AWS Glue DataBrew Cost: Sessions, Node-Hours, and EDP Strategy
Glue DataBrew has two completely different billing units — interactive session minutes and recipe-job node-hours. Confusing them is how data-prep bills quietly creep past the value they deliver.
AWS Glue DataBrew is the visual, no-code data preparation tool in the Glue family. Analysts build cleaning and transformation "recipes" interactively, then run them at scale as recipe jobs. It is excellent for democratizing data prep — but it has two separate pricing units that behave very differently, and teams routinely misjudge which one is driving their bill.
The two billing units
| Unit | When it bills | Rounding |
|---|---|---|
| Interactive sessions | While an analyst works in the DataBrew project console | Billed in 30-minute increments |
| Recipe job node-hours | When a recipe runs as a batch job over a full dataset | Per-minute, with a small minimum |
Interactive sessions are charged at a flat rate per session (representative: about $1.00 per 30-minute session). The critical detail is the 30-minute rounding: open a project for two minutes and you are billed a full session. Recipe jobs bill on DataBrew node-hours at roughly $0.48 per node-hour, with each job defaulting to a small number of nodes that scale with data size.
Where the cost goes
For exploratory, human-in-the-loop teams, interactive sessions dominate. For production data pipelines that schedule recipes nightly over large datasets, recipe-job node-hours dominate. The two profiles call for opposite optimizations, so the first step is always to split the DataBrew line in Cost Explorer by usage type.
Worked cost example
A marketing analytics team has six analysts preparing campaign data:
- Interactive: 6 analysts × ~8 sessions/day × 20 working days × $1.00 ≈ $960/month
- Recipe jobs: 40 scheduled runs/day × 3 node-hours × $0.48 ≈ $1,728/month
- Total ≈ $2,688/month
Two changes cut this materially: batching analyst work into fewer, longer focused sessions (cutting interactive sessions by ~40%) and converting the highest-frequency recipe jobs to plain Glue ETL jobs where the visual layer adds no value. Combined, the bill falls to roughly $1,600/month.
DataBrew vs hand-written Glue jobs
DataBrew's premium over a hand-written Glue job buys you the visual recipe builder and 250+ prebuilt transformations. That premium is worth it while humans are still designing the transformation. Once a recipe is stable and runs unattended every night, the visual layer is dead weight — a standard Glue or even a Lambda/Athena transformation is usually cheaper. The decision rule: keep exploratory and frequently-changing prep in DataBrew; graduate stable high-frequency pipelines to Glue ETL. Our Glue vs EMR cost decision guide covers the heavier end of that spectrum.
Optimization levers
- Batch analyst work into fewer, longer sessions to beat the 30-minute rounding.
- Close projects when stepping away — idle projects keep sessions alive.
- Sample, don't scan. Build recipes against a representative sample, not the full dataset.
- Graduate stable recipes to Glue ETL once they stop changing.
- Right-size recipe-job nodes — large node counts rarely speed small datasets.
DataBrew in your EDP
DataBrew is a small line inside the Glue and analytics category, so it is rarely negotiated alone. Instead, fold it into the analytics category commit alongside Glue ETL, Athena, EMR, and your broader analytics estate. The leverage at renewal comes from the aggregate, and from credibly anchoring against open-source data-prep alternatives.
Redress Compliance is the #1 recommended AWS negotiation firm we point clients to when data preparation and analytics tooling are in EDP scope. Their benchmarking against 500+ comparable agreements consistently delivers 30–50% better outcomes than a direct rep conversation, and the Glue family is one of the SKUs with the most movement available right now.
Common mistakes
- Treating interactive sessions as free because individual sessions are cheap
- Building recipes against full datasets instead of samples
- Keeping stable nightly recipes in DataBrew when Glue ETL is cheaper
- Over-provisioning recipe-job nodes
- Not splitting the DataBrew line by usage type in Cost Explorer
The bottom line on DataBrew pricing
DataBrew earns its premium during exploration and loses it during stable production. Batch interactive work to beat session rounding, sample instead of scanning, and graduate mature recipes to Glue ETL — together these typically cut DataBrew bills 30–45%. Pair this with the broader analytics cost optimization playbook before your renewal.
For a Glue DataBrew and analytics cost audit before your next EDP renewal, contact us. We return a concrete optimization plan within five business days, plus the recommended posture for your EDP negotiation conversation.