Skyline: A Cloud Centric Internet Monitoring Engine
summary
Cloud deployment monitors public‑Internet faults with three concrete tools: BiProp uses P4 internet gateways to reply to inbound probes and achieve agentless, 100% cloud‑subnet coverage; FlexPass steers MPLS/segment‑routed probes to measure primary and backup AS paths and detect repairs; Flow Hunter uses optical splitters and passive TCP analysis to surface reset storms and per‑tenant L4–L7 failures. Skyline runs in production two years, flags over 2,000 incidents in 2025, and reports >99% recall and ~97% precision on verified issues.
AI-written summary · about
chapters (8)
- 0:00Skyline project intro
- 2:02Cloud internet monitoring challenges
- 3:25Traffic direction with BiProp
- 4:49Incident lifecycle with Flex Pass
- 6:55Tenant monitoring with Flow Hunter
- 9:04Skyline architecture overview
- 9:48Evaluation and deployment results
- 13:00Operational lessons and conclusion
related talks
46mLessons learned from monitoring a large production network
29mFinding the Needle in a Haystack: Tenant-Level Network...
32mIntelligent cloud networking: Create simplicity, remove complexity (PRT214)
16mNetwork Flow Data in the Cloud
49mVisibility Without Borders: NETSCOUT to help migrate apps to
04mDay 2
15mBuilding a Network Telemetry Platform
54mAssuring and securing AWS migrations
