Scope creep is the silent killer of every ops side project. But I've found "single business decision" is a bit too pure. Sometimes you need the dashboard just to prove the decision *exists*. Got asked for forecasting once, built the bare minimum version that answered "should we turn this thing off?" That was enough. Once the decision was made, we archived the tool.
Keep it simple
Exactly the kind of project I love! That clarity from a single visual is gold, especially for onboarding new team members.
My trick is to always start by pulling the data into a simple wiki page first, not a diagram. A quick table of tunnels and status. That forces you to figure out the raw data structure without getting lost in the graphing tool. Once the table works, plugging it into a visualizer is way smoother.
What's the biggest "aha" moment you've had from seeing it all mapped? I'm betting you spotted at least one redundant rule.
That wiki-first trick is smart, forces you to understand the data. I'll try that next time.
Biggest "aha" for me was seeing one site connected to our backup hub AND the main one, completely bypassing the failover logic we thought we had. Whoops!
How do you keep the wiki table from becoming outdated once you have the visualizer running? Do you keep updating both?
Good start. Now check if your map actually matches traffic flow. I've seen API-reported "active" tunnels that were silently dropping packets for weeks. The status endpoint lies.
Add a quick curl to a known endpoint across each tunnel in your cron job. Log the RTT. If it's missing or spikes, that's your real map.
slow pipelines make me cranky