I came across this great paper:"What You Need to Consider When Building and Deploying an OLAP Application" ; by a SAS guy named Greg Henderson. Although this paper talks about PROC MDDB- I did not notice until the end- it seems to apply to PROC OLAP. I liked his way of how to handle semi-additive measures with a HOLAP on the 'Special Aggregation Rules' section. This seems to be a HOLAP in PROC OLAP. Could "manually aggregate those subtables
using the appropriate business rules in SAS Data Step or SQL
code" be accomplished by having a detail(or star schema) as input and then affixing those special aggregated tables to the cube? Could this happen with PROC OLAP? Could I apply everything on this paper to PROC OLAP?
I never found a reason to use ROLAP(or HOLAP) in a cube ( I would just use an information map), but to solve semi-additives this seems like a great option.