I’d like to setup a Knative Serving for a bunch of revisions (50k, most of them scaled to 0) The serving code will be fairly long-lived, up to 1 min, and bi-directionally communicate with a caller. It’d be great if one bi-directional session would be counted as concurrency = 1.
Is this supported? Would protocol would be better? Pure http/2, gRPC, websocket?
So far I tried to look for the solution in the docs. No luck.
