We use it here at Streak for streaming log processing. We have it setup such that our backend and client side logs are streamed to our "dataflow job" where we d

We use it here at Streak for streaming log processing. We have it setup such that our backend and client side logs are streamed to our "dataflow job" where we do some pretty simple processing/transformations and then it gets outputted to BigQuery. Sounds simple but there is a lot of complexity its hiding when you're streaming at a large enough scale. We know because we built our infrastructure for this at first and it sucked. As for pricing, its just consumes your regular google compute engine instances, so its all based on how big your jobs are and how long they run for.