Anchor Modeling

[Lars Rönnbäck]

We have recently introduced the concept of anchor modeling, which is a modeling technique that yields maximum flexibility. It's been proved over the last four years at our customers to work well in a data warehousing environment, where surrounding systems often change. Until now, we have only been using this technique on SQL Server, since its optimizer is intelligent enough to filter out tables that are not needed even if they are left joined into the view used in the query. I would like to use the same technique on MySQL, but am not sure if the optimizer does the same kind of optimization (Oracle won't). Since you can order the physical table data on the storage media similar to a clustered index in SQL Server the performance when using "latest views" should be good.

Does anyone know if the optimizer will remove tables from the execution plan that it won't need, even if they are joined into the view you are using?

If you want to read up on anchor modeling, please download the fact sheet, and the presentation from the recent TDWI European conference from our homepage: http://www.anchormodeling.com
Regards,
Lars