What is a lakehouse, and how does Iceberg enable it?

A lakehouse is an architectural pattern combining a data lake's low-cost, open, scalable storage (typically cloud object storage) with a data warehouse's reliability guarantees — ACID transactions, schema enforcement, consistent snapshots — that were historically only available in a proprietary, tightly-managed warehouse system.

Before table formats like Iceberg existed, data lakes offered the low cost and openness but lacked those warehouse-grade reliability guarantees, forcing teams to either accept weaker consistency in the lake or duplicate data into a separate warehouse system specifically to get transactional guarantees — an expensive, operationally complex pattern that also introduced staleness between the two copies.

Iceberg is one of the table formats that makes the lakehouse pattern practical: by adding ACID transactions, schema evolution, and consistent snapshots directly on top of plain files in object storage, it lets a single copy of data serve both traditional data-lake workloads (large-scale batch processing, machine learning) and warehouse-style workloads (reliable, consistent BI and analytics) without needing to duplicate data between two fundamentally different systems.