Google Cloud uses regions, subdivided into zones, to define the geographic location of physical computing resources. Google stores and processes your data only in the region you specify for all features of AI Platform (Unified) except for data labeling tasks and any feature in experimental status.
Choosing your location
You can choose any supported location when you create a dataset, train a custom-trained model that does not use a managed dataset, or when you import an existing model. You should typically use the region closest to your physical location or the physical location of your intended users, but check that the AI Platform service you want to use is supported in your region. There is no global location.
For operations other than creating a dataset or importing a model, you must use the location of the resources you are operating on. For example, when you create a training pipeline that uses a managed dataset, you must use the region where the dataset is located.
Specifying the location using Cloud Console
When you use Google Cloud Console, you specify the location by using the location dropdown menu:
Specifying the location using the AI Platform (Unified) API
You specify the location for an AI Platform API request by using the appropriate regional endpoint.
For example, to make a request in the europe-west4 region, use
the following endpoint:
https://europe-west4-aiplatform.googleapis.com
To make a request in the us-central1 region, use the following
endpoint:
https://us-central1-aiplatform.googleapis.com
For a full list of supported endpoints, see the API reference documentation.
When you specify a resource, you use the name of the resource's region as the
location. For example, a dataset in the us-central1 region would be specified
using the following path:
projects/PROJECT/locations/us-central1/datasets/DATASET_ID
Available regions
AI Platform is available in the following regions:
Americas
- Iowa (us-central1)
Europe
- Netherlands (europe-west4)
Asia Pacific
- Taiwan (asia-east1)
Google Cloud also provides additional regions for products other than AI Platform.
Service availability
Some AI Platform services are not available in all regions. The following table lists the services that are available in each region:
Americas
| Region | Iowa us-central1 |
|---|---|
| AutoML for image data (training, online predictions, and batch predictions) | |
| AutoML for tabular data (training, online predictions, and batch predictions) | |
| AutoML for text data (training, online predictions, and batch predictions) | |
| AutoML for video data (training and batch predictions) | |
| Custom model training | |
| Custom model online predictions | |
| Custom model batch predictions | |
| Data Labeling |
Europe
| Region | Netherlands europe-west4 |
|---|---|
| AutoML for image data (training, online predictions, and batch predictions) | |
| AutoML for tabular data (training, online predictions, and batch predictions) | |
| AutoML for text data (training, online predictions, and batch predictions) | |
| AutoML for video data (training and batch predictions) | |
| Custom model training | |
| Custom model online predictions | |
| Custom model batch predictions | |
| Data Labeling |
Asia Pacific
| Region | Taiwan asia-east1 |
|---|---|
| AutoML for image data (training, online predictions, and batch predictions) | |
| AutoML for tabular data (training, online predictions, and batch predictions) | |
| AutoML for text data (training, online predictions, and batch predictions) | |
| AutoML for video data (training and batch predictions) | |
| Custom model training | |
| Custom model online predictions | |
| Custom model batch predictions | |
| Data Labeling |
Region considerations
Training with accelerators
Accelerators are available on a region basis. The following table lists all the available accelerators for each region:
Americas
| Region | Iowa us-central1 |
|---|---|
| NVIDIA Tesla K80 | |
| NVIDIA Tesla P4 | |
| NVIDIA Tesla P100 | |
| NVIDIA Tesla T4 | |
| NVIDIA Tesla V100 |
Europe
| Region | Netherlands europe-west4 |
|---|---|
| NVIDIA Tesla K80 | |
| NVIDIA Tesla P4 | |
| NVIDIA Tesla P100 | |
| NVIDIA Tesla T4 | |
| NVIDIA Tesla V100 |
Asia Pacific
| Region | Taiwan asia-east1 |
|---|---|
| NVIDIA Tesla K80 | |
| NVIDIA Tesla P4 | |
| NVIDIA Tesla P100 | |
| NVIDIA Tesla T4 | |
| NVIDIA Tesla V100 |
If your job uses multiple types of GPUs, they must all be available in a single
zone in your region. For example, you cannot run a job in us-central1 using
NVIDIA Tesla T4 GPUs, NVIDIA Tesla K80 GPUs, and NVIDIA Tesla P100 GPUs.
While all of these GPUs are available for jobs in us-central1, no single zone
in that region provides all three types of GPU. To learn more about the zone
availability of GPUs, see the comparison of GPUs for compute
workloads.
Insufficient resources
Demand is high for GPUs and for compute resources in the us-central1 region.
You may get an error message in your job logs that says: Resources are
insufficient in region: <region>. Please try a different region..
To resolve this issue, try using a different region or try again later.
Cloud Storage bucket requirements
Some AI Platform tasks, such as importing data from your local computer, use a Cloud Storage bucket.
Create your Cloud Storage buckets in the same region that your AI Platform job runs in.
Use the following settings when creating Cloud Storage a bucket to read and write data for your AI Platform job.
- Location type:
Region. - Location: The region that your AI Platform (Unified) job runs in, such as
us-central1,europe-west4, orasia-east1. - Storage class:
Standard(sometimes displayed in the Cloud Storage browser asRegional).
- Location type:
Restricting resource locations
Organization policy administrators can restrict the regions available where you can use AI Platform by creating a resource locations constraint. Read about how a resource locations constraint applies to AI Platform
Resource locations constraints don't apply to DataLabelingJob
resources.