question about scaling proccess/revise postgrest README

[mananamenos]

Hi, in postgrest github readme you state: Being stateless to allow horizontal scaling. If there were a need to scale, what would be the differences comparing it with traditional haskell web server. In traditional server, i suppose you would need a load balancer first to scale horizontally. However in case of postgrest, you run ./postgrest which creates a process (like a server), which would need to some kind of load balancing also in case of too many requests, right? Or do you want to say that only db layer scaling would be needed (like using EnterpriseDb). Maybe someone could give me a high level picture of high level, basic execution steps in a scenario where scaling would be needed. Thanks

[steve-chavez]

The README needs revising, I'm also not sure about that sentence related to the Hasql library.

However postgrest is like another pg client, you can scale it horizontally. Though ideally it'd live on the same box as pg connected with a unix socket.

Or do you want to say that only db layer scaling would be needed (like using EnterpriseDb).

Not necessarily, replicas(pg + pgrst) could be added for read performance.

I'll move this to docs, I'll see about adding a section about scaling in the next version.

[wolfgangwalther]

I think the README is fine as it is in that regard.