Loading the Data Warehouse Across Various Parallel Architectures
Vijay V. Raghavan · Very Large Data Bases · 1996
Loading data is one of the most critical operations in any data warehouse, yet it is also the most neglected by the database vendors. Data must be loaded into a warehouse in a fixed batch window, typically overnight. During this period, we need to take maximum advantage of the machine resources to load data as efficiently as possible. A data warehouse can be on line for up to 20 hours of a day, which can leave only a window of 4 hours to complete the load. The Red Brick loader can validate, load and index at up to 12GB of data per hour on an SMP system.