Research of Real-time Data Warehouse Storage Strategy Based on Multi-level Caches

Abstract

Real-time data warehouse extend the application of traditional data warehouse. It can not only support tactical queries for enterprise but also provide much variable tactical decision support effectively. For these reasons, it is very meaningful to research on the structure of real-time data warehouses. This paper introduced the background of real-time data warehouse and proposed the strategy of real-time data warehouse which is based on double mirror replication mechanism. The strategy is composed of double steps. First we used double mirror replication mechanism to enable continuous loading data in the real-time data warehouse with minimum impact in query execution time. Second we proposed incorporating multi-level caches into the data warehouse structure which is based on real-time partition and gave the process of design and implementation with details. We differentiated between queries with various data freshness requirements, and used multi-level caches to satisfy these different requirements.