Automatically Correcting Networks with NEAt
summary
NEAt intercepts SDN controller updates and automatically rewrites forwarding rules to enforce policy graphs (reachability, segmentation, chaining, load‑balance) by slicing packet space, building per‑flow forwarding graphs, and solving a constrained optimization that minimizes edits. Prototype in Python runs with Mininet and on a real enterprise trace with 1M forwarding entries, detects and repairs violations including loops and missing load‑balancing, completes detect+repair in about 120 ms, and outperforms a synthesizer baseline by roughly 10^2–10^3× faster.
AI-written summary · about
chapters (16)
- 0:00Intro
- 0:32What is a network supposed to do?
- 1:26A Typical Enterprise Network
- 1:46Network errors are common
- 3:52Network Error Auto-Correct
- 4:40Challenges
- 6:41Design of NEA
- 8:12Application Mode
- 12:46Repair Algorithm
- 13:52Repair - Basic Reachability
- 14:48Repair - Generalized Reachability
- 16:16Repair - Loop Freedom
- 18:38Scalability Challenge and Solution
- 19:39Implementation & Evaluation Setup
- 20:19Enterprise Network Trace Study
- 20:46Repair vs. Synthesis
related talks
18mNetKeeper: Enhancing Network Resilience with Autonomous Network Configuration...
19mNetEdit: Fine-grained Network Tuning at Scale
25mDataplane equivalence and its applications
24mReal Time Network Policy Checking Using Header Space Analysis
30mUsing Python to design, configure, and measure large-scale networks
17mTraffic Engineering for AI Training Networks
14mBrain-on-Switch: Towards Advanced Intelligent Network Data Plane via NN-Driven Traffic...
20mContra: A Programmable System for Performance aware Routing