Composite metrics allow you to create a metric that is a functional composition of other metrics, thereby allowing you to apply a formula to a group of metrics.
For example, you could use a Composite metric to find the sum of transaction rates across 100 web servers, the standard deviation of server access time, the average of a specific metric over a specified time window, or the average concurrency across a large cluster of databases.
Within Circonus, each composite metric is its own check and has exactly one metric. The name of that metric is chosen by the creator of the check.
Each composite metric has a formula that defines how the value is calculated. Here is a simple example to calculate the average duration of requests for users:
(metric:average(1,"duration",60000) + metric:average(2,"duration",60000)) / metric:average(10,"users",60000)