Amazon Glue Pricing

With Amazon Glue, you only pay for the time your ETL job takes to run. There are no resources to manage, no upfront costs, and you are not charged for startup or shutdown time. We charge you an hourly rate based on the number of Data Processing Units (or DPUs) used to run your ETL job. A single Data Processing Unit (DPU) provides 4 vCPU and 16 GB of
memory. We bill for jobs and development endpoints in increments of 1 second, rounded up to the nearest second. 

There are three types of Amazon Glue jobs: Apache Spark, Spark Streaming, and Python shell. 

Apache Spark and Spark Streaming job runs require a minimum of 2 DPU. By default, Amazon Glue allocates 10 DPU to each Apache Spark job and 2 DPU to each streaming job. Jobs using Amazon Glue version 0.9 or 1.0 have a 10-minute minimum billing duration, while jobs that use Glue versions 2.0 and later have a 1-minute minimum.

For Python shell jobs, you can allocate either 1 DPU or 0.0625 DPU. By default, Amazon Glue allocates 0.0625 DPU to each Python shell job. These jobs have a 1-minute minimum billing duration. 

Interactive Sessions is optional, and billing applies only if you use it for interactive ETL code development. We charge for Interactive Sessions based on the time the session is active and the number of DPUs. Interactive sessions have configurable idle timeouts. Amazon Glue Interactive Sessions require a minimum of 2 DPUs and have a default of 5 DPU. There is a 1-minute minimum billing duration for each provisioned Interactive Session. Amazon Glue Studio Job Notebooks provide a built-in interface for Interactive Sessions. We do not charge for the Job Notebooks but do charge for the Interactive Sessions they use.

Development endpoints is optional, and billing applies only if you use it for interactive ETL code development. We charge for development endpoints based on the time the endpoint is provisioned and the number of DPU. Development endpoints do not time out. Development Endpoints require a minimum of 2 DPUs and have a default of 5 DPU. There is a 10-minute minimum billing duration for each provisioned development endpoint. 

Amazon Glue Studio data previews allow you to test your transformations during the job authoring process. Each Amazon Glue Studio data preview session uses 2 DPUs, runs for 30 minutes, and stops automatically.

Pricing   

• ¥3.021 per DPU-Hour for each Apache Spark or Spark Streaming job, billed per second with a 1-minute minimum (Glue version 2.0 and later) or 10-minute minimum (Glue version 0.9/1.0)  
• ¥3.021 per DPU-Hour for each Python Shell job, billed per second, with a 1-minute minimum
• ¥3.021 per DPU-Hour for each provisioned Development Endpoint, billed per second with a 10-minute minimum
• ¥3.021 per DPU-Hour for each Interactive Session billed per second with a 1-minute minimum. 
• ¥3.021 per DPU-Hour for each Amazon Glue Studio data preview session, billed in 30-minute units and invoiced as development endpoints
  

Additional charges

If you ETL data from data sources such as Amazon S3, Amazon RDS, or Amazon Redshift, you are charged standard request and data transfer rates. If you use Amazon CloudWatch, you are charged standard rates for CloudWatch logs and CloudWatch events.

Pricing example

ETL job example: Consider an Amazon Glue Apache Spark job that runs for 15 minutes and uses 6 DPU. The price of 1 DPU-Hour is ¥3.021. Since your job ran for 1/4th of an hour and used 6 DPUs, We will bill you 6 DPU * 1/4 hour * ¥3.021, or ¥4.532. 

Amazon Glue Studio Job Notebooks and Interactive Sessions example: Suppose you use a notebook in Amazon Glue Studio to interactively develop your ETL code. An Interactive Session has 5 DPU by default. If you keep the session running for 24 minutes or 2/5th of an hour, you will be billed for 5 DPUs * 2/5 hour at ¥3.021 per DPU-Hour or ¥6.042. 

ML Transforms example: Similar to Amazon Glue jobs runs, the cost of running ML Transforms, including FindMatches on your data will vary based on the size of your data, the content of your data, and the number and types of nodes that you use. In the following example, we used FindMatches to integrate points of interest information from multiple data sources. With a data set size of ~11,000,000 rows (1.6GB), a size of Label data (examples of true matches or true no-matches) of ~8,000 rows (641kb), running on 16 instances of type G.2x, then you would have a labelset generation runtime of 34 minutes at a cost of ¥54.781, a metrics estimation runtime of 11 minutes at a cost of ¥17.723, and a FindMatches job execution runtime of 32 minutes at a cost of ¥51.558.