What would you like to happen?

Much like #20687 to read to BigQuery, The Java SDK contains an implementation which uses the BigQuery Storage API to write to BigQuery. We should implement the same for Python.

In our case, we would like to use it instead of using the FILE_LOADS ingestion method in streaming mode, where errors during the loading process are not propagated to the pipeline. We are currently using it as we do not need STREAMING_INSERT features, and hence want to save on cost.

Issue Priority

Priority: 3

Issue Component

Component: io-py-gcp