Datadog
A monitoring platform that gathers logs, metrics and the path of a request across your services in one place. An agent sits on each machine and ships everything, and you build dashboards and alerts on top. The value is not in any single one of those three streams, it is in walking from an alert to the offending trace to the log line without switching tools.
Strengths
- Logs, metrics and traces in one place: you go from the alert to the offending line without switching tools
- Answers which of your services caused the outage, a question error tracking alone cannot handle
- Covers infrastructure as much as application code, so one on-call rotation for both
Limitations
- Billed on log volume and host count: the real cost shows up at the end of the month
- With no several services to correlate, you pay for a capability you never exercise
- Setup and tuning need someone who knows what they want to measure, otherwise dashboards stay green and empty
Best for
- A production application spread across several services that call each other
- An on-call team that has to find the cause of a slowdown, not just be told about it