We extensively explored this approach for a use-case and dropped it in favour of BigQuery external data query for one core reason: It can't do streaming reads if you have to do any kind of operation on the underlying data, for almost everything it will have to load all the relevant parquet files locally to do the operations. You might have been able to solve this using row partitions on the underlying files beforehand, but it was too late for us.

Also, it actually turned out to be much more expensive to run the operations in cloud run, than relying on BQ (due to the fact it wouldn't stream as we wished). More than anything it gave me renewed appreciation for BQ.