← All work
Server lifecycle automation: provisioning and decommissioning
Outcome: Both ends of the lifecycle became one-request, auditable operations. Provisioning reliability climbed from 60% to 99% and went fully agentless, while decommissioning became a staged, reversible flow that saved countless hours and let a mistaken request be caught and rolled back before any data was lost.
The problem
The server lifecycle was manual at both ends. Standing a server up ran as a click-through process with brittle, agent-based registration that failed to land cleanly about 40% of the time. Tearing one down meant touching a long chain of systems by hand: slow, easy to get wrong, and leaving orphaned records scattered across the environment.
What I did
- Rebuilt provisioning at the health system as a fully agentless process on a CI/CD-managed Ansible Automation Platform, owning the Windows and RHEL path end to end, from IPAM reservation to a compliant, managed host.
- Built a ServiceNow-driven decommissioning workflow that orchestrated teardown across SCCM, Active Directory, DNS, VMware, SCOM, CyberArk, and NetBackup from a single request form.
- Made both ends auditable and repeatable, driven from a request form rather than a runbook of manual steps.
- Staged the decommission deliberately, with a two-week powered-off hold before final teardown, so a server flagged in error could still be recovered.
- Added cross-team visibility and a revert path so Operations could watch application-team tickets and reverse a wrong shutdown before anything was destroyed.