bigquery: set PreferredMinStreamCountwhile creating CreateReadSessionRequest

[k-anshul]

Is your feature request related to a problem? Please describe.
While creating CreateReadSessionRequest we are setting MaxStreamCount as 0 which as I understand will leave the decision on server to choose the correct stream count.
On my testing on various public and private data sources I always found the number of stream to be 1.
I would like to set PreferredMinStreamCount as readClientSettings.maxWorkerCount as hint to the server for desired minimum streams.

Describe the solution you'd like
I would like to set PreferredMinStreamCount as readClientSettings.maxWorkerCount as hint to the server for desired minimum streams.

Describe alternatives you've considered
Since there is no option in sdk that allow user to configure this there are no other alternatives.

Additional context
https://cloud.google.com/bigquery/docs/reference/storage/rpc/google.cloud.bigquery.storage.v1#createreadsessionrequest

[alvarowolfx]

@k-anshul I think the configuration that you want to tweak here is the maxStreamCount instead of maxWorkerCount, because that's what we send to the Storage Read API as can be checked here. And the default value for it is already 0, which will signal the backend to produce as many streams to generate some reasonable throughput. But in some scenario, like ordered queries, our client and the backend can produce only one stream, so ordering can be guaranteed.

In the light of that, can you give more details on your use case ? I'm assuming that you're using the Storage API through the EnableStorageReadClient method. Maybe what is happening on your case is that your queries are ordered, which will cause only one stream to be generated.

[k-anshul]

@alvarowolfx Thanks for your reply.
I am running a select * from table limit <num> query on both public and private sources so ordering is not required in my case. Also I confirmed that maxStreamCount being set is zero. I also confirmed that I am never getting more than 1 stream (on similar queries).
I was referring to additionally setting PreferredMinStreamCount equal to readClientSettings.maxWorkerCount or some multiple to provide a hint to server to give more than 1 stream. I tested with setting this and could verify that server is returning more than 1 streams now. I can also raise a PR for this change.
Sample public datasets on which I ran query :

SELECT * FROM `bigquery-public-data.covid19_open_data.compatibility_view` LIMIT 10000000