Kimball University: Keep to the Grain in Dimensional Modeling - InformationWeek

InformationWeek is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

IoT
IoT
Software // Information Management

Kimball University: Keep to the Grain in Dimensional Modeling

When developing fact tables, aggregated data is NOT the place to start. To avoid "mixed granularity" woes including bad and overlapping data, stick to rich, expressive, atomic-level data that's closely connected to the original source and collection process.

Ralph Kimball Ralph Kimball
The power of a dimensional model comes from a careful adherence to "the grain." A clear definition of the grain of a fact table makes the logical and physical design possible; a muddled or imprecise definition of the grain poses a threat to all aspects of the design, from the ETL processes that fetch the data all the way to the reports that try to use the data.

What, exactly, is the grain? The grain of a fact table is the business definition of the measurement event that creates a fact record. The grain is exclusively determined by the physical realities of the source of the data (see related article, "Declaring the Grain").

All grain definitions should start at the lowest, most atomic grain and should describe the physical process that collects the data. Thus, in our dimensional modeling classes, when we start with the familiar example of retail sales, I ask the students "what is the grain?" After listening to a number of careful replies listing various retail dimensions such as product, customer, store and time, I stop and ask the students to visualize the physical process. The salesperson or checkout clerk scans the retail item and the register goes "BEEP." The grain of the fact table is BEEP!

Once the grain of the fact table is established with such clarity, the next steps of the design process can proceed smoothly. Continuing with our retail example, we can immediately include or exclude possible dimensions from the logical design of the retail fact table. Benefiting from the very atomic definition (BEEP), we can propose a large number of dimensions including date, time, customer, product, employee (perhaps both checkout clerk and supervisor), store, cash register, sales promotion, competitive factor, method of payment, regional demographics, the weather, and possibly others. Our humble little BEEP turns into a powerful measurement event with more than a dozen dimensions!

Of course, in a given practical application, the design team may not have all these dimensions available. But the power of keeping to the grain arises from the clarity that the grain definition supplies to the design process. Once the logical design has been proposed, the design team can systematically investigate whether the data sources are rich enough to "decorate" the BEEP event with all these dimensions in the physical implementation.

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Previous
1 of 2
Next
Comment  | 
Print  | 
More Insights
InformationWeek Is Getting an Upgrade!

Find out more about our plans to improve the look, functionality, and performance of the InformationWeek site in the coming months.

Slideshows
Blockchain Gets Real Across Industries
Lisa Morgan, Freelance Writer,  7/22/2021
Commentary
Seeking a Competitive Edge vs. Chasing Savings in the Cloud
Joao-Pierre S. Ruth, Senior Writer,  7/19/2021
News
How CIO Roles Will Change: The Future of Work
Jessica Davis, Senior Editor, Enterprise Apps,  7/1/2021
White Papers
Register for InformationWeek Newsletters
Video
Current Issue
Monitoring Critical Cloud Workloads Report
In this report, our experts will discuss how to advance your ability to monitor critical workloads as they move about the various cloud platforms in your company.
Slideshows
Flash Poll