You’re using Docker. You’re on Kubernetes. You’re building microservices. And now you’re fully ready to scale to arbitrary workloads and gracefully handle partial failures. Or are you? In this talk, former Twitter engineer Oliver Gould describes a critical component of every cloud native application: the “service mesh” that defines how services communicate with each other, shed load, and propagate failure. He traces the lineage of the service mesh from client-specific implementations, to libraries like Finagle, Hysterix, and Stubby, to modern proxy-based incarnations like Linkerd. If you’re writing cloud native applications today, the service mesh is a critical part of ensuring that they can be fast, scalable, and reliable.