Chaos Engineering Practice Under Ultra-large-scale Cloud Native Edge Computing
Talkat KubeCon + CloudNativeCon Europe 2025 · on Thursday, 2025-04-03· at 16:45· Level 1 | Hall Entrance S10 | Room A·
summary
KubeEdge claims stable operation at 100,000 edge nodes and management of over one million pods by using a multi‑instance cloud‑core / edge‑core design, an efficient bi‑directional cloud‑edge messaging channel, and an incremental data push model that avoids surge loads on the API server. Scalability tests use HMAK and cluster‑node tools, report P99 read latencies under 1s and pod‑standup P99 ~4ms, and show edge‑core memory ≈70MB; ChaosMesh is used to inject faults and validate reliability.
AI-written summary · about
chapters (13)
- 0:00Chaos Engineering at scale
- 1:09Edge computing growth
- 4:03KubeEdge project overview
- 7:35Service level indicators
- 9:23Scalability testing challenges
- 10:30Large-scale cluster tools
- 12:40Deployment model config
- 14:26Test results and SLO compliance
- 16:56Network and data strategies
- 17:17Cloud-edge message channel
- 18:45Incremental data transmission
- 20:31Future edge testing
- 21:15Q&A
related talks
20mCentralized, distributed or both? - Architectures for Kube...
43mSystems @Scale Summer Q&A Week 2
42mThe Challenges and Solutions of Open Edge Infrastructures
55mChaos Engineering for Cloud Foundry Platform
31mAn Adventure in Multi Cloud with Kubernetes and Distributed Data
01mEdge Computing
21mKeynote: CERN Experiences - Ricardo Rocha & Clenimar Filemon
39mPracticing Cloud-Native Resliency Is a Challenge: Decoding It with Chaos...