Frameworks like Kubernetes have solved most of the challenges of dealing with stateless applications. But for stateful applications, we are still clinging to the ancient wisdom that state shall be someone else's problem: just put it in a database! Because of that, we are still struggling with the same issues of data consistency and complex failure semantics as a decade ago. For bigger applications, we are still far away from the easy development and operations experience that is the promise of Serverless.

In this talk, we will do a deep dive into Apache Flink's newest addition: "Stateful Functions", an API and library on top of Flink that attempts to solve that very problem of scalable and resilient distributed applications. Users provide containerized, polyglot event-driven functions that can be scaled in a lightweight manner. A Stream Processor (Apache Flink) handles the messaging between functions and state consistency in a fault tolerant way, subsuming the role of the database.
This talk offers an in-depth walkthrough of how the Stateful Functions runtime works, as well as the building blocks that are provided to users to structure a Stateful Functions application.

Stateful Functions is currently being migrated into the Apache Flink project. At the moment, code, docs, and examples are still under https://statefun.io/.