Everyone knows what their services do. Almost nobody knows what else stops when one of them dies. Map the dependencies, kill something, and watch the cascade — the answer is usually smaller than your lab and larger than you expected.
Killed = what you switched off. Hatched amber = collateral, things that cannot work without it.
Tick what each service needs in order to work. Be honest about DNS.
| Service | Needs |
|---|
A dependency you forgot to draw is still a dependency. The three people always miss: DNS, the hypervisor host underneath a container, and the switch everything is plugged into. A Peira Labs field tool · peira.dev.