The performance of SAP HANA calculation views is critically dependent on their design and complexity. Aggregating data as early as possible within the view's data flow is a core optimization principle. This reduces the size of intermediate result sets that need to be processed by subsequent nodes, such as joins, leading to significantly faster execution. Similarly, limiting the number of stacked calculation views (i.e., views built on top of other views) is a crucial best practice. Deeply stacked models increase compilation time and complexity for the query optimizer, which can result in less efficient execution plans.

A. Avoid union pruning: Union pruning is a performance optimization feature that excludes irrelevant data sources in a union node based on filters. Avoiding it would degrade, not improve, performance.

B. Partition large tables: While partitioning underlying tables improves query performance, it is a physical data storage strategy, not a direct modeling technique applied within the calculation view's design itself.

1. SAP HANA Cloud

SAP HANA Database Modeling Guide for SAP Web IDE for SAP HANA - Section: Performance Guidelines for Calculation Views. This guide explicitly states

"Filter and aggregate data at the earliest possible stage." It also advises

"Keep models simple. Avoid creating a deep stack of calculation views because this can lead to poor performance." This directly supports answers D and C.

2. SAP HANA SQL and System Views Reference - Section: MDEVEXPENSIVESTATEMENTS. Analyzing expensive statements often reveals issues caused by complex models with many stacked views or late aggregation

reinforcing the importance of the principles in answers C and D.

3. SAP HANA Administration Guide - Section: Managing Tables. This guide details table partitioning as a general performance technique for managing large tables at the physical layer

confirming that option B is a valid but less direct technique compared to the modeling-specific advice in C and D.