The Building That Runs the Cloud
The Alarm Has to Work Too
Uptime depends on what you can see as much as what’s running.
A chiller can be humming along while the screen that watches it has gone blank.
That's a dangerous place to be. If temperature readings stop arriving, nobody knows the room is warming until a server throttles itself. If an alarm path breaks, the first warning may be a tenant on the phone.
So in a data center, the service check has to cover more than "is the device running?" It has to ask whether the team can still see it, hear from it and reach it the way they need to.
Two checks, every time
Every Findavi action ends with two checks. Each one has to pass on its own.
Two checks. Both must pass.
| After this action | Security check | Service check |
|---|---|---|
| Restrict a cooling controller | The unknown address can’t be reached | Temperature readings, alerts and the link to the building management system still work |
| Shut off an unapproved vendor session | The session is blocked and can’t reconnect | The approved emergency access process still works |
| Isolate an access control device | The attack path is contained | Authorized staff and emergency exits keep the access they need |
| Push a new device setting | The old weak setting is gone | The device reconnects normally to its approved systems |
When to keep the case open
If either check fails, the issue stays open. If either check can't be proven, the issue stays open too.
That sounds strict. It should be. "We think it's fine" is how door five happens.
A closed ticket should mean two things you can show anyone. The bad path is gone. The building is still doing its job.