Plan a deploy
Set up the deploy: the changes, teams, stages, steps and gates.
Do this weeks before the deploy, with every team. Each team fills in its own steps.
1. Overview
Fill in Deploy name, Planned start and Must be finished by.
2. Changes
One row per change: a feature, a fix, a data model change. Give each an ID such as F-01, and a Business owner. Steps, smoke tests and data migration rows point at a change by its ID.
3. Teams
One row per team taking part, including a team that has not said what it needs to do yet. A team with no steps shows No steps listed.
4. Stages
List the stages in Order: for example Pre-deploy, Deploy, Data migration, Post-deploy checks, Business testing.
- Planned wait after (hrs): time after the stage that is not work, such as overnight or waiting for testers.
- Point of no return: tick it when rolling back gets much harder once the stage is done.
- What must be true to go on: what the gate checks.
5. Steps
Every team adds its steps to the one Steps tab. For each step: ID, Run order, Stage, Team, Change (ID), Step, How to do it, Owner and Backup owner.
Then the estimate, the waits and handoffs, and the rollback. Each has its own guide: Estimates and practice runs, Order, waits and handoffs.
6. Rollback for every step
Fill in Rollback steps and Rollback hours. A step without them makes its team Not ready, and its time is missing from "Hours to roll back everything".
7. Gates
On the Gates tab, add one row for every team that must say Go at the end of each stage: After stage and Team. On the day, each team fills in Call.
8. People and the rollback plan
List who is on the deploy and when they are available on People. On Rollback plan, write what counts as P0 to P3 for each system, what you do, and who decides.