Dataflow Batch or Streaming insert to BigQuery clarification

Question

Given that there is cost involved with Streaming inserts to BigQuery, does using Dataflow to read from a Bounded Collection (or a batch mode) incur cost from the BigQuery perspective? And I believe only when reading from Unbounded collection(in streaming mode) and writing to BigQuery is considered a Streaming insert to BigQuery?

Answer 1

You are correct, as specified in the sources:

If the input is bounded, then file loads will be used. If the input is unbounded, then streaming inserts will be used.

You can overwrite the default behaviour by using .setMethod(Write.Method.FILE_LOADS) or .setMethod(Write.Method.STREAMING_INSERTS) on the BigQueryIO builder object.