Hi Team, we have self hosted temporal server on k8s. I want to understand the throughput that a temporal worker can serve, of course it would depend on workflow and activities, but still what could be the rough number. My use case is such that i have about 200 workflows each on dedicated task queues, but each workflow can have about million of runs per day, so how the scaling of temporal server look like, how many worker pods, history pods, number of shards, etc. ?Also should the retention policy of the namespace be small to prevent any outages ? What are the possible bottlenecks at this scale and how do we mitigate them ?
Thanks !