Read multiple records simultaneously from Amazon SageMaker Feature Store using the new BatchGetRecord API
Amazon SageMaker Feature Store is a fully managed, purpose-built repository to store, update, retrieve, and share machine learning (ML) features. Customers can retrieve features from SageMaker Feature Store at low millisecond latencies for real-time use-cases using the GetRecord API. This API allows customers to retrieve features from a single feature group and access one record per API call. Until now, to read multiple records at a time from SageMaker Feature Store, customers needed to call the GetRecord API multiple times and manage parallelization of the API calls to achieve lower latency, which increased operational complexity.
Amazon SageMaker Feature Store is a fully managed, purpose-built repository to store, update, retrieve, and share machine learning (ML) features. Customers can retrieve features from SageMaker Feature Store at low millisecond latencies for real-time use-cases using the GetRecord API. This API allows customers to retrieve features from a single feature group and access one record per API call. Until now, to read multiple records at a time from SageMaker Feature Store, customers needed to call the GetRecord API multiple times and manage parallelization of the API calls to achieve lower latency, which increased operational complexity.
Today, we announce a new BatchGetRecord API that provides customers greater flexibility and higher throughput when retrieving records from SageMaker Feature Store. With the BatchGetRecord API, customers can access multiple records per API call, across multiple feature groups while maintaining low latency, typically in milliseconds, at inference. Customers can do this by simply providing record identifiers, feature names, and feature group names as parameters to the BatchGetRecord API.
A single BatchGetRecord API call can contain as many as 100 records and can query up to 10 feature groups. To learn more, please view the documentation here. To get started, log into the Amazon SageMaker console.