Warehousing , Olap , and Data

Rapidly improving computing and networking technology enables enterprises to collect data from virtually all its business units. The main challenge today is to extract useful information from an overwhelmingly large amount of raw data. To support complex analysis queries, data warehouses were introduced. They manage data, which is extracted from the different operational databases and from external data sources, and they are optimized for fast query processing. For modern data warehouses, it is common to manage Terabytes of data. According to a recent survey by the Winter Corporation (2003), for instance, the decision support database of SBC reached a size of almost 25 Terabytes, up from 10.5 Terabytes in 2001 (Winter Corporation, 2001). Human analysts cannot “digest” such large amounts of information at a detailed level. Instead they rely on the system to provide a summarized and task-specific view of selected data. Consequently, efficient summarization and aggregation of large amounts of data play a crucial role in the analysis process. The goal of Online Analytical Processing (OLAP) is to support this style of analysis at interactive response times for massive data collections. OLAP applications often maintain aggregate information in data cubes. Pre-computed aggregate values that speed up query execution for group-bys, cross-tabs, and subtotals can be easily included in the model (e.g., CUBE operator, which is discussed later).