Question:
Hi,
Had a question...for SEM-BPS, transactional cube is used for planning. All the queries are based on this transactional cube only. Some queries run really slow because of the complex structures involved in rows section (that is strcutures with selections restricted to hierarchy nodes or range of hierarchy nodes). Will it be a good idea to push all the data to a basic cube and create complex reports on it or is there another alternative?
Usually in an SEM environment, do they just create queries on transactional cubes only? What is the norm?
Thanks!
Answer:
Hi,
U can datamart the data to another basic infocube and generate the report the from there.
if the report is slow use the T/C: Rsrt and analyse the reasons.
May be your report is trying to whole lot of stuff, which are not required. Analyse the performance of the report thru stats cube.
Regards,
Ram.
Answer:
Performance of the query depends very little on whether the cube(s) where the data is retreived from is "Basic" or "Transaction" although "Basic" cubes are optimized for reading and "Transaction" cubes are optimized for writing data. Performance with the queries has to do with data model, data volume and query complexity. In my opinion, pushing data to a basic cube alone would not improve performance of the query.
Please refer to OSS Notes that address query performance and possible solutions.
There is no norm with regards to query creation on transactional cubes used in SEM. It is driven by reporting requirements. For example, if the budget/forecast data is stored in transaction cube and actual data is stored in basic cube, then one would create a query on a multicube if the reporting requires comparison of actual data to budget/forecast.
Regards
Narasimha Devisetty