Show TOC

Performance Optimization in Data WarehouseLocate this document in the navigation structure


The performance of the data warehouse processes depends on various factors. Good system performance begins with the design of the data model. A Layered Scalable Architecture (LSA) is the foundation of a performance-optimized data warehouse. Therefore you need to develop a strategy for configuring your data model according to concrete requirements and for extensively planning and implementing your model according to the LSA guidelines. This allows you to prevent architecture-related bottle necks later on. Besides the clearly devised layer and data architecture, data logistics also plays a role here (for example, preventing processes, such as the activation of DataStore object data, from locking each other and causing delays). In addition, you need to note the relevant database specifications when configuring the system. Tactical possibilities are available for improving the runtime of your operational system. Usually performance only becomes an issue after a your BW system has been operating for a certain time (once the data volume has increased significantly). You can improve performance by noting the tips explained here.

The following section explains the individual measures for improving performance in the data warehouse. The documentation covers the topics of extraction, loading processes and further processing processes in the data warehouse with regards to performance. It also describes aspects of data retention, load distribution and possibilities for runtime analysis. We have only concentrated on the possibilities for the BW system, designing processes and objects as well as improving performance at runtime. We did not cover architectural aspects, data logistics and database specifications. The tips are aimed at experienced system administrators or consultants, who can use the tips as a checklist for modeling or for the administration of the operational system.