Amazon Machine Learning endpoints and quotas

This section lists the the service endpoints and service quotas for the service. To connect programmatically to an AWS service, you use an endpoint. For more information, see AWS service endpoints.

In addition to the standard AWS endpoints, some AWS services offer the following endpoints in some or all of the AWS Regions that the service is supported in:

• IPv4 endpoints — These endpoints support only IPv4 requests and have the following format: service-name.region.amazonaws.com

• Dual-stack endpoints — These endpoints support both IPv4 requests and IPv6 requests and have the following format: service-name.region.api.aws

• FIPS endpoints — These endpoints comply with the Federal Information Processing Standards (FIPS) and can support either IPv4 requests or dual-stack (IPv4 and IPv6) requests. FIPS endpoints have the following format: service-name-fips.region.ip-endpoint-type, where ip-endpoint-type is amazonaws.com (for IPv4 requests) or api.aws (for IPv4 or IPv6 requests).

Service quotas, also referred to as limits, are the maximum number of service resources or operations for your AWS account. For more information, see AWS service quotas.

Note

AWS recommends that you use Regional STS endpoints within your applications, and avoid using global (legacy) STS endpoints. Regional STS endpoints reduce latency, build in redundancy, and increase session token validity. For more information about configuring your applications to use a Regional STS endpoint, see AWS STS Regionalized endpoints in the AWS SDKs and Tools Reference Guide. For more information about global (legacy) AWS STS endpoints, including how to monitor for use of this type of endpoint, see How to use Regional AWS STS endpoints in the AWS Security blog.

Service endpoints

Region Name	Region	Endpoint	Protocol
US East (N. Virginia)	us-east-1	machinelearning.us-east-1.amazonaws.com	HTTPS
Europe (Ireland)	eu-west-1	machinelearning.eu-west-1.amazonaws.com	HTTPS

Service quotas

Name	Default	Adjustable	Description
Batch prediction input records	Each supported Region: 100,000,000	Yes	The maximum number of records of batch prediction input.
Batch prediction input size	Each supported Region: 1 Terabytes	Yes	The maximum size (in TB) of batch prediction input.
Classes for multiclass ML models	Each supported Region: 100	Yes	The maximum number of classes for multiclass ML models.
Job runtime	Each supported Region: 7	No	The maximum runtime length (in days) for any job.
ML model size	Each supported Region: 2 Gigabytes	No	The maximum ML model size (in GB).
Observation size	Each supported Region: 100 Kilobytes	Yes	The maximum size (in KB) of each observation.
Rate of real-time prediction requests per endpoint	Each supported Region: 200	Yes	The maximum number of requests per second that you can perform with each real-time prediction endpoint.
Recipe complexity	Each supported Region: 10,000	Yes	The maximum recipe complexity (number of processed output variables).
Simultaneous jobs	Each supported Region: 25	Yes	The maximum number of simultaneous jobs.
Tags per object	Each supported Region: 50	No	The maximum number of tags per object.
Total RAM for all real-time prediction endpoints	Each supported Region: 10 Gigabytes	Yes	The maximum total RAM (in GB) for all real-time prediction endpoints.
Total rate of all real-time prediction requests	Each supported Region: 10,000	Yes	The maximum total number of requests per second that you can perform with all of your real-time prediction endpoints.
Training data size	Each supported Region: 100 Gigabytes	Yes	The maximum size (in GB) of training data.
Variables per data file	Each supported Region: 1,000	Yes	The maximum number of variables in a data file (schema).

For more information, see Amazon ML Quotas in the Amazon Machine Learning Developer Guide.