It’s Time To Kill Staging: The Case for Testing in Production

It’s Time To Kill Staging: The Case for Testing in Production

Staging environments have been a bottleneck for 20 years, but new isolation methods enable safe testing in live systems. DoorDash and Uber have already eliminated staging, reducing deployment delays by 70%.

Why This Matters

Staging environments are a relic of an era when infrastructure duplication was costly. The ideal model would test code in production with zero risk, but staging’s fidelity gap—missing real data, traffic, and IAM policies—creates dangerous bugs. Staging also introduces bottlenecks: 50 developers merging code can cause test failures due to conflicting deployments, not code quality. This slows velocity and inflates costs, with some teams spending $1M+ annually on staging infrastructure.

Key Insights

• “Staging bottlenecks cost companies hours in CI/CD delays”: 40-minute test suites and multihour deployment queues are common in microservices-heavy organizations.
• “Request-level isolation with Kubernetes enables safe production testing”: Sandboxes route test traffic to ephemeral services while using production dependencies.
• “DoorDash and Uber eliminated staging for faster, reliable deployments”: These companies reduced infrastructure costs by 60% while improving test accuracy.

Practical Applications

• Use Case: DoorDash uses request isolation to test microservices without staging, ensuring real-world data interactions.
• Pitfall: Overlooking data isolation headers can corrupt production databases if test writes are not strictly routed to ephemeral stores.

References:

• https://dev.to/signadot/its-time-to-kill-staging-the-case-for-testing-in-production-521a