Cloud monitoring is a complex thing and several successful startups and services are out there to meet your needs. As you grow, you will find that the cost begins to outweigh the benefit and you will be tempted to search for an alternative.

I would like to share how we migrated from Datadog to a scaled open-source solution, as a combination of Prometheus, AlertManager, Grafana, and cut our ~2000 host setup substantially while improving our observability at the same time.

I will also present some in-house algorithms for smarter anomaly detection using nothing but Prometheus!