An independent read of the system before the next commitment.
An architecture review examines the system as it is: its boundaries, constraints, and operational model, read against the decision in front of you, by a reviewer with no stake in the implementation that follows.
When to bring this in.
- A consequential change is pending
- Growth, replacement, or a platform shift will commit years of budget. The design deserves an outside read first.
- The architecture is contested
- Credible engineers disagree. A review replaces the strongest voice with the strongest evidence.
- Confidence is asserted, not shown
- The system may well be fine. A review turns that maybe into a statement with reasons.
What we examine.
- System boundaries and coupling
- Where the seams are, which ones hold, and which ones transmit every change.
- Quality attributes and constraints
- What the system must guarantee, and where the design depends on hope.
- Data and integration architecture
- How information moves, where it is owned, and what each integration assumes.
- Deployment and operational model
- How the system reaches production and behaves when parts of it fail.
- Security and dependency exposure
- The surfaces an attacker or an abandoned dependency would find.
- Changeability and migration risk
- What the next change will cost, and which decisions are still reversible.
What you receive.
- The question, stated
- The decision under review, agreed before the work starts.
- Evidence, not impressions
- Findings tied to the code, the architecture, and the runtime behaviour actually observed.
- A recommendation and its rejected alternative
- What we would do, what we would not, and why both.
- A record you can re-examine
- The reasoning stands on its own after we leave.
How it runs.
- Frame
- Agree the question and what evidence would settle it.
- Examine
- The system, its code, and its operations.
- Compare
- Options weighed against constraints, with rejected paths kept visible.
- Recommend
- A direction stated plainly, with its costs.
- Record
- The review document, written to be argued with.
Put the architecture in front of independent eyes.
Say what decision is coming. The review is scoped to the question that cannot wait. The proposal states the scope, the evidence required, the duration, and the fee before the engagement begins.