PHASE 03 // IMPLEMENT

recfo@implement:~/runbooks/s5-03
recfo@implement:~/runbooks/s5-03 $ cat README.md
S5-03 · Optimize Usage & Cost · Usage Optimization

Identify Business-Value Recommendations

Why

Advisors optimise what exists but never ask whether it should exist at all. Decommissioning saves 100%. Non-prod environments are typically 30–50% of spend but receive little attention. Old logs, forgotten POCs, and orphaned infrastructure accumulate silently.

What

Evaluate resources for business-value relevance: should this exist? Is dev running 24/7? Are there zombie workloads with zero traffic? Can we delete old logs beyond retention?

How

Identify Zombie Workloads

Cross-reference cost data with usage metrics (CPU, network, request count). Resources with >30 days of near-zero usage are decommission candidates. Require product owner sign-off before action.

Audit Non-Production Environments

Calculate non-prod spend as a percentage of total. Identify resources running 24/7 that could be scheduled (dev, staging, sandbox). Target: non-prod compute should run <40% of hours (business hours + buffer).

Review Data Retention

Identify logs, snapshots, and backups exceeding retention policy. Common wins: CloudWatch log groups with infinite retention, daily snapshots kept for years, S3 buckets with no lifecycle policy.

Deliverable Checklist

  • Zombie workload scan completed (cost vs usage cross-reference)
  • Non-prod scheduling opportunities identified with savings estimate
  • Data retention audit completed
  • Findings documented as tickets with product owner sign-off required