Why
Cloud advisors cover common patterns but miss service-specific optimisations. A per-service checklist ensures systematic coverage — non-compute services (storage, networking, databases) are frequently skipped in ad-hoc reviews but often represent 40–60% of spend.
What
Create a per-service optimisation checklist covering compute, storage, networking, and databases.
How
Build the Checklist
| Service Category | Checklist Items |
|---|---|
| Compute | Right instance family? Right size? Graviton/AMD? Spot eligible? Scheduling? |
| Storage | Lifecycle policies? Tiering? Unattached volumes? Snapshot retention? |
| Database | Right instance class? Read replicas needed? Reserved? Storage auto-scaling? |
| Networking | NAT gateway vs endpoints? Cross-AZ traffic? CDN for static? Egress paths? |
| Containers | Pod requests vs actual? Node group sizing? Spot for non-critical? HPA tuned? |
| Serverless | Memory right-sized? Provisioned concurrency needed? Duration optimised? |
Run Against Top Workloads
Apply the checklist to your top 10 workloads by spend. Document findings as optimisation tickets.
Deliverable Checklist
- Per-service checklist created
- Applied to top 10 workloads
- Findings documented as tickets