For AirByte on Kubernetes, which components benefit from replication factor > 1? ie: do any of the components benefit from the ability to horizontally scale? Coming from deploying this via docker-compose there weren't many options, and I don't really see an operations guide for running AirByte on K8s -- so without knowing more about the internal architecture I'll be mostly guessing.

This variable is applicable to Kubernetes deployment where you can have multiple workers to achieve higher parallel sync connections or servers to have a HA instance

@Marcos Marx (Airbyte) thanks. So only

worker

and

server

benefit from higher concurrency?

temporal

,

webapp

,

cron

, and

connector-builder-server

do not? And what does the higher concurrency for

worker

and

server

enable it to do? • More workers = more capacity for jobs? • More servers = better response for api calls from workers (in addition to redundancy for outage resilience)? Are there diminishing returns after a certain number? (eg:, after 3 servers you're bottleneck will be postgres, or after 10 workers your bottleneck will be temporal)

Not really, temporal shouldn't so intensive resource consumer as worker. The others services (temporal webapp, cron...) they don't consume too much resources. As I said before, more servers allow you to have high availability in case in one instance break, worker it is needed to achieve more jobs. The temporal of course can be scale horizontal but you must achieve a lot of concurrent syncs to need it