Too many dimensions per fact?

[chade25]

Is it possible to have too many dimensions per fact table? Right now I have 19 dimensions on one fact table, and I could add more.

[hkandpal]

Hi,

you can have as many as you can , but after 20 it is better you see if you can combine some dimension into a Junk dimn.

thanks

HImanshu

[BoxesAndLines]

It starts to increase the load times due to all the indexing normally required for FK indexes. That said, I've blown by 20 many times. :-)

[Mike Honey]

Each new dimension decreases the flexibility of your model - it is much easier to add or change attributes in an existing dimension than add a new dimension. So I tend to lean in that direction e.g. a single Customers dim with many attributes rather than Customer Type dim, Customer Location dim, etc.

Another consideration is the increasing complexity of the end user experience: whether coding star joins or using an OLAP tool you may be making life harder for your testers, report authors and users.

Balanced against that, it can be confusing when junk dimension attributes have absolutely no association with each other.

[Sponsored content]