ETL (extract, transform, load)

ETL is a data integration pattern in which data is extracted from source systems, transformed into a target structure using a separate processing engine, and then loaded into a destination such as a data warehouse. Transformation happens in flight, before the data lands, so what arrives in the destination is already cleaned and modeled.

Its main alternative is ELT, which loads raw data first and transforms it afterward using the destination's own compute. ETL made sense when warehouse compute was expensive and limited, since transformation happened on dedicated servers; it remains common where data must be cleaned, masked or validated before it can be stored at all, such as for compliance reasons. ETL is typically implemented as a data pipeline with scheduled or triggered jobs, sometimes combined with change data capture to pick up only changed records.

ETL matters wherever data needs to be shaped or restricted before it reaches its destination. Pitfalls include transformation logic that is hard to test or version because it lives inside a proprietary tool, and pipelines that become brittle as source schemas change, since transformation happens far from where analysts can inspect it through ordinary data modeling.