Iris: Expressive Traffic Analysis for the Modern Internet
summary
Iris implements a compiler that compiles Rust subscriptions into an optimized pipeline by decomposing connection-layer state machines into incremental filters and actions, deduplicating work across subscriptions and fusing runtime pipelines. It exposes connection- and application-layer types to user code, runs hundreds of concurrent analysis tasks, and processes 100Gbps+ on a single commodity server (100–120Gbps on Stanford traffic) while keeping complex tasks compact (video inference in <200 lines Rust) and yielding an order-of-magnitude developer effort reduction.
AI-written summary · about
chapters (10)
- 0:00Traffic analysis challenges
- 1:21Building systems from scratch
- 4:11Limitations of existing tools
- 5:54Introducing Iris framework
- 6:58Iris programming model
- 9:02Connection state machines
- 10:49Iris compiler and runtime
- 13:34Evaluating scalability and performance
- 15:43Real-world applications
- 16:28Conclusion and future
related talks
16mIris-Wasm: Robust and Modular Verification of WebAssembly Programs
50mHigh Performance Network Programming in Rust
36mFast, secure, safe: the web that can still be
14mFOSS in building Telecom Networks
36mInformation Reliability Service - a novel approach to fraud detection
30mIntegrating Rust into Tor: Successes and Challenges
59mWhat's the deal with unsafe Rust?
41mNetwork Observability