A conversation about data pipelines is never complete without discussing ingestion practices, such as ETL and ELT (extract, transform, load, or extract, load, and transform). ETL transforms the data before loading it into a data warehouse, while ELT loads the data and allows the transformation to be handled within the data warehouse.
Each practice is rooted in strong business needs, and are necessary parts of modern data flow practices. However, discussion of these practices is often couched in a competitive narrative, asking which one is better. You’ll find any number of blogs out there with a title including “ETL vs ELT” in some way, shape or form. However, there are strong reasons why both are seen in use today, and neither one is going away anytime soon. So in this article, we’ll cover these two methods, the reasons they are so often pitted against one another, the situations in which one or the other thrive, and why, with Kestra’s unique capabilities, you might want to consider a hybrid solution. Let’s get started, shall we?
First, let’s look at the stages involved in both processes. As you might guess from the acronyms, both use the same stages, just in a different order.