Thank you, Andy. I should have mentioned this is on DB2. I guess option #2 is my best bet.
Any insight into whether it is faster to read twice from the cube using run_rsdri_query vs. reading once, writing the results to a cluster, then reading back from the cluster?