Not sure why you need to stress test SAS libraries that are pulling data from Oracle. Have you done any testing so far with just single queries? If not you should start with these and compare performance with other SQL query tools like Oracle's SQL*Plus. If run times are similar then you can be fairly certain SAS queries are working efficiently.
LIBNAME query performance will be primarily influenced by two things: Oracle server performance and the performance of the network pipe between Oracle and SAS servers. In my experience the performance of SAS itself is not a major factor. The main limit is often network performance, particularly when extracting large data volumes. For example if your Oracle and SAS servers are not in the same data centre and you are pulling data over long distances this can really slow down performance.
Once you've confirmed single queries are working efficiently, then I'd suggest you set them up as SAS batch jobs. As long as you pre-assign your SAS Oracle LIBNAMES (I'm assuming here you have defined them in SAS metadata) then they will work fine in batch. It is very easy to run multiple SAS batch jobs simultaneously and this will mimic the performance of multiple simultaneous EG sessions running the same queries, plus they are a lot easier to set up.
