We’ve just released our Prometheus integration. You can now send metrics from Prometheus to Anomify with a simple Prometheus configuration rule. Check out the support docs to see how to integrate Anomify into your workflow.

Why are we supporting Prometheus?

Prometheus is the most widely used time-series database. It has alerting rules but they take time to organise and tweak, and they only work against metrics you know to set them up against. The rest of your metrics go unobserved.

Anomify analyses your unobserved Prometheus metrics by

• accounting for seasonal changes.
• detecting anomalies.
• correlating across events.

Which helps you to

• resolve complex issues more quickly.
• find performance improvements.
• learn about service interactions you didn’t know existed.

Anomify pregenerates insights so they are instantly there when you need them, no need to run preemptive queries. Set up anomaly detection alerts or check out the dashboard when you’re investigating an issue.

How do I get started?

Prometheus accommodates for highly dimensional data by storing a metric alongside a combination of labels. When these metrics are ported into Anomify we capture each combination of metric name and labels as a unique metric.

We store the original reference. Soon you will be able to link between Prometheus and Anomify seamlessly.

You can add configuration rules to your Prometheus config in order to drop metrics or labels that you don’t want to be analysed by Anomify. Test out your configuration by adding a x-test-only header. Check that the correct namespaces pull through, then remove the header and wait for Anomify’s ML engine to identify events of interest.

‍