ELT

Back to DuckDB Data Engineering Glossary

ELT (Extract, Load, Transform) is a modern data integration process that reverses the order of traditional ETL (Extract, Transform, Load) workflows. In ELT, raw data is first extracted from various sources and loaded directly into a target data warehouse or lake without prior transformation. The transformation step occurs afterwards within the destination system, leveraging its processing power and scalability. This approach allows for greater flexibility, as analysts can transform data on-demand and iterate on transformations without re-extracting or re-loading. ELT is particularly well-suited for cloud-based data warehouses like Snowflake, BigQuery, or Redshift, which can handle large-scale data transformations efficiently. Tools like dbt have popularized ELT by providing a framework for managing and executing in-warehouse transformations using SQL.