What is Production Readiness?

#What production readiness covers

A readiness review checks several categories. Observability: does the service emit the signals needed to detect and diagnose problems. Alerting: do alerts exist for user affecting conditions, and do they page someone real.

Operability: is there a runbook, a tested rollback path, and a documented owner. Capacity: is expected load known and has the service been tested against it. Dependencies: what does the service require, and what happens when each of those fails.

#Why readiness reviews matter

Nearly every gap they catch is far cheaper to close before launch than after.

• Cheap fixes: adding instrumentation before launch costs hours, adding it mid incident costs an outage.
• Clear ownership: the service has a named owner before it needs one at 3am.
• Known capacity: load limits are measured rather than discovered under real traffic.
• Consistent baseline: every service meets the same operational floor.

#Limits and failure modes

Readiness checklists become theater when they are completed for approval rather than for correctness. A service can be marked ready with alerts that page a rotation nobody staffs and a runbook that describes an earlier architecture.

Readiness is also a snapshot. A service that met the criteria at launch drifts as dependencies change, traffic grows, and ownership moves. Without periodic re review, the checklist describes a state that stopped being true some time ago.

#Production readiness vs launch approval

Launch approval is a decision about whether to proceed, typically involving product, business, and timing considerations. It answers whether the organization wants this live now.

Production readiness is a technical assessment of whether the service can be operated safely. The two are frequently conflated, and when readiness becomes a gate that business pressure can override, it loses its function. Keeping them separate lets a launch proceed with explicitly accepted operational risk, which is a legitimate decision, rather than pretending the risk was assessed away.

#Key takeaways

• Readiness covers observability, alerting, operability, capacity, and dependency failure behavior.
• Gaps are far cheaper to close before launch than during the first incident.
• Checklists become theater when completed for approval rather than for correctness.
• Readiness is a snapshot that drifts, so periodic re review is required.

#Frequently asked

Who conducts a production readiness review?

Usually an SRE or platform team with operational experience, working with the team that built the service. External review catches gaps the authors have stopped seeing.

What is the most commonly missing item?

A tested rollback path. Teams frequently have a theoretical rollback that has never been executed, which means it has never been verified.

Should readiness ever block a launch?

It should be able to, or it carries no weight. The alternative is explicit acceptance of the identified risk by someone accountable for the outcome.