Does SageMaker Feature Store use Athena queries (data virtualization) to make its queries performant?

Question

SageMaker boasts rapid performance for reading data from batch and online. I am wondering what is happening behind the scenes. Are these Athena queries behind the scenes that the feature stores uses? Is it accessing materialized views, or are these queries running against partitioned data in S3 directly (and thus faster due to partitioning rather than materialized views)?

Answer 1

The video goes into the high-level details of FeatureStore and this E2E sample shows how to query against the offline data store.