Threaded Reads provide a way to load data in parallel from Google BigQuery by using multiple connections to your data source. You can use the following options to implement threaded Reads:
Use the DBSLICEPARM= options to specify how to manage threaded Reads and optionally, the maximum number of threads. Set DBSLICEPARM=THREADED_APPS or DBSLICEPARM=ALL to implement threaded reads. The DBSLICEPARM= data set option overrides any other DBSLICEPARM= option. The DBSLICEPARM= LIBNAME option overrides any value that you set for the system option.
For Google BigQuery, when you use one of the DBSLICEPARM= options, you must also set READ_MODE=STORAGE, which specifies that SAS/ACCESS uses the Storage API for Google to transfer data. Threaded Reads can be implemented only via the Storage API for Google. The Storage API moves data from Google to SAS by using multiple streams. Each stream contains approximately equal amounts of data.
Optionally, you can explicitly control how a table is partitioned by using the DBSLICE= data set option. Use this option to provide expressions that divide a table into separate blocks of data. When you use this option, make sure that partitions do not overlap, and make sure that the entire data set is covered by the expressions that you provide, including any NULL values. Also, if you provide a value for the maximum number of threads, then ensure that you do not provide more expressions than threads.