Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»

Beyond indexing, update, and basic data management at the physical block level are some other very basic differences between the data management capabilities and philosophies of general-purpose transaction processing DBMSs and data warehouse-specific DBMSs. Perhaps the most basic difference is the ability to physically organize data in an optimal fashion for different kinds of access. A general-purpose DBMS typically physically organizes data for optimal transaction access and manipulation. Organizing in this fashion allows many different types of data to be gathered according to a common key and efficiently accessed in one or two I/Os. Data that is optimal for informational access usually has a very different physical profile. Data that is optimal for informational access is organized so that many different occurrences of the same type of data can be accessed efficiently in one or two physical I/Os.

Data can be physically optimized for transaction access or DSS access, but not both at the same time. A general-purpose, transaction-based DBMS allows data to be optimized for transaction access, and a data warehouse-specific DBMS allows data to be physically optimized for DSS access and analysis.

Changing DBMS Technology

An interesting consideration of the information warehouse is changing the DBMS technology after the warehouse has already been populated. Such a change may be in order for several reasons:

■■ DBMS technologies may be available today that simply were not an option when the data warehouse was first populated.

■■ The size of the warehouse has grown to the point that a new technological approach is mandated.

■■ Use of the warehouse has escalated and changed to the point that the current warehouse DBMS technology is not adequate.

Скачать в pdf «Building the Data Warehouse»