Critical Flow Reliability Assessment
Know what is verified and what needs evidence.
Critical Migration
Availability can survive a migration while transaction continuity fails. Follow the flow across old and new states, including rollback.
Questions worth answering
- What happens to in-flight work at cutover?
- How will consistency be checked across both states?
- When does rollback cease to be safe?
The outcome to protect
Protect in-flight work and data consistency as responsibility moves between environments.
How we investigate
Examine dependency changes, cutover sequencing, data validation, rollback limits, access, and post-cutover observation.
Decisions & outcomes
- Cutover dependency and ownership map
- Continuity and reconciliation checks
- Explicit rollback limits and open questions
Follow the last write across the cutover
A fictional scenario, not a customer result.
The question to resolveWill every accepted account update appear exactly once after a database migration?
- Source accepts update
- Replication catches up
- Writes switch target
- Records reconciled
Rehearsal evidence
The rehearsal reconciled the copied dataset. Writes accepted during the final handover window were excluded from that comparison.
- Evidence basis
- Tested
- Verification result
- Partially verified
- Freshness
- Fresh
Unproven assumption
Replication lag suggests a short catch-up window, but it does not establish that every acknowledged write reached the target.
- Evidence basis
- Inferred
- Verification result
- Unverified
- Freshness
- Unknown
The resulting decision
Define the write boundary, rehearse catch-up and reconciliation, and agree an abort point before changing the production writer.
Scope of this example
The example distinguishes data-copy success from end-to-end correctness. It does not certify migration completeness or a zero-loss outcome.
Relevant service
Choose one critical flow.
Tell us what must work and what you need to know. We agree the scope before delivery.