Customer Reviews

Clickstream Data Warehousing

5 star:		(5)
4 star:		(0)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

This is the first book on data warehousing and the web that actually goes into detail about web-based data. Instead of the obligatory chapter about fields in a server log and "what is a cookie", this book has several chapters of detailed explanation about web server data and the kinds of problems in it. One of the most useful sections describes the all the steps in the building of clickstream ETL. It also has a complete set of clickstream data models, although we already designed ours by the time I read the book.

The book goes into a fair amount of detail on certain topics, like server log files, and the generalized architecture of modern web sites. However, it doesn't really drill into any detail on the issues that may come up, such as cookieless browsers, ideas for cleansing dirty URLs, and methodology issues (when do you not count a pageview?).

As for the construction of the warehouse itself, it does provide some good generalized schemas that you can use for ideas, but nothing that you couldn't dream up yourself with a few days on a project.

A good overview, but no substitute for on-the-job learning.

This book is an excellent treatment of architecture and detail for constructing a clickstream data warehouse. There are useful examples and explanations of clickstream data, data warehouse design, hardware requirements, performance recommendations, and the ET&L process. The authors seem to take a no fear attitude in revealing problems and suggesting solutions.

After reading Kimball's Data Webhouse Toolkit, I was ready for a book that would drill down into the details and show how to put his ideas into practice. Sweiger's book really delivers---it doesn't avoid the really hard choices you need to make, and provides very clear directions on how to implement the techniques described.

The coverage of technical stuff like server log files and using cookies for tracking user identity is the most useful treatment of these topics I've read anywhere. But I really like the fact that the author provides a detailed, step-by-step plan for implementing a clickstream DW--with lots of examples that show he's really done this and isn't just talking from 5,000 feet up.

Mark and Mark have always been known to deliver the kind of detail practitioners demand. They don't disappoint with this book. It's very practical for the typcial implementation.

For those who are looking for "better practice" I offer recommendations that are admittedly outside of their area of expertise. The total approach -- staffing, activities, etc. -- is reflective of the 'typical' IT environment. Web deployment, however is not 'typical IT'. A single box on their team org chart, labeled 'webmaster' should be, in reality an entire collective of specialists who design and develop commercial Web sites (refer to the roles noted at [URL]). It would be roles such at these who would work to establish the requirements, with techniques not covered in this book.

Two other significant considerations not covered by their exhaustive detail are: establishing benchmarks to base metrics against and determining additional feedback loops (to be added to the Web designs) to enhance the data.