AWS infrastructure assessment
Read-only and data-driven, to answer why CPU spikes and why deploys hang.
My role
I ran the assessment and wrote the report and the proposal (14 of 17 commits).
- 615
- security checks
- 24%
- of annual spend flagged as waste
- 111k
- lines of Java reviewed
Problem
The client, an engineering consultancy whose systems serve public agencies, reported CPU spikes and Jenkins deploys hanging, and wanted rolling deploys, observability and redundancy.
Solution
Read-only collection under a restricted IAM policy, using Steampipe and Prowler, plus a source code review. I correlated CPU with network traffic and showed that 25% of the spikes lined up with deploy artifact uploads and 75% were normal load fighting for CPU with PostgreSQL, which ran on the same host as the app.
Highlights
- 615 Prowler checks, with continuity and access risks prioritized.
- 24% of annual spend flagged as immediate waste, including extended support on an end-of-life database that cost more than the instance itself.
- Reviewed 111k lines of Java and 301 migrations. Auth was already stateless, which took 60 hours off the scope.
- A proposal split into modular workstreams, with pricing.
Result
Delivered with a transformation roadmap and a commercial proposal.