HANA Modelling

[Former Member]

Hello ,

I have to model an analytical view based on a table which has approximatelyt 120 columns and it has to be joined to  60 text tables . I have built it and tried to test its performance. its taking several minutes to a million records. what are alternate ways of modelling? Please suggest.

Thanks,

Sri

[rindia · Accepted solution]

Hi Sri,

It is better to build an Attribute view with many text tables and then use this Attribute view with fact table in Analytic view.

Regards

Raj

[Former Member · Accepted solution]

but how would it improve the performance ?

[former_member184768 · Accepted solution]

Hi Sri,

Before improving the performance, let's identify where is the issue in performance. .

As you might be aware, the Attribute view joins to Analytic view in Logical view might perform better as compared to the all the joins in the data foundation. But this is just one point, there could be other scenarios too.

Hence it is important first to identify the performance bottleneck. Please check the visual plan for the query and find out where the issues are occurring. You can refer to the wonderful blog by Lars on Visual plan.

Regards,

Ravi

[former_member184768]

Hi Sri,

Please check the join types in the model. Check if you have Referential joins which may avoid joining the text tables when the description is not queried.

Please also look into the performance bottlenecks, check if you can have some filters defined, which can be pushed down during the query execution. Check where the query is taking most of its time. Based on the initial analysis you may find alternatives to the modeling.

Regards,

Ravi

[rindia]

Hi Sri,

If you are thinking of how to push down the filter during the query execution, then Ravi had already written an excellent blog on this and can be found here.

You will certainly get some knowledge on this.

Regards

Raj