Prior to describing the issue I will outline what I have at hand:
0. HANA SPS03 SP21
1. T_ITEMS -- Sales Order Items, 4.5 Mio records
2. ANV_ITEMS -- Analytical View created on top of T_ITEMS, using some Master Data for Profit Center, Plant etc
3. T_COND -- Sales Order conditions, 25 Mio records. Contains Condition Type, Ship-to and Amount fields, which I will use below.
4. ANV_COND -- Analytical View on top of ANV_COND, using some Master Data for e.g. Condition Type
5. CLV_REP -- graphical Calculation View created as a Join of ANV_ITEMS and ANV_COND using Order Number and Order Item.
So my steps:
A. In HANA Studio, I run a SELECT statement against the ANV_COND, selecting Ship-to, Condition Type and Amount fields -- it takes 300 msec. Fine.
B. I modified that SELECT statement to run against the CLV_REP --all of a sudden it takes 2 minutes!
C. I modified the CLV_REP, so instead of ANV_COND Joined to ANV_ITEMS it would be ANV_COND joined to T_ITEMS -- the runtime drops to 12 seconds.
OK, question time.
Q1: Does SAP recommend against joining two Analytical Views in a Calc View? I didn't see such a recommendation, but maybe I missed it while reading the manuals.
Q2: Why at all the Join had an impact on performance, since in all three scenarios all the 3 fields used reside in the same base table?