Toward Orchestration of Complex Networking Experiments
Talkat 13th USENIX Workshop on Cyber Security Experimentation and Test (CSET '20) · on Monday, 2020-08-10·
summary
MAGI converts a YAML-based specification into orchestrated runs on emulation testbeds, launching Python agents via lightweight daemons and a central orchestrator that schedules event streams and triggers. It runs on DETER for seven years, scales from student projects with ~30 clients to deployments with 2,000 web clients and 100-node Tor experiments, and reproduces an IEEE‑39 power‑bus experiment over an 18-node comms topology. Results show repeatable execution, asynchronous large‑scale concurrency, and daemon feedback for graceful failure handling.
AI-written summary · about
chapters (12)
- 0:00Start
- 0:43Goal of a Networking Experiment
- 1:15What Makes an Experiment Complex
- 1:55The Goal of a Networking Experiment
- 4:42Key Highlights of the Magi Tool
- 5:52Agent Directive
- 6:38Event Streams
- 7:20Orchestrator
- 8:40Daemons
- 9:11Case Studies
- 10:30Feed Lab Feedback Loops
- 13:42Key Takeaways
related talks
43mNext generation config mgmt: autonomous systems
16mAutomatic Parallelization of Software Network Functions
31mTaming Your Data Networks with the Power of NetDevOps
11mOrchestrating thousands of bots from the cloud
24mContaining Agent Chaos
16mReducing Operational Barriers to Model Training
15mEngineering Production-Grade Multi-Agent Systems in the Cloud
28mMagma Abstractions and Interfaces A Systems Approach to Access Networks