Author: Randall Root and Caryn Mason
Aimed at: database admins and developers working on BI projects
Reviewed by: Kay Ewbank
You can put forward a strong argument that Business Intelligence is the only real point of having a database.
Why keep data unless you're going to use it to make your business more effective, or in some other way to use the data to provide insights that can't be seen by 'just looking' at the records? The problem in the past has been that BI products are large, cumbersome, expensive, and hard to use. Whether or not you like SQL Server as a database, there's no denying that Microsoft has, over recent versions, put together some BI tools that don't cost a fortune, and that in theory are simple enough for anyone to use. Root and Mason's book assumes no prior knowledge of BI or the BI elements of SQL Server. It isn't aimed at programmers in particular, just at database techies who need to understand BI.
The book kicks off with a general introduction to BI solutions, followed by a '10,000 foot view of a BI project'. This chapter would be an excellent one to give to anyone you're working with who has no idea what a BI project involves, as it lays out in fairly understandable terms just what is needed. The next chapter looks at how to plan a BI project, with questions you should ask such as 'why do we need it?', 'what are we building?', 'when will we need it?', and 'how will we finish it?' In my experience, most of these would, if honest, be answered with 'no idea', but having some clue might prevent some of the disasters that masquerade as BI projects. The chapter also covers concepts like locating the data and estimating the cost.
Having laid out the high level requirements, Root and Mason go on to some of the basic necessities, starting with a chapter on designing a data warehouse. This discusses star and snowflake structures, dimensional patterns, and surrogate keys. The next chapter looks at creating the warehouse using SQL Server Management Studio, and this is followed by a chapter on ETL - Extract, Transform and Load - the processes involved in taking data and placing it in the data warehouse in a usable format. The practicalities of ETL are covered in the next two chapters, looking at how to use SSIS (SQL Server Integration Services) to work over the data.
Once the data warehouse is created, the next stage of a BI project in SQL Server is to use SSAS to analyze the data.SQL Server Analysis Services lets you create data cubes made up of aggregate data that you can run reports on. Root and Mason have good in-depth descriptions of the concepts of setting up data cubes, and how to design and configure the dimensions. There's also a chapter on additional cube and dimension configurations that looks at the 'extras' - the other tabs you get when crating cubes and dimensions where you can specify actions, partitions, aggregations, perspectives and translations.
There are just two chapters on creating reports, one showing how to use SQL queries, the other MDX queries. I'd have liked more coverage of this area, but I can see how the space requirements make that difficult. There's a chapter looking at creating reports in Excel, and two chapters look at creating and configuring reports with SSRS - SQL Server Reporting Services.
Overall, I liked this book. It gives a good balanced view of what BI is all about, and how to go from a mass of data to a system that you can run queries and reports against. If you don't really understand SQL Server's BI facilities, it's a good read, and it contains enough detail and descriptions to enable you to create a fully featured BI system.