AWS drops price for managed Prometheus service by 84%

I'm the worst kind of person to reply, because my brain was already damaged by Borgmon before Prometheus was a thing. But I've run both at relative scale, and I'll take prometheus anyday. Datadogs query language is lacking and they've never added decent support for ratio-of-rate metrics, which are most of the metric queries you should be alerting on. Datadog is expensive, and because it's expensive you'll hold back on some of your most important and expressive metrics. Prometheus scales very far vertically - Far enough that I've yet to use it's sharding features, despite abusing tags badly.

Prometheus's query language is a sharp tool, and you can mess it up. It takes getting used to, and Grafana is a must for making charts navigable to executives (which is one of the top selling points of Datadog: Execs can see and build dashboards for the metrics they want). I would not be likely switch to another system - Even Monarch, Google's next generation monitoring system, has significant downsides compared to what you can do with Prometheus.