Flightplan: Dataplane Disaggregation and Placement for P4 Programs
summary
Flightplan automatically splits a monolithic P4 program into cooperating subprograms, generates resource dependency rules from code, and uses a planner-based split-and-place algorithm to map pieces to heterogeneous dataplanes (programmable switches, FPGAs, x86). A thin runtime OS-like layer handles handover and fault detection. The implementation, tests and browser demo run testbed experiments and simulations across real switches, FPGAs and servers, and show measurable gains in bandwidth, energy and latency while preserving original P4 behavior.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:24Dataplane Programmability
- 1:14Current paradigm: dataplane program
- 2:11Example: "Crosspod" Program
- 5:08Heterogeneity
- 7:51Example program (Crosspod)
- 8:27Abstract program
- 8:49Abstract Resource Semantics
- 9:14Runtime: Fault Detection
- 9:38Resources & Behavior vs Functions Testbed experiment
- 10:03Plan comparison (Testbed + Simulation)
- 10:20Multiple Programs vs Runtimes vs Splits in same network (Simulation)
related talks
19mExoPlane: An Operating System for On-Rack Switch Resource Augmentation
17mHiP4-UPF: Towards High-Performance Comprehensive 5G User Plane Function on P4...
14mMultitenancy for Fast and Programmable Networks in the Cloud
01h 08mThe Network is the Computer: A Programming Language Perspective
15mWhen P4 Meets Run-to-completion Architecture
15mSynthesizing Runtime Programmable Switch Updates
16mSequence Abstractions for Flexible, Line-Rate Network Monitoring
25mDataplane equivalence and its applications