Efficiency on the way out of the Data Vault: Speeding through the Mart – from Requirement to Delivery
Regardless of the format that a front-end tool requires, a report will always
Event Details
Efficiency on the way out of the Data Vault: Speeding through the Mart – from Requirement to Delivery
Regardless of the format that a front-end tool requires, a report will always have Dimensions and KPI.
Usually, one does build a dimensional model on top of the data vault. And done in the right way, it
can be done quickly.
In reality we often encounter redundancy in the calculation of KPI, performance trouble, long and
strange discussions about how much change in the dimensions will be displayed (slowly changing
dimensions 0 through 7) and business users who will answer any ‘this or that’ question with ‘Both!’.
The level of automation on the data mart is often quite low.
In 2015 Kimball did complain that the dimensional model as last layer of a relational Inmon-style
warehouse is not properly described. At least not in a way he used to describe his dimensional
approach. And he is right, if we look at the differences that a relational model with a full history of
data as a base for dimensional model brings to the table – we can make the data mart way more
efficient.
In this talk we will look at these differences and find ways to improve the gathering of requirements,
automatically avoid redundancy in calculations and find new patterns to automate the Data Mart.
We will use these findings to create a growth path, where we start easy with a Raw Mart and add
concepts on the way to the most complex Information Marts. Gain efficiency (by method).
