Notes from a change-day checklist that actually got used
The checklist was one page. Before traffic moved: confirm the application path under study still emits the agreed continuity properties. During the cutover: capture a five-minute baseline of those properties at both old and new edges. After: compare error and latency distributions only on sessions that kept identity continuity.
What they skipped previously was the middle step. Without a dual-edge baseline, every spike looked like a circuit failure. With it, they saw a configuration mismatch on the application side that would have been blamed on the managed link for weeks.
We now include a change-day checklist in Session Continuity Mapping and refresh it during Ops Briefing months when major overlay work is planned. The document is deliberately short; long runbooks do not get opened at 02:00.
If your next cutover already has a runbook for routers but nothing for application measurement, ask for a briefing slot. The cost of an empty measurement window is usually higher than the cost of a two-hour review.