‎2024 Aug 28 3:23 AM - edited ‎2024 Aug 28 12:35 PM
Dear Experts,
I am facing some performance issues with datasphere modeling. Here is my design structures:
Background: We have to create SAC report based on COPA Actual/Plan Data and some other External Data as well.
Modeling Steps refer to above:
1. Preload COPA Actual Data(20 M Records), Budget Data(1 M Records), External Data(10 K Records) to local Table and store table data in memory. Then create views based on local tables and make projections to reduce the no. of columns. Filters and Aggregations are also applied to reduce the total no. of records.
2. Use Input parameters to limit amount of data access.
3. Create UNION for Actual and Plan data
4. Create UNION for External data and above(3)
5. Aggregation based on final data structure
6. Various Calculated column(simple calculation) created
7. Fact source view created(Measures: 110 / Attribute: 45) and used by Analytic Model. Till this point, no. of records have reduced to 2M
SAC then use this Analytic Model for Reporting
Now here comes the trouble:
1. When I Preview data in the Analytic Model, it takes 30~40 seconds to show the result, the acceptable response time is within 10 secs.
2. While the model is consumed by SAC, as we create more calculations(YTD, Last year YTD...etc) SAC crashes.
Is there any suggestion for performance improvement ?
Thanks a lot
Request clarification before answering.
Hello Charlielin,
i would recommend you to have a look at his Blog https://community.sap.com/t5/technology-blogs-by-members/performance-analysis-in-the-sap-datasphere/...
After Reading this Blog you should be familiar on how to analyse the Performance on each step done by the SAP HANA Database thats underneath the SAP Datasphere.
Best Regards
Julian
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
| User | Count |
|---|---|
| 9 | |
| 7 | |
| 6 | |
| 4 | |
| 4 | |
| 2 | |
| 2 | |
| 2 | |
| 2 | |
| 2 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.