"Big data" started as a buzzword about size. Fifteen years later its real legacy is architectural: the way organisations store, process and use data has been rebuilt at least three times. This article traces that transformation, from the enterprise data warehouse through Hadoop data lakes and cloud warehouses to today's lakehouses and real-time, AI-ready platforms, and explains what each shift solved and what it broke. 1990s–2000s Enterprise DW ETL, star schemas, appliances ~2006–2015 Hadoop data lake HDFS, MapReduce, Hive, cheap storage ~2012– Cloud warehouse storage separated from compute ~2019– Lakehouse open table formats on object storage 2020s Real-time + AI streaming, semantic layers, LLMs Constant through every era: model the business, test the data, govern access. The tools changed; the discipline did not.
Practical tutorials and architecture guides for data engineering, analytics, BI and AI.