Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»

■■ Using bit maps

■■ Having multileveled indexes

■■ Storing all or parts of an index in main memory

■■ Compacting the index entries when the order of the data being indexed allows such compaction

■ Creating selective indexes and range indexes

In addition to the efficient storage and scanning of the index, the subsequent access of data at the primary storage level is important. Unfortunately, there are not nearly as many options for optimizing the access of primary data as there are for the access of index data.

Compaction of Data

The very essence of success in the data warehouse environment is the ability to manage large amounts of data. Central to this goal is the ability to compact data. Of course, when data is compacted it can be stored in a minimal amount of space. In addition, when data can be stored in a small space, the access of the data is very efficient. Compaction of data is especially relevant to the data warehouse environment because data in the warehouse environment is seldom updated once inserted in the warehouse. The stability of warehouse data

minimizes the problems of space management that arise when tightly compacted data is being updated.

Another advantage is that the programmer gets the most out of a given I/O when data is stored compactly. Of course, there is always the corresponding issue of decompaction of data on access. While it is true that decompaction requires overhead, the overhead is measured in CPU resources, not I/O resources. As a rule, in the data warehouse environment, I/O resources are much more scarce than CPU resources, so decompaction of data is not a major issue.

Скачать в pdf «Building the Data Warehouse»