Service Migration

Service migration moves a running workload to a new runtime, account, region, or architecture. Treat it as a production change with explicit ownership and rollback.

Migration patterns

• Rehost: move with minimal application change.
• Replatform: move to a new runtime with limited code change.
• Refactor: change application architecture or boundaries.
• Replace: move capability to a different product or service.
• Retire: remove unused or superseded capability.

Readiness checklist

Before migration, confirm:

• Current owners and escalation paths.
• Deployment and rollback process.
• Configuration and secret sources.
• Data stores and dependencies.
• Network ingress, egress, DNS, and certificates.
• Dashboards, alerts, logs, and traces.
• Load, capacity, and performance baseline.

Cutover flow

Validation

Validate:

• Critical user journeys.
• Dependency calls.
• Background jobs and scheduled tasks.
• Metrics and alerts.
• Error rates and latency.
• Cost and scaling behavior.
• Security and access controls.

Watchouts

• DNS TTLs and client caching can delay rollback.
• Background workers can process duplicate work during cutover.
• Environment variables often hide undocumented dependencies.
• Cleanup must include old credentials, routes, alerts, and dashboards.