This document explains the pricing for Cloud Data Fusion. To see the pricing for other products, read the Pricing documentation.

For pricing purposes, usage is measured as the length of time, in minutes, between the time a Cloud Data Fusion instance is created to the time it is deleted. Although the rate for pricing is defined on the hour, Cloud Data Fusion is billed by the minute. Usage is measured in hours (30 minutes is 0.5 hours, for example) to apply hourly pricing to minute-by-minute use.

If you pay in a currency other than USD, the prices listed in your currency on Google Cloud SKUs apply.

Cloud Data Fusion pricing is split across two functions: pipeline development and execution.

For pipeline development, Cloud Data Fusion offers the following three editions:

For pipeline execution, you are charged for the Managed Service for Apache Spark clusters that Cloud Data Fusion creates to run your pipelines at the current Managed Service for Apache Spark rates.

In addition to the development cost of a Cloud Data Fusion instance,̦ you are billed only for any resources that you use to execute your pipelines, such as:

• Managed Service for Apache Spark
• Cloud Storage
• Networking
• BigQuery

★ For building replication jobs, BigQuery flat-rate pricing is recommended, not on-demand pricing.

Currently, pricing for Cloud Data Fusion is the same for all supported regions.

Consider a Cloud Data Fusion instance that has been running for 24 hours, and there are no free hours remaining for the Basic edition. Based on the edition, the instance charge for Cloud Data Fusion is summarized in the following table:

★ Note: Cloud Data Fusion instances, once provisioned, always need to be available. After you delete instances, they cannot be recovered and any pipeline data is lost. For estimated monthly costs, refer to the Pricing overview.

During this 24-hour period, you ran a pipeline that read raw data from Cloud Storage, performed transformations, and wrote the data to BigQuery every hour. Each run took approximately 15 minutes to complete. In other words, the Managed Service for Apache Spark clusters that were created for these runs were alive for 15 minutes (0.25 hours) each. Assume that the configuration of each Managed Service for Apache Spark cluster was the following:

The Managed Service for Apache Spark clusters each have 24 virtual CPUs: 4 for the master and 20 spread across the workers. For Managed Service for Apache Spark billing purposes, the pricing for this cluster would be based on those 24 virtual CPUs and the length of time each cluster ran.

Across all runs of your pipeline, the total charge incurred for Managed Service for Apache Spark can be calculated as:

Managed Service for Apache Spark charge = # of vCPUs * number of clusters * hours per cluster * Managed Service for Apache Spark price

= 24 * 24 * 0.25 * $0.01

= $1.44

The Managed Service for Apache Spark clusters use other Google Cloud products, which would be billed separately. Specifically, these clusters would incur charges for Compute Engine and Standard Persistent Disk Provisioned Space. You will incur storage charges for Cloud Storage and BigQuery, depending on the amount of data your pipeline processes.

To determine these additional costs based on current rates, you can use the billing calculator.