Pricing

IMPORTANT: The Cloud Dataflow pricing model has changed, effective January 9, 2017. Dataflow jobs executed on or after January 9 are billed according to the new Cloud Dataflow pricing model. Dataflow jobs executed prior to January 9 were billed according to the pricing model detailed in this section.

The service charge for using the Cloud Dataflow managed service is expressed in hours and calculated over the total GCEU (Google Compute Engine Unit) hours created by the Google Compute Engine instances used during a job’s lifetime. There are separate service charges for batch and streaming mode.

A GCEU is a unit of CPU capacity used to describe the compute power of Compute Engine instance types. Each Google Compute Engine instance for Dataflow job carries its own GCEU unit value.

A Dataflow job will consume the following resources, each billed at their own pricing:

A Dataflow job might consume the following resources, each billed at their own pricing, including but not limited to:

If your Dataflow pipeline reads a BigQuery table with Dataflow's BigQueryIO API, the read operation performs a table export that uses BigQuery quota. These table exports do not have billing impact.

However, if you use BigQueryIO to read with a query string, those queries are subject to the BigQuery pricing model.

If your pipeline reads from Cloud Storage (using Dataflow's TextIO API, for example), those read operations are billed according to Cloud Storage pricing.

Future releases of Cloud Dataflow may have different service charges and or bundling of related services.

Dataflow Mode Price per GCEU per hr
Batch $0.01 2
Streaming $0.015 2
If you pay in a currency other than USD, the prices listed in your currency on Cloud Platform SKUs apply.

1 All Google Compute instances for a given Dataflow job have a 3-minute clock-time minimum, and are billed based on sustained use price rules.

2 All GCEUs computed for a given job have a 3-minute clock-time minimum.

Send feedback about...

Cloud Dataflow Documentation