Separating Deployment from Release
What you'll learn: How to safely get code into production without immediately activating it for users.
The Core Idea
In traditional software delivery, deployment (moving code to production servers) and release (making features available to users) happen simultaneously. But what if you could break that coupling? Separating deployment from release means you push code to production infrastructure, but control when and to whom that code becomes active—independently of the deployment itself.
Think of it like a restaurant preparing meals behind closed doors. The food is ready and on the counter (deployed), but the chef decides exactly when to ring the bell and send each dish to specific tables (release).
How It Works
Feature flags (also called feature toggles) are conditional statements in your code that control whether a feature executes. When deployed, the feature's code exists in production, but remains dormant until you flip the flag:
if feature_flag_enabled("new_checkout_flow", user):
new_checkout()
else:
old_checkout()
Dark launches take this further: you deploy new functionality but route zero user traffic to it initially. The code is live, instrumented, and observable, but not yet active. You can gradually enable it for internal users, beta testers, or small user percentages before a full release.
Why This Matters
This separation provides tremendous safety and flexibility. You've already met the "Production-Ready Definition" concept—now you can deploy production-ready code before you're ready to release the feature to customers. If something goes wrong, you toggle the flag off instantly without redeploying. You can test in production safely, perform gradual rollouts, and schedule releases independently of engineering deployments.
Key Takeaway: Separating deployment from release using feature flags and dark launches gives you independent control over what code exists in production versus what features users actually see, dramatically reducing deployment risk.