Bulk loading to a Google BigQuery table might result in poor performance when using SAS/ACCESS® Interface to Google BigQuery


If you specify BULKLOAD=YES as a LIBNAME or data set option when loading data into a Google BigQuery table, performance issues might result, including the possibility of slower performance than when loading without the BULKLOAD=YES option.

This issue occurs because a BULKLOAD operation is currently occurring for every block of data. The frequency with which the BULKLOAD operation occurs is determined by either of the following:

This issue does not occur with the FEDSQL procedure or with the use of the SAS® Data Connector for Google BigQuery in SAS® Viya®.

The workaround is to not specify the BULKLOAD=YES option and instead use the INSERTBUFF= option. Select a value up to 32,767 based on the memory resources that you have available on your system.