1

I don't know if it's a simple question but does a data warehouse have different rules when it comes to normalization forms?

At the moment I am creating some tables in a data warehouse, and they simply do not fulfill the normalization forms rule. I have tried to find a method to fulfill them, but it seems to be pointless.

2

Normalization is not the holy grail. Normalize as much as possible. In a data warehouse you can group some data to get better performance. Also a data warehouse is about querying data. Less on users maintaining the data.

Go for a design that will give the information in an easy and quick way. Normalize if you can but in my opinion this is not the first goal.

1
  • Thanks for the answer, yes i have tried to normalize as much as i can, but i reached a point where it was not possible or pointless to do anylonger. Nov 26 '15 at 15:58
3

Actually it is common to denormalise in a data warehouse.

For instance I will have the full address of customers with the state and the country as columns in my Customer dimension. The data will be duplicated a lot for the customers living in the same city, but I won't have to do two costly joins to retrieve the state and then the country from two other tables.

Most of Data Warehouse are modelled as Star Schemas, where the fact table (the event you want to measure) is directly joined to dimensions (e.g. Time, Customer, Product, ...). It can be a constellation if you have multiple fact tables sharing the same conformed dimensions. You can also have some kind of normalisation where for instance the State and the Country would be in another dimension joined to the Customer dimension, but some reporting tools behave better with a pure Star Schema.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.