What is a data lakehouse?

A data lakehouse is a data platform that stores data as open files but adds the ACID transactions, schema enforcement, and query performance you'd expect from a warehouse. It replaces the older pattern of a lake for raw data and a separate warehouse for anything trustworthy enough to report on.

How a lakehouse differs from a lake or a warehouse

A data lake stores anything cheaply but gives you no guarantees about correctness: two writers can corrupt the same file, and there's no built-in way to enforce a schema. A data warehouse gives you those guarantees but at the cost of copying data into a closed, often expensive system. A lakehouse keeps data in open file storage and adds a table format, like Apache Iceberg, that provides the transaction and schema guarantees a warehouse has, without the copy.

Why it matters

Fewer copies of data means fewer places for it to drift out of sync. A lakehouse is also engine-agnostic: because the tables are open, more than one query engine can read the same data, instead of locking it inside one vendor's storage format.

How OptimaFlo implements this

OptimaFlo's medallion architecture, Raw, Clean, and Ready, is a lakehouse implementation: every layer is stored as an Apache Iceberg table in your own AWS or GCP account, and OptimaFlo's AI data team does the modeling work of moving data from one layer to the next.

Related terms

• Apache Iceberg: The open table format that gives OptimaFlo's lakehouse its ACID and schema guarantees.
• Medallion architecture: The Raw, Clean, and Ready layers that make up OptimaFlo's lakehouse.
• AI data team: The AI roles that build and maintain the pipelines feeding the lakehouse.