Yeah, I’ve seen that. If you treat each map as a micro‑service you can spin up a server when the queue rises and tear it down when traffic drops. The key is to keep your load balancer smart: monitor ping, packet loss, and player count per region. Then set thresholds that trigger a pod replication. It keeps latency low without overspending. Also think about state sharding—don’t let one map hog memory, split it into zones, each with its own container. Keeps the battle flow smooth and the lag in the dust.