What is AWS Glue? A plain-English explainer

What ETL is

ETL stands for extract, transform, load: pulling data out of a source system, reshaping or cleaning it into a usable form, and loading it into a destination where it can be queried or analysed. Our what-is-etl explainer covers the concept in depth, and our what-is-a-data-pipeline explainer covers the broader idea of moving data through a series of automated steps, of which ETL is one common pattern.

Doing this by hand does not scale once data sources multiply and grow — it needs to be automated, monitored and repeatable, which is exactly the job a managed ETL service like Glue is built to do.

What Glue actually does

Glue has two main pieces that work together. The Data Catalog is a central metadata store that keeps track of what data exists, where it lives and what shape it is in — effectively an index of your data across different sources, so that other tools can discover and query it without needing to know the underlying details themselves.

On top of that, Glue runs serverless ETL jobs: you define how data should be extracted, transformed and loaded, and Glue provisions and runs the compute needed to do it, without you managing servers or clusters directly. The result is data that has been cleaned, reshaped and catalogued, ready to be queried or used for analytics and machine learning.

Where Glue fits

Glue’s job typically sits between raw data sources and the places that data ultimately needs to live for analysis — commonly a data lake or a data warehouse. Our data-lake-vs-data-warehouse explainer covers the difference between those two destinations; Glue is largely indifferent to which one you are feeding, since its role is preparing and moving the data rather than storing it long-term.

Because it is serverless, Glue removes the need to provision and manage the infrastructure that ETL jobs would otherwise require, which is a large part of its appeal for teams that want the pipeline without the operational overhead.

Common uses

Glue is commonly used to build pipelines that move data from operational databases and other sources into a data lake or warehouse, to catalogue data so it can be discovered and queried by tools like Athena, and to run scheduled or event-driven transformation jobs that keep downstream analytics up to date.

• Cataloguing data across multiple sources so it can be discovered and queried
• Running serverless ETL jobs to clean and reshape data
• Feeding a data lake or data warehouse with prepared data
• Preparing data for analytics and machine learning workflows

When you would use Glue

Glue makes sense whenever data needs to move from where it is generated to where it can be analysed, and you would rather not build and manage that pipeline’s infrastructure yourself. It is a natural fit for teams already working within AWS who need a repeatable, managed way to discover, transform and load data at whatever scale their sources demand.

Glue and pricing

Like most AWS services, Glue is pay-for-use — you are charged for the compute resources consumed while jobs run and for the Data Catalog based on usage, rather than a flat subscription. Exact rates change over time, so check the AWS pricing page for current figures rather than relying on a fixed number here.

Where Glue appears in certification study

Glue is central to the AWS Data Engineer Associate exam, which covers data-integration and ETL services in depth, and it also appears in the Solutions Architect Associate exam as one of the standard building blocks for data-oriented architectures. Data-integration and ETL concepts more broadly are relevant to any certification with a data-engineering focus, including platform-agnostic equivalents on other clouds.

This article deliberately stays at that introductory level. How Glue jobs, crawlers and the Data Catalog are actually configured is exam-depth material, and that is what our /revision study library is built to cover.