The Backbone Audit: A Checklist You Will Actually Use

A quarterly walk: drift, single points, doc rot, capacity creep, and the four questions to ask each box.

Audits fail from ambition: sixty pages nobody executes. This one fits a shift.

Quarterly, per box

• Config drift vs. blueprint — diff it, explain every line
• Single points of failure: double-check the "dual" supplies that share one breaker
• Firmware age and the CVE list against your RUNNING version, not the latest
• Capacity slope: ports, queues, and the graphs that bend before the page fires

Per path

Failure-domain walk: each pair's failover: each pair's failover actually exercised, not assumed from logs that auto-refresh the wrong counter.

The real test

Nobody may survive documentation retirement — reach each thing in under 3 steps, from inside the building if Google is unreachable. Explain your network to a new hire in one morning? If the walk takes more, the rot started above the core. Complete the audit; bank the fixes in the normal change mill, not an unusually brave Friday.