arpnetworking/mad repository overview

Metrics Aggregator Daemon (MAD) aggregates metric samples into configurable time buckets (e.g. 1 second, 1 minute, etc.) published by metrics client libraries (e.g. Java, NodeJS, Ruby, etc.) or published from supported data sources (e.g. Telegraf, Collectd, Graphite, etc.) to compute a variety of statistics (e.g. min, max, sum, count, average, percentiles, etc.). The statistics are reaggregatable and are published together with supporting data to configurable destinations.

To run MAD via Docker simply invoke:

docker run -p 7090:7090 arpnetworking/mad:${VERSION}

This will start a plain dependency free instance with most supported sources enabled. The only sink that is enabled is Metrics Portal which can stream per-second aggregates directly from MAD. However, other supported destinations will require additional configuration. We recommend using Cluster Aggregator to at least act as a connection/request aggregator to your data store although it can also function as a reaggregator for data stores which do not support complex statistics like histograms (e.g. InfluxDb). We recommend you either use this as your base image and layer custom configuration on top or use volume mounts to inject your custom configuration. More information about the project and its configuration is available on Github at:

https://github.com/ArpNetworking/metrics-aggregator-daemon⁠

MAD is also available as a JAR in Maven Central as well as a TGZ and RPM in Github Releases.