The cutover weekend, hour by hour
Uneventful cutovers are engineered, not lucky. Here's the shape of a boring one.
Good cutovers are boring. The excitement all happened weeks earlier, in dry runs where surprises are cheap. By the time the real weekend arrives, the run is a rehearsed sequence with timestamps, owners, and a decision framework — not a war room improvising.
Before Friday
At least two full dry runs against production-scale extracts, each ending in full reconciliation. The second dry run exists to prove the fixes from the first. If dry run two produced surprises, there is a dry run three — and the cutover date moves before the corner gets cut. That discipline is easier to hold when the price is fixed and the delay is our cost, not a change request.
The weekend itself
Friday evening: source freeze, final delta extract, checksums agreed before anything transforms. Saturday: load, then the reconciliation harness runs — every record, every total, exceptions queued with owners. Sunday morning: the go/no-go, made against pre-agreed criteria written down weeks before, when everyone was calm. Numbers decide, not adrenaline. Rollback is a rehearsed script, not a threat.
Monday
Users arrive to the new system. Hypercare has named engineers and a triage channel, and the evidence pack is already being compiled — because reconciliation ran as part of the load, not as an afterthought. Two weeks later, compliance files a signed pack, and the migration ends the way it should: quietly.
Got a cutover date? Tell us the source system, the target, and the deadline — we'll tell you within 48 hours whether we can hit it and what the Assessment will cost.
freddie@godwit.uk