This document discusses real-time supply chain analytics using machine learning, Kafka, and Spark. It outlines four key requirements for real-time supply chain databases: supporting massive data ingestion, serving as a system of record while providing real-time analytics, integrating with familiar ecosystems, and allowing for online scaling. The document then introduces MemSQL as a database platform that can meet these requirements using an in-memory approach. It provides an example called MemEx that combines MemSQL, Kafka, and Spark with machine learning for global supply chain management and real-time predictive analytics.