An overview of Lakehouse Architecture

Presented the DBT, Trino, Iceberg data lakehouse architecture that we implemented at Check24 to 200+ data scientists and engineers at the Check24 Data Science Summit, showcasing the platform’s impact and best practices.

The presentation explains the transition from traditional SQL-based data warehouses to a lakehouse architecture, highlighting the roles of DBT, Trino, and Apache Iceberg in building a modern data platform.

Motivation: Why Move to a Lakehouse

• Traditional warehouses have limitations in scalability, flexibility, and cost.
• A lakehouse combines:
  • Data lake scalability (cheap storage, flexibility)
  • Warehouse reliability (ACID guarantees, structured querying)
• Goal: unified platform for raw + processed + analytics-ready data.

Technology Choices (Rationale)

• Parquet → efficient columnar storage
• Iceberg → reliable table management on object storage
• Trino → fast distributed SQL query engine
• DBT → structured, maintainable transformation workflows

For more details read this report