This document summarizes a talk given by Tyler Treat about using simple solutions for complex distributed systems problems. Some key points: - Distributed systems are inherently asynchronous and unreliable, but many try to build them as if they are synchronous. - Exact delivery guarantees are expensive and impossible at scale. Replayable and idempotent delivery are better alternatives. - NATS is a simple, high performance, and highly available messaging system that embraces asynchronous communication. - Workiva uses NATS as a messaging backplane between microservices for pub/sub, RPC, and load balancing. Running a local NATS daemon per VM improves performance.