Not so much whether the code will be performant enough to service concurrent websocket requests, but other operational concerns such as, websocket balancing across the cluster, authentication and access logs, failure modes, or usage isolation, just to name a few things off the top of my head.