Project Real Analytics Services technology Discussion (1)

This white paper provides a technical discussion on Analysis Services design and best practices in Project REAL. It discusses in depth the details of each type of object, such as data sources, data source views, dimensions, hierarchies, attributes, measure groups, split tables, and so on. It also points out how to create a SQL Server 2005 service package that automatically creates a measurement group partition table on the basis of a relational database partition table.

About Project REAL

Project Real is Microsoft's effort to provide best practices for creating business intelligence applications. These programs are built on the basis of Microsoft®sql server™2005 and implemented on a real customer background. This means that real customer data can be placed inside the system and can respond to the same problems that customers will encounter in the development process. These issues include:

Pattern Design-relational mode and analytic service mode

Implementation of data extraction, data transformation, data loading (ETL) process

Design and development of client system, including data report and interactive analysis

Classification of product Systems

Management and maintenance of operational systems, including continuous updating of data

Through our work experience in this real-world deployment environment, we get a complete understanding of how to use these tools. Our goal is to focus on all the issues that large companies encounter in their own actual deployment.

This white paper provides a technical discussion on Analysis Services design and best practices in Project REAL. We discussed in depth the details of each type of object, such as data sources, data source views, dimensions, hierarchies, attributes, measure groups, split tables, and so on. It also points out the important problems that we encounter in the process of advancing.

To view an overview of Rroject real, view the Project real:technical Overview white paper. A significant portion of the data, tools, and examples are generated in the life cycle of Project REAL. To find the latest information, you can go to Project REAL Web site This connection to see the relevant information (http://www.microsoft.com/sql/bi/ProjectReal/).

Note: This article is only a draft, and it includes a number of constructive practices that are based on our earlier experience with the Community Technology Preview (CTP) work in SQL Server 2005. Before the release of the product, the descriptions in the white paper are accurate. The functionality of the product described in the document may vary. In the future, a better practice scenario may be offered. SQL Server 2005 is the development tool that we use in these good practice routines.

Introduction

This article reviews the technical design of Project REAL Analytics services and discusses various impact design issues. We assume that the reader is already familiar with the design of the analysis service and has practiced the pattern used by Project REAL. For example, we assume that the reader already knows the existence of a multiple-vendor dimension. Our discussion focuses on why it exists (and the alternatives we consider before finalizing the design).

In this paper, we examine the various types of analysis service objects that are applied in multidimensional design. Start with physical mode objects, such as data sources and data source views. Next we'll talk about logical objects, such as dimensions, user-defined hierarchical relationships, attribute hierarchies, and measure groups, and so on. Next, drill down to the metric group features, such as Split, collection (aggregate) design, and the proactive cache (proactive caching). This section finally discusses other logical designs, including calculations, key performance indicators (KPIS), activities, perspectives, custom assemblies, user-defined functions (UDFS), and MDX scripts.

In the last chapter, we discuss in detail the two optional and reasonable design scenarios in the design phase of the analysis service pattern. We provide the goals, and we think about what we are going to do, and that is what we have achieved.

This article ends with an introduction to the service-side settings, which mainly discusses why we need to change these configurations.

Project Real Design relies heavily on segmentation (partitioning), which defines hundreds of such split tables in all measurement groups, and in Appendix A we show how we solve the management problems that we create and manage partitioned tables in various databases.