Benchmark response times using BIA.

Question

Hi all
I ran some queries using BIA and logged their processing times. Now, I want to compare these processing times with "benchmark response times using BIA" just to make sure that these queries are performing optimally. Could you pls provide pointers/links to the same.
Regards
Niranjana

Answer 1

Following are the steps we do in order to get the benchmark data.
Assume you have Indexed a cube in BIA.
Without BIA:
1. Goto transaction RSRT2-->Select the query and click "Execute + Debug"
2. Under section "BIA Server"-->Check "Donot user BIA Index"
3. Under section "Others"--> Check "Display statistics data" and "Do not user cache"
4. Execute....After query complete click on Back button to see the statistics
5. Sum up the "Duration column" and Subtotal by column "Event Text"
6. We are interested on the item "datamanager" - note down the value
With BIA:
7. Goto transaction RSRT2-->Select the same query and click "Execute + Debug"
8. Under section "BIA Server"-->DONOT Check "Donot user BIA Index"
9. Under section "Others"--> Check "Display statistics data" and "Do not use cache"
10. Execute....After query complete click on Back button to see the statistics
11. Sum up the "Duration column" and Subtotal by column "Event Text"
12. We are interested on the item "datamanager" - note down the value
13. Click on "Aggregation layer"-->You should note the on the "aggregation" column the cube name should be suffixed with "$X" ( which indicates it uses BIA)
14. Here "viewed at" is sum of other two front columns...
In the above way you can benchmark between non-BIA vs BIA query performance. if you see even with BIA the performance is slow...then it could be bcoz of OLAP time...lot of calculation in the frontend etc...
In general we see the performance of BIA is fantastic...except sometimes the numbers are not matching between BIA and non-BIA queries...where we need to rebuild the indexes.
Hope it helps
Gopi