"A Trio of Interesting Snowflakes" article

[AlanB]

Does anyone know where to find a copy of Kimball's "A Trio of Interesting Snowflakes" article from Intelligent Enterprise Magazine, June 2001?

The link in Articles and Papers only shows part of the article. http://www.kimballgroup.com/html/articles_search/articles2001/0106IE.html?TrkID=IE200106

Links to Intelligent Enterprise get redirected to Information Week which no longer has the article, at least not if you search on Kimball & snowflake. Even Google failed me!

[ngalemmo]

Its a shame they seemed to have lost the archive... maybe they should change their name to Information Weak.

[AlanB]

Its a shame they seemed to have lost the archive... maybe they should change their name to Information Weak.

Nice touch! Maybe someone should talk to them about DLM.

An email's been sent to the InformationWeek.com Sr Director of User Experience. Let's see where that leads.

[Nancy Rinn]

Thanks for letting us know about the missing text! We now have the complete article available online: http://kimballgroup.com/html/articles_search/articles2001/0106IE.html?TrkID=IE200106

Nancy

[AlanB]

Thanks Nancy.

Having a large, complex customer dimension, I thought I'd better read up on this.

[hang]

Nice article, but still missing some relevant schema diagrams.

I am working on a BI system managing national job seekers with 20 million records in the dimension. The original design has denormalised all the lengthy text attributes, as well as low cardinality attributes, into a monster dimension. You can imagine about the sheer volume of redundancies in the dimension making queries really slow and inefficient.

People can easily get carried away by the guidelines in dimensional modelling, that is normalised facts and denormalised dimensions. However in some cases, eg. monster dimensions, if the guidelines are misused as dogmatic rules, the consequence is sever. I don’t like people setting rigid rules in any design. If there is a rule for dimension denormalisation in dimensional modeling, I strongly suggest the monster dimension be excepted.

[Sponsored content]