Data Warehouse Fundamentals

Most people learn data warehousing on the job, in whatever order the job happens to demand. This course puts the order back: six lessons that build from what a warehouse is to where it runs, so the vocabulary, the modeling decisions, and the loading patterns land as one connected picture instead of a pile of terms.

By the end you can define a data warehouse by the three properties that distinguish it from every other database in the building, and apply a threshold test for when building one is worth the effort. You can explain why analytical and operational workloads pull schema design in opposite directions, and why the same tables cannot serve both well. You can place the conceptual, logical, and physical levels of a data model, explain what normalization optimizes for and why analytics deliberately relaxes it, and describe the round-trip by which a model becomes a database and an existing database becomes a model. You can name the major modeling approaches for a warehouse, say what each optimizes for, and choose sensibly between them at the level a project kickoff demands. You can explain how data gets in and stays current: ETL versus ELT, incremental loads, change data capture, load ordering. And you can reason about where a warehouse runs, and, just as usefully, where a warehouse is the wrong tool.

Working engineers and analysts who are new to warehousing, or self-taught and conscious of the gaps. If you have built application databases but never an analytical one, or you have inherited a warehouse and want the reasoning behind its shape, the course meets you where you are. It assumes you know what a table and a SQL query are; it assumes nothing about warehousing itself.

It is not a platform tutorial. If you want click-by-click instruction for one vendor's product, this is the wrong course: the lessons teach techniques, and the techniques apply across tools. It is also not the deep dimensional treatment. Star schemas, grain, surrogate keys, and slowly changing dimensions are named here and taught properly in the next course, which assumes this one.

Six lessons, about an hour of video in total, each paired with a written note that stands alone as a tutorial. The notes are free to read on the site. The video lessons arrive by email as they ship, in order, starting with the first. There is nothing to install and no dataset to download; the walkthroughs are demonstrations, not exercises. Watch a lesson on a commute, read the note at your desk.

The course is in production now. Lessons publish as they are recorded, and the curriculum is the commitment, not an aspiration.

The written reference covers everything the course introduces, in more depth. Start with these.