What is a 'rolling restart' in a web server deployment and why is it preferred over a full restart for zero-downtime deployments?

• A.A rolling restart reboots every instance at the same moment (a coordinated reload), loads the new code into memory, and returns the pool to service after a short pause. This ensures traffic is queued at the load balancer during the cycle, achieving zero downtime for clients that retry. In contrast, a full restart would cause a partial outage, a delay of several minutes
• B.A rolling restart replaces the instances one at a time (or in small batches) with a second environment of the same size, then switches the load balancer to the new pool in a single step, and the old pool is deleted at once. This gives all users the new version together, achieving zero downtime. In contrast, a full restart would cause a complete outage, a step that doubles capacity for a short window
• C.A rolling restart takes instances out of rotation one at a time (or in small batches), restarts them with new code/config, and returns them to rotation before proceeding to the next batch. This ensures at least some instances are always serving traffic during the restart cycle, achieving zero downtime. In contrast, a full restart would cause a complete outage