rolling deploy
also: rolling update, staged deployment, canary deployment (similar concept)
A deployment strategy where updates are gradually rolled out to a subset of servers or instances at a time, keeping the service running throughout the process without downtime.
Rolling deploy is a release technique commonly used in containerized environments like Kubernetes and traditional server farms. Instead of updating all instances simultaneously (which causes downtime), new versions are deployed to a few instances while others continue serving traffic, then gradually expanded to all instances.
This approach maintains service availability and allows quick rollback if issues are detected. For example, with 10 web servers, you might update 2 at a time, wait for health checks to pass, then proceed to the next batch. If the new version has problems, you can stop the deployment before all instances are affected.
Rolling deploys work especially well with container orchestration platforms. In Kubernetes, you can control the rollout with parameters like maxSurge (new pods created before old ones terminate) and maxUnavailable (max pods allowed to be down during update), ensuring smooth transitions with zero or minimal downtime.