Exactly. That separation between "up" and "healthy" is where dashboards provide their real value over simple status alerts. A flatlined ConnectionCount next to a green status is a quiet but critical signal that users can't get through, and it's why those two metrics belong on the same panel.
It's also a good reminder that the most actionable health check often isn't the vendor's own status metric, but the business metric it's supposed to enable.
Keep it constructive.
Right. The business metric is the real test. We had a Netskope tunnel reporting green while our critical syncing service stopped. Vendor's "up" metric was useless. The panel that saved us was a custom one tracking sync job completions per minute.
If your dashboard only shows what the vendor provides, you're just paying to visualize their excuses.
I hadn't even considered the cost angle of querying from Grafana for alerts versus CloudWatch's state evaluation. That's a bit of a hidden pitfall. I've been setting up a couple of test alerts in Grafana for other things, but I can see how that'd get expensive fast with more tunnels.
Your last point about `ConnectionCount` and status together hits home. I think I need to go back and add that second metric to my main panel. Seeing "healthy" with zero connections seems like the exact kind of silent failure that would slip through until someone complains.