I have created some queries on a custom cube with about 8M rows of data. I have compressed and partitioned the cube ()CALMONTH) and built a relavent aggregate that is being picked up by the query. Also, it seems like the OLAP cache is working properly.
However, the run times are still quite long (over 65 sec) for queries selecting larger data sets (say 2 months of data). When I watch the query running in sm50, I notice it spends most of its time running report CL_RSDM_READ_MASTER_DATA. In my cube, I have made several attributes on one characteristic (Account#) navigational, and some of the query filtering is done on these nav attributes. Also, the report displays several of the attributes (nav and display) from this characteristic - I am not sure if that is causing the long reads from the master data P tables? I have played around with creating indexes on the X table for some of these attributes but that seems to make no difference either.
Any clues? Am I missing something simple?