Demystifying OLAP Data Cubes

A Brief But Thorough Overview of OLAP Data Cubes and Their Place in The Modern Data Stack

Photo by Milad Fakurian on Unsplash

You may have Googled what data cubes are and given up trying to understand shortly thereafter. Typically, the abandonment happens on the Wikipedia page, paragraph two.

You’re not alone.

Data cubes are not a straightforward topic in data engineering. Instead, it’s a classic case of asking for a definition five times and receiving five different versions. The ambiguity owes itself to the namesake and unfortunate universal metaphor used to describe data cubes: a diagram of a 3D cube.

Nearly every article covering data cubes starts with this representation and most likely ends there, too. The explanations don’t provide examples of how the concept of a data cube is implemented in a data pipeline.

So let’s see if we can fill the gaps and demystify this topic.

What Are Data Cubes?

Generally speaking, a data cube is a design pattern where a measure (like sales) is aggregated across multiple dimensions (like region, store, and product).

The design pattern is implemented primarily in two contexts: 1) as a pre-aggregated table in a relational database, and 2) as a data…