What’s the big idea? After scrolling and clicking I’m not sure what this gets me. What’s the durability guarantee? Is this at-least-once, at-most-once, or thoughts-and-prayers message delivery? What happens when a node crashes?
elixir has an even better offering baked into phoenix. unlike node which is limited to 1 process and necessitating a complex IPC cluster, erlang's vm can use all cpus on the host as well as connect different machines. The end result is a much more robust websocket solution that can relay state to whichever machine resumes the connection.
architecture docs are scant, but it appears this horizontally scales channels by adding "broker" instances and horizontally scales consumers by adding "worker" instances? i.e. a broker only needs to ever broadcast a message to the maximum number of worker instances in the cluster.