Deployment management processes and activities
Domain 2: Processes and activities
This domain tests the end-to-end deployment workflow and the reasoning behind approach selection. Expect sequencing questions and questions on deployment plans, verification, and rollback.
Core rules
- Typical activities: plan scope, approach, resources, schedule, and dependencies; design the deployment steps, verification, rollback, and communication; build or assemble the deployment unit; test in a controlled environment; deploy to target; verify; and close with records and lessons.
- A deployment plan states what moves, where, when, by whom, in what order, how success is verified, and how to roll back. Dependencies and maintenance windows belong in the plan.
- Approaches: big bang, all at once; phased, by site, user group, or component; continuous, through an automated pipeline; push, centrally controlled; and pull, where the target initiates. Automation suits repeatable, low-variance deployments; manual control suits high-risk or regulated cutovers.
- Verification is not optional. Compare actual state with expected state, check health checks and monitoring, and confirm the deployment record.
Worked example
A retailer must deploy a new POS version to 400 stores. Big bang is rejected because a failure would stop all sales. A phased deployment starts with 10 pilot stores. Success criteria: no P1 incidents for 48 hours, transaction latency within threshold, and reconciliation balanced. After the pilot passes, waves of 50 stores follow, with rollback to the previous image if criteria fail. The reasoning is risk reduction through limited blast radius, not speed alone.
Traps
- Treating deployment as only the final copy step.
- Omitting rollback or verification.
- Choosing big bang by default.
- Forgetting documentation, data migration, or configuration changes.