The Microsoft® Data Warehouse Toolkit: With SQL Server™ 2005 and the Microsoft® Business Intelligence Toolset

10.2. SQL Server Data Mining Architecture Overview

Microsoft SQL Server 2005 Data Mining offers a rich, well-tuned, integrated, and easy-to-use data mining environment. For those of you who worked with data mining in SQL Server 2000, the 2005 release is a great leap forward. In this section we give an overview of the data mining environment using the high-level architecture drawing presented in Figure 10.1 as a guide.

From a system point of view, integrating data mining into the overall product allows the data mining service to take advantage of the functionality offered by the rest of the system. For example, point A in Figure 10.1 shows how data mining models are built using the Analysis Services dimensional engine, leveraging its ability to load data and quickly perform the base statistical calculations like sums, averages, and counts. The data mining server can also easily pull case data from both relational and Analysis Services databases as seen at point B in Figure 10.1.

Point C in Figure 10.1 shows how the typical developer will first experience data mining by creating a BI Studio Analysis Services project and then using the Data Mining Wizard to create a new data mining structure and an initial data mining model. The mining structure is a new construct that provides a metadata layer allowing several mining models to work with the same input data. Each mining model in a mining structure can have different algorithms and parameters. The wizard provides model building guidance ...