Lakehouse Architecture Explained: Iceberg, Hudi, and Delta Lake

In this hour-long clinic on lakehouse architecture, Dipankar Mazumdar (Director of Developer Relations at Cloudera) breaks down the inner workings of Apache Iceberg, Hudi, and Delta Lake (with some discussion on DuckLake). This video is sponsored by MotherDuck: https://motherduck.com/ What you'll learn: - Why lakehouse architecture is replacing traditional data warehouses - The key differences between Iceberg, Hudi, and Delta Lake (and when to use each) - Why Hudi excels at streaming workloads while Iceberg dominates analytics - How to migrate legacy data warehouses to lakehouse architecture - The future of table format interoperability (v4 specs, Lance, DuckDB) Highlights: - Real-world use cases from Notion, Netflix, and Uber - The evolution of data systems: OLTP → OLAP → Data Lakes → Lakehouse - Deep dive into Dipankar's new book on lakehouses - Why simplicity matters: getting a lakehouse running is easy but maintaining it is hard Resources mentioned: - Dipankar's new book: https://a.co/d/07PGkH5R - Cloudera Community: community.cloudera.com - Iceberg Community: https://iceberg.apache.org/community/ My Links: Substack: https://thefulldatastack.substack.com/ LinkedIn: https://www.linkedin.com/in/hoytemerson/ X: https://x.com/HoytEmerson My content company: https://www.earlysignal.tech/ Timestamps: 00:00 - Introduction and Deepankar's role at Cloudera 11:12 - Early days of Iceberg and explaining open table formats 18:53 - Market adoption of lakehouse and table formats 25:03 - Deep dive into Apache Hudi 37:25 - Iceberg as open standard and table format specialization 42:37 - DuckLake and Lance discussion 43:33 - Competition in Lakehouses 51:15 - Book "Engineering Lakehouse" 51:56 - Book details and evaluation framework 56:08 - Where to find resources