HANA experts,
I wanted to pose a question that our developers are pondering here. In summary, does the number of fields defined on a given Analytical View affect overall performance?
Here's a hypothetical scenario to illustrate:
Given a existing columnar HANA table T existing with 100 fields and loaded with 100,000,000 rows of data.
1. Create an analytical view AV1 over the above table T but w/5 high level fields defined (company code, cost center, year, quantity, amount). Create calculation view CV1 over AV1 w/5 fields and 5 calculated measures. In total CV1 has 10 fields total now.
2, Create an analytical view AV2 over entire table T w/100 fields defined. Create calculation view CV2 over AV2 w/100 fields and 5 calculated measures defined the same as the ones in CV1. In total CV2 has 105 fields total now.
Now the question is if query Q1 is run over CV1 with those 10 fields defined and a query Q2 run over CV2 with equivalent same 10 fields as CV1, will the performance of Q1 = Q2, even though CV2 has a much wider definition?
Keep in mind both CVs are ultimately pulling from same base table with 100 fields acros and 100,000,000 rows of data.
From the HANA literature out there, I'm led to believe that above scenario would result in the same performance of Q1 and Q2 due to "pruning/object stripping". However, as sometimes we've seen, theory doesn't necessarily match reality so I wanted to gauge your opinions.
If it turns out Q1=Q2, the thought is to go ahead and define it as AV2 and CV2 to give the most reporting flexibility over the Calculation View.
Best Regards,
Bruce