AWS Feature Store

Commercial Information

Vendor Name

AWS Feature Store

History

Developed internally by AWS

Stand-alone vs. Platform

Part of the Amazon SageMaker platform

Delivery Model

Fully-managed cloud service

Clouds Supported

AWS

Pricing Model

Consumption Pricing

Service Level Guarantees

None

Support

24 x 7 support & response time guarantees

Feature Store Capabilities

Feature Definitions

Not available

Automated Transforms

Not available, requires setting up transformations using Data Wrangler or Glue Databrew, and setting up pipelines with SageMaker Pipelines or Airflow

Feature Ingestion

Batch ingestion with Spark or ingestion API into offline & online store

Streaming ingestion with Spark Streaming or ingestion API into offline & online store

Storage and Feature Processing Infrastructure

Online storage: DynamoDB

Offline storage: S3

Feature Sharing and Discovery

Web UI

Training Dataset Generation

Dataset generated from offline storage using AWS SDK

Online Serving

Serving endpoint / API for online data

Monitoring and Alerting

Not available

Security and Data Governance

ACL and RBAC

SSO

Data encryption at rest and in flight

Integrations

Batch data: S3, Athena, Redshift

Streaming data: Any streaming source

Amazon SageMaker Feature Store

  • What Data Sources Can Amazon SageMaker Feature Store Ingest From?

    Amazon Kinesis Data Firehose. You can also create features in data preparation tools such as Amazon SageMaker Data Wrangler, and store them directly into SageMaker Feature Store.

  • Does Amazon SageMaker Feature Store Support Streaming Data Sources?

    Streaming data is fed into SageMaker Feature via a synchronous PutRecord API. This requires buildout to make successful streaming ingestion and adds a point of failure to an ingestion pipeline.

  • Does AWS SageMaker Feature Store Have A Web UI?

    The primary web UI for the SageMaker Feature Store is a notebook.

  • How Does AWS SageMaker Feature Store Manage Batch Data Transformations?

    Sagemaker uses Data Wrangler and Spark to ingest data into the feature store, these are scheduled via lambda functions.

Reviews

There are no reviews yet. Be the first to write one.