establishes a threshold of acceptance where 90% of the observed values occur.
- p95 - establishes a threshold of acceptance where 95% of the observed values occur.
- p99 - establishes a threshold of acceptance where 99% of the observed values occur.
Aggregation Frequencies
Considering that we’re talking about time series, it makes sense to set an Aggregation Frequency that defines the time interval considered on aggregation functions.
Statful uses this interval to buffer the metrics before calculating and storing the final result of the aggregation.
The following aggregation frequencies are supported:
- 10 seconds
- 30 seconds
- 60 seconds
- 120 seconds
- 180 seconds
- 300 seconds