Data marts are the business user interface of your data warehouse. I have tried to explain how to design an enterprise data warehouse in my first article. This article will give some best practice tips for designing useful data marts for your business teams. The question was in my previous article:

"What kind of data marts do you need? "

Your business teams’ needs can differ from each other. They may need their own data format where, for example, the finance department has requirements other than the sales or marketing department.

Security requirements must also take into consideration. Different departments should not reach or see others’ data. You may not want to prefer other departments to see all the financial data on your systems.

On the other hand, there will exist shared data between departments, to keep the single point of truth, the data mart design should also have some common structures on your system.

Data marts are focused on mostly fast reading of the data. The data should sometimes be kept in summary formats. So your data marts may contain some aggregated structures or calculated metrics based on your business team need.

Keeping history is very important for your trend analysis and your past situation of your company. However, keeping all the transaction data may cost you a lot. So you can create some end-of-month or maybe end-of-week structures in your data marts.

So we can consolidate the questions above as the following lines:

• What do business teams need? Do they have different requirements and need different formats of data?
• Is security of data a concern for your company?
• Does the data mart have common structures like dimensions or facts?
• Do business teams need aggregated data?
• Do business teams need new calculated metrics?
• What kind of historical data is needed?
• Should there be a life cycle of a data mart?

#big data #data analysis #data warehouse #data warehouse architecture #data warehouse concepts #data mart