The Matrix is a proposed alternative to the Bridge Tables.
When building virtual fact tables, we need to use a materialized basis, in order to have a good query performance. The Matrix is a materialized table containing the hash keys that handle the relations among tables: not only facts to dimensions, but also facts to facts. The Matrix is a large table, but it uses less storage space than the Bridge Tables.
We will start by presenting a brief history of the Data Warehouse. Then we will illustrate a problem that exists today in the traditional Dimensional Modeling: the redundancy. Then we will introduce the Matrix, and we will make a practical example that explains why the Matrix ensures zero redundancy and minimized storage. We will show that it’s possible to deliver to the end users the basis for a multi-fact analysis, even in the scenario of non-conformed dimensions, all based on the Matrix.
This solution has been tested in several client implementations, but not yet on massive data volumes. The open question is: will it scale?