Author: Konrad Semsch

In this post, we discuss how to implement Aurora ML performance optimizations to perform real-time inference against a SageMaker endpoint at a large scale. More specifically, we simulate an OLTP workload against the database, where multiple clients are making simultaneous calls against the database and are putting the SageMaker endpoint under stress to respond to thousands of requests in a short time window. Moreover, we show how to use SQL triggers to create an automatic orchestration pipeline for your predictive workload without using additional services.