Amazon SageMaker Feature Store in-memory online store for low latency feature retrieval
Amazon SageMaker Feature Store now supports a fully managed, in-memory online store, which enables you to retrieve features for model serving in real time for high throughput ML applications. The new online store is powered by ElastiCache for Redis, which is a blazing fast in-memory data store built on open-source Redis.
Amazon SageMaker Feature Store now supports a fully managed, in-memory online store, which enables you to retrieve features for model serving in real time for high throughput ML applications. The new online store is powered by ElastiCache for Redis, which is a blazing fast in-memory data store built on open-source Redis.
With this launch, you can select the new in-memory online store when you create a feature group, then use familiar SageMaker APIs for reading and writing data. You can use capabilities such as time-to-live (TTL) to manage storage size, monitor using CloudTrail logs to track API calls, as well as operational metrics in CloudWatch for invocations, errors, latency, and more. SageMaker manages the infrastructure so you don't need to worry about operational tasks like scaling, availability, and security.
This capability is available in 18 regions globally, including US East (N. Virginia, Ohio), US West (Oregon, N. California), Canada (Central), Europe (Frankfurt, Ireland, London, Paris, Stockholm), Asia Pacific (Hong Kong, Mumbai, Seoul, Singapore, Sydney, Tokyo), Africa (Cape Town), and South America (São Paulo). To learn more, please view the documentation here. To get started, go to SageMaker Studio from the Amazon SageMaker console.