Implementing Custom Aggregates in Gnocchi

Adding custom aggregates in Gnocchi is desirable. Aggregates such as exponentially weighted moving averages need rolling time windows and require retention of full time series. In Gnocchi, these aggregates can be configured as separate classes, loaded as extensions on run-time, and computed on the full data retrieved from a GET request. The parameters necessary to specify the aggregate, such as the time window length, are passed through the GET request as keywords, e.g.:

GET /v1/entity/UUDI/measures?aggregation=ewma&window=60s