When you specify a  job, you may pass to the running environment a set of property values to apply to the execution of the job. Overrides are applied to individual jobs. 

These property values override any settings applied to the project.

Dataflow Execution Properties

Default execution settings:

By default, runs your job in the us-central1 region on an n1-standard-1 machine. As needed, you can change the geo location and the machine where your job is executed.

Tip: You can change the default values for the following in your project settings. See Project Settings Page .

Making changes to these settings can affect performance times for executing your job.

Tip: For more information on how the following settings affect your jobs, see Run Job on Cloud Dataflow.

SettingDescription
Regional Endpoint

A regional endpoint handles execution details for your job, its location determines where the job is executed.

Zone

A sub-section of region, a zone contains specific resources for a given region.

Select Auto Zone to allow the platform to choose the zone for you.

Machine Type

Choose the type of machine on which to run your job. The default is n1-standard-1.

Note: not all machine types supported directly through .

For more information on these regional endpoints, see https://cloud.google.com/dataflow/docs/concepts/regional-endpoints.

For more information on machine types, https://cloud.google.com/compute/docs/machine-types.


Advanced settings:

SettingDescription
VPC Network mode

If the network mode is set to Auto (default), the job is executed over publicly available IP addresses. Do not set values for Network, Subnetwork, and Worker IP address configuration.

As needed, you can override the default settings configured for your project for this job. Set this value to Custom.

NOTE: Avoid applying overrides unless necessary. These network settings apply to job execution. Preview and sampling use the default network settings.

  1. Specify the name of the VPC network in your region.
  2. Specify the short or full URL of the Subnetwork. If both Network and Subnetwork are specified, Subnetwork is used. See https://cloud.google.com/dataflow/docs/guides/specifying-networks.
  3. Review and specify the Worker IP address configuration setting. See below.

For more information:

NetworkTo use a different VPC network, enter the name of the VPC network to use as an override for this job. Click Save to apply the override.
Subnetwork

To specify a different subnetwork, enter the URL of the subnetwork. The URL should be in the following format:

regions/<REGION>/subnetworks/<SUBNETWORK>

where:

  • <REGION> is the region identifier specified under Regional Endpoint. These values must match.
  • <SUBNETWORK> is the subnetwork identifier.

If you have access to another project within your organization, you can execute your job through it by specifying a full URL in the following form:

https://www.googleapis.com/compute/v1/projects/<HOST_PROJECT_ID>/regions/<REGION>/subnetworks/<SUBNETWORK>

where:

  • <HOST_PROJECT_ID> corresponds to the project identifier. This value must be between 6 and 30 characters. The value can contain only lowercase letters, digits, or hyphens. It must start with a letter. Trailing hyphens are prohibited.
Click Save to apply the override.

For more information on these settings, see Project Settings Page.

SettingDescription
Worker IP address configuration

If the VPC Network mode is set to custom, then choose one of the following:

  • Allow public IP addresses - Use workers that are available through public IP addresses. No further configuration is required.
  • Use internal IP addresses only - workers use private IP addresses for all communication.
    • If a Subnetwork is specified, then the Network value is ignored.
    • The specified Network or Subnetwork must have Private Google Access enabled.
Autoscaling Algorithms

The type of algorithm to use to scale the number of Google Compute Engine instances to accommodate the size of your job. Possible values:

  • Throughput based - Scaling is determined by the volume of data expected to be passed through .
  • None - None algorithm is applied.
    • If none is selected, use initial number of workers to specify a fixed number of Google Compute Engine instances.
Initial number of workersNumber of Google Compute Engine instances with which to launch the job. This number may be adjusted as part of job execution. This number must be an integer between 1 and 1000, inclusive.
Maximum number of workers

Maximum number of Google Compute Engine instances to use during execution. This value must be greater than the initial number of workers and must be an integer between 1 and 1000, inclusive.

Service account

Email address of the service account under which to run the job.

Labels

Create or assign labels to apply to the billing for the jobs run in your project. You may reference up to 64 labels.

NOTE: Each label must have a unique key name.

For more information, see https://cloud.google.com/resource-manager/docs/creating-managing-labels.

Notes on behavior: