Aggregation rules are fantastic, especially if you run a large fleet (for example, aggregating away the instance/pod labels from histograms on ingest makes them much cheaper to store). Drop rules are a lifesaver when someone adds a bad metric and you need to stop it ASAP.

Their internal observability tooling is great. Their profiler is excellent for discovering cost-saving opportunities (by adding aggregation or drop rules). They also provide out-of-the-box dashboards for understanding how you are using the system, which helps answer questions like "how much of my limit is being used by job X?"

Extremely fast to query.

Very easy to onboard if you're already using Prometheus. The chronosphere collector is mostly compatible with existing Prometheus configs, and their frontend can import any of your existing Grafana dashboards.

Wonderful support. The Chronosphere team has been fantastic, not just during the implementation phase but also afterward. They've also been great at surfacing non-obvious issues to us in places where they haven't quite automated everything yet (for example, misconfigured notifiers and alerts). Review collected by and hosted on G2.com.

Prometheus/PromQL (which Chronosphere is based on) is still a complex system to learn. If you're already using it, Chronosphere is a fantastic drop-in replacement for Prometheus. However, if you or your users are new, expect to take some time learning the language and data model. If you don't know what you're doing, it's easy to make a query that looks mostly right but isn't quite. Review collected by and hosted on G2.com.