Cloud & reliability
A production-readiness checklist for fewer blind spots
Access, deployment, observation and recovery: questions to settle before production.
Before deployment: know what is changing and where
A dependable release starts with an identifiable version, a known environment and an explicit dependency list.
Data changes need their own sequence, compatibility checks and a route back to a stable state.
- Identify the version and change set.
- Confirm environments and dependencies.
- Verify access and secrets without exposing them in logs.
- Prepare a recovery approach appropriate to the change.
During deployment: make the path verifiable
Every important step should produce a state people can interpret. Automated checks reduce omissions, while stop criteria and final responsibility remain explicit.
After deployment: observe journeys and dependencies
Availability alone may not be enough. Early checks should cover critical journeys, dependency exchanges and failures that prevent use.
Alerts are useful when they identify the service, signal and first available context for action.
Finish with an operable handover
Delivery includes access, ownership, observation points and the response to common incidents. These should not depend on one person’s memory.