Sequence Abstractions for Flexible, Line-Rate Network Monitoring
Talkat 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI '24) · 2024-04-16..04-18·
summary
A new system called FLM compiles sequence-monitoring programs to P4 that run on programmable switch ASICs, splitting specs into state management plus a regular expression and using SMT-based synthesis to produce a tiny microprogram. The paper shows 15 real monitoring tasks map to DFAs of 4–19 states and 2–32 symbols, the compiled P4 fits hardware pipeline stages and adds under 15% extra header fields and instruction words, and typical compilation plus vendor P4 compile time is about 2 minutes.
AI-written summary · about
chapters (6)
- 0:00Sequence abstractions for monitoring
- 2:46FLM system architecture
- 3:52Detecting DHCP misuse
- 7:49Compiling FLM patterns to hardware
- 12:44Evaluating FLM flexibility and performance
- 15:32Summary
related talks
01h 08mThe Network is the Computer: A Programming Language Perspective
12mFlightplan: Dataplane Disaggregation and Placement for P4 Programs
15mWhen P4 Meets Run-to-completion Architecture
15mSynthesizing Runtime Programmable Switch Updates
18mLiveness Verification of Stateful Network Functions
20mContra: A Programmable System for Performance aware Routing
14mModular Switch Programming Under Resource Constraints
15mOn Temporal Verification of Stateful P4 Programs