02/13/2023, 8:39 AM
hi, i'm going to inject spicedb to my project. the thing is that we use graphql subscriptions and i wanna check availability every event (check permission) for every connected client before sending. By my guess it will be least of 1000 checks per second. what cluster configuration is needed for my request?


02/13/2023, 9:38 AM
it largely depends on the schema and the dataset, so it's hard to tell from the outside. With that said, an in an optimistic scenario of a relatively simple schema, for 1K I'd start with 3 replicas with at least 2 cores each, and then scale out as you see fit, potentially up to 5 pods.