Idle Resource Detection and Cleanup

Idle resources are 5-15% of any mature cloud bill. Cleanup is mechanical once detection is in place.

Why idle accumulates

Resources outlive their creators. Engineers leave; environments are forgotten; volumes detached for ‘just in case’ sit forever.

Without active cleanup, idle resources only grow.

Four detection patterns

• 1. Unattached volumes (no instance for >30 days).
• 2. Old snapshots (older than retention policy).
• 3. Unused load balancers (no traffic for 14 days).
• 4. Idle dev environments (no logins for 30 days).

Auto-cleanup pipeline

Daily scan; tag for archive; weekly archive; monthly delete unless tagged keep-alive.

Pipeline runs forever; the savings recur.

Owner-of-record opt-out

Owners can tag for keep-alive with a justification. Justification reviewed quarterly.

Without an opt-out path, real legitimate resources get nuked.

Antipatterns

• Manual cleanup. Always falls behind.
• Aggressive auto-delete with no notice. Outages.
• Cleanup without owner-of-record. No accountability.

What to do this week

Three moves. (1) Apply this lever to your highest-spend workload. (2) Measure the dollar impact for one month. (3) Roll the practice out to the next two services if the savings hold.