AWS Glue is a serverless ETL service for data preparation and integration. Pricing is based on DPU-hours (Data Processing Units) consumed during job execution. See the Glue pricing page for current rates. The key cost driver is how efficiently your jobs use compute — poorly written Spark jobs can consume 10x more DPU-hours than necessary for the same data volume.
TL;DR: Glue ETL jobs cost $0.44 per DPU-hour with a minimum of 2 DPUs (minimum charge $0.88/hour). Glue Data Catalog stores 1 million objects free, then $1.00 per 100K objects/month. Crawlers cost the same $0.44 per DPU-hour. Use Glue 4.0 with auto-scaling, minimize crawler frequency, and optimize Spark partitioning to reduce costs by 40-60%.
Glue Component Pricing
ETL Job Pricing
| Component | Cost |
|---|---|
| Apache Spark ETL | $0.44 per DPU-hour |
| Apache Spark Streaming | $0.44 per DPU-hour |
| Python Shell | $0.44 per DPU-hour (0.0625 DPU minimum) |
| Ray (ML workloads) | $0.44 per DPU-hour |
| Minimum DPUs (Spark) | 2 DPUs |
| Minimum DPUs (Python Shell) | 0.0625 DPU |
| Billing increment | 1-second (minimum 1 minute) |
One DPU = 4 vCPUs + 16 GB memory.
Data Catalog
| Component | Cost |
|---|---|
| First 1M objects stored | Free |
| Beyond 1M objects | $1.00 per 100,000 objects/month |
| First 1M requests | Free |
| Beyond 1M requests | $1.00 per million requests |
Crawlers
| Component | Cost |
|---|---|
| Crawler runtime | $0.44 per DPU-hour |
| Minimum | 2 DPUs |
| Billing increment | 1-second (minimum 10 minutes) |
Real-World Cost Examples
| Scenario | DPU-Hours | Monthly Cost |
|---|---|---|
| Small ETL: 10 jobs, 5 DPU, 10 min each, daily | 250 | $110 |
| Medium ETL: 20 jobs, 10 DPU, 30 min each, daily | 3,000 | $1,320 |
| Large ETL: 50 jobs, 20 DPU, 1 hr each, daily | 30,000 | $13,200 |
| Python Shell: 100 lightweight jobs, daily | 19 | $8.33 |
Hidden Costs
Crawler Over-Runs
Crawlers have a 10-minute minimum charge. Running a crawler that finishes in 30 seconds still costs 10 minutes x 2 DPUs x $0.44/DPU-hour = $0.15. Running unnecessary crawlers hourly costs $108/month.
Development Endpoints (Deprecated)
Interactive development endpoints cost $0.44/DPU-hour and were often left running. AWS now recommends Glue Studio notebooks instead, which are charged per session.
Interactive Sessions
| Component | Cost |
|---|---|
| Glue Studio notebooks | $0.44 per DPU-hour |
| Minimum | 2 DPUs |
| Idle timeout | Configurable (default: 30 min) |
Developers forgetting to stop sessions can accumulate significant costs.
Cost Optimization Strategies
1. Use Glue 4.0 with Auto-Scaling
Glue 4.0 auto-scales workers based on workload, starting with 2 DPUs and scaling up only when needed. This can reduce DPU-hours by 30-50% compared to fixed-size Glue 2.0/3.0 jobs.
2. Use Python Shell for Lightweight Jobs
Python Shell jobs use as little as 0.0625 DPU (1/32 of a standard DPU). For simple transformations, file moves, or API calls, Python Shell costs $0.003/hour vs $0.88/hour for minimum Spark jobs.
3. Schedule Crawlers Wisely
Run crawlers only when schema changes are expected — not on a fixed schedule. For static schemas, skip crawlers entirely and define tables manually in the Data Catalog.
4. Optimize Spark Partitioning
Repartition data to match the parallelism of your DPU allocation. Too many small partitions waste overhead; too few large partitions leave DPUs idle.
Related Guides
FAQ
How does Glue pricing compare to EMR?
For small to medium ETL workloads, Glue is simpler but more expensive per compute-hour. EMR with Spot instances can be 60-70% cheaper for large, long-running jobs. Glue's advantage is zero infrastructure management.
Is the Glue Data Catalog free?
Effectively yes for most teams. The first 1 million objects and 1 million requests per month are free. Only data lake environments with massive schema catalogs exceed this.
Can I reduce Glue costs without rewriting jobs?
Yes. Enable auto-scaling (Glue 4.0), right-size the number of workers, reduce crawler frequency, and set shorter timeouts. These changes alone typically reduce costs 30-40%.
Lower Your Glue Costs with Wring
Wring helps you access AWS credits and volume discounts to lower your Glue costs. Through group buying power, Wring negotiates better rates so you pay less per DPU-hour.
