This document discusses stream processing at scale using Samza at LinkedIn. It describes how LinkedIn uses Samza for stream processing tasks like tracking analytics from 1.4 trillion events per day and content standardization. Samza provides scalable, fault-tolerant stream processing by partitioning streams and replicating state across clusters of hundreds of nodes. The document outlines LinkedIn's large Samza deployments and upcoming features to improve local state storage, configuration, and development.