Rolling Deployments
Practice this topic in a realistic system design interview
Most deployments have two goals: keep the service online and avoid paying for a second full copy of production.
A rolling deployment does this by replacing servers or containers gradually.
Some instances start running the new version. Some old instances finish their current work and shut down. The rest of the service keeps serving users.
The tricky part is the mixed-version window: the period when old and new code run at the same time. During that window, both versions may use the same database, load balancer, queues, and caches. Many rolling deployment failures happen there.
This chapter explains how a rolling update moves through the service, how health checks and graceful shutdown keep it safe, what the mixed-version window requires, and when rolling is the right fit.