Reading failure modes before funding a platform year

What an architecture review should pressure-test when leadership wants a multi-quarter microservices programme.

A programme slide that says “move to microservices” is not a risk register. Before funding a year of extraction, ask four questions that fit on one page:

  1. Which data stores are shared across the proposed service list?
  2. Which releases today require coordinated deploys?
  3. Which on-call rotations already blur across those boundaries?
  4. What happens to a customer journey if the new network hop fails half the time for ten minutes?

If the answers are vague, the money will buy ceremonies rather than safer releases.

A fixed-scope architecture review exists for this moment: interviews, a written risk register, and a readout timed to a decision date — not an open-ended discovery that never closes.