Lesson 17 / 21

Rolling Deployments

Replace instances gradually, one batch at a time, with no extra environment.

How it works

Instead of two full environments, a rolling deployment replaces old instances with new ones a few at a time — take one out, deploy new, health-check it, add it back, repeat — while the rest keep serving traffic.

Cheaper, but slower to reverse

Rolling deployments need no duplicate infrastructure, unlike blue-green. The tradeoff: rollback means rolling the old version back out the same gradual way, and two versions run side-by-side during the rollout.

Resurfacing a highway lane by lane

Instead of closing the whole highway (blue-green) or a tiny test patch (canary), crews close one lane at a time and reopen it before moving to the next — traffic never fully stops, but there's a window where old and new surface coexist.

Quick check: Which deployment strategy needs no duplicate full-size environment?

  • Blue-green
  • Rolling deployment
  • Both require the same extra capacity
Answer

Rolling deployment — Rolling deployments replace instances gradually within the existing fleet, unlike blue-green which needs a full second environment.