How often is a stream archived to a dataset?

I have created a stream and pushed 3 rows into it using the stream-proxy API.

After 30 minutes the first archive ran and my 3 rows show up in the dataset.

Few questions:

  • Where is this 30 minutes documented?
  • Is it static or can it be configured?

I think this should be configurable on our side so we can make the tradeoff calculation between latency and costs.

Thanks