Deploying Stateful FaaS on Streaming Dataflows
summary
A system compiles stateful functions and their message flows into a streaming dataflow that runs on Apache Flink, making state local to operators and relying on Flink checkpointing for exactly-once semantics. A Scala-based runtime called the Rho prototype exposes pure, stateful and orchestrating functions via function namespaces, routes requests through per-namespace queues (Kafka today), and demonstrates linear scalability with latencies under 0.07 s per function in basic benchmarks.
AI-written summary · about
chapters (5)
- 0:00Introduction and motivation
- 13:13Programming model and runtime
- 23:16Demo and execution semantics
- 30:21Performance and research directions
- 33:38Q&A session
related talks
42mInterview and AMA with the creators of Stateful Functions
55mStateful Functions: Polyglot Event-Driven Functions for Stateful Distributed Applications
35mHands-On Stateful Serverless Applications with K8s and Stateful Functions
24mStateful Functions: Flexible like a Microservice, Consistent like a Monolith
34mHands-on Stateful Serverless Applications with Kubernetes and Stateful Functions
41mKeynote: Introducing Stateful Functions 2.0: Stream Processing meets Serverless Applications
27mStateful Functions 3.0: Towards a Cross-Platform Stateful Serverless Stack
38mPolyglot Actor Model on Flink Stateful Functions