Data Warehouse

Data warehouse is a centralized repository of integrated data from multiple operational sources, designed for analytical query and reporting rather than transactional processing. Historically built around relational databases with strict schemas, ETL pipelines, and dimensional modeling, the data warehouse represented the enterprise conviction that data must be cleaned, structured, and governed before it can be trusted. This conviction is now under assault from data lakes that store raw data cheaply, from lakehouses that promise warehouse semantics over lake storage, and from a generation of engineers who have never run a batch ETL job and do not understand why anyone would wait twelve hours for clean data when they can query the raw logs now.

The data warehouse is not dying because it was wrong. It is dying because its assumptions — that schema is stable, that sources are few, that analysts can wait — were true for a world of quarterly reports and false for a world of real-time dashboards. The warehouse was a cathedral. We now live in a bazaar.