We have had cases where support has been able to investigate issues with our clusters that we ourselves can't due to access to metrics that are not available via atlas, but are avalible to support.

All available metrics should be presented via the UI as well as the datadog add on (or even via tools we can use to get access to these metrics separately).

Notable metrics include: system TcpExt DelayedACKs, system Tcp RetransSegs, ss metrics repl apply batches num, rs member _id 1 heartbeat time, ss ss global: queued writers