Book Image

Expert Cube Development with Microsoft SQL Server 2008 Analysis Services

Book Image

Expert Cube Development with Microsoft SQL Server 2008 Analysis Services

Overview of this book

Microsoft's SQL Server Analysis Services 2008 is an OLAP server that allows users to analyze business data quickly and easily. However, designing cubes in Analysis Services can be a complex task: it's all too easy to make mistakes early on in development that lead to serious problems when the cube is in production. Learning the best practices for cube design before you start your project will help you avoid these problems and ensure that your project is a success. This book offers practical advice on how to go about designing and building fast, scalable, and maintainable cubes that will meet your users' requirements and help make your Business Intelligence project a success. This book gives readers insight into the best practices for designing and building Microsoft Analysis Services 2008 cubes. It also provides details about server architecture, performance tuning, security, and administration of an Analysis Services solution. In this book, you will learn how to design and implement Analysis Services cubes. Starting from designing a data mart for Analysis Services, through the creation of dimensions and measure groups, to putting the cube into production, we'll explore the whole of the development lifecycle. This book is an invaluable guide for anyone who is planning to use Microsoft Analysis Services 2008 in a Business Intelligence project.
Table of Contents (17 chapters)
Expert Cube Development with Microsoft SQL Server 2008 Analysis Services
Credits
About the Authors
About the Reviewers
Preface
Index

Summary


In this chapter, we've discussed how to deal with some common problems we might face when designing our Analysis Services dimensions:

  • Grouping and banding members and numeric values

  • Slowly Changing Dimensions

  • Junk Dimensions

  • Ragged hierarchies

We've found that we can create new attributes on a dimension to group members and create bandings for numeric values using some simple SQL. This can have a positive impact on the user's experience. We've also found that with a bit of thought, the three different types of SCD can all be modeled satisfactorily. With ragged hierarchies we've learnt that the parent/child hierarchies are very flexible but have several problems, especially with query performance, and should be avoided where possible. We have also seen that using a regular user hierarchy and the HideMemberIf property is a better option, although not without its own limitations.