Edge Computing and Kubernetes BOF
summary
Discussion catalogs concrete edge use cases and hard numbers: video caching and IoT gateways at tens of thousands of sites, healthcare pre-positioned imaging caches, manufacturing anomaly detection, cloud robotics and IPTV multicast. Architecture choices land on one cluster per site for isolation, with alternatives (central master, multi-rack proximal clusters) rejected for partition risk. Tooling experiments include a demo using CoreDNS + anycast+BGP to route to nearest cluster, use of multi-cluster registries, and concerns about 20 ms 5G control loops, 500 ms video tolerances, and finite edge resource eviction/priorities.
AI-written summary · about
chapters (9)
- 0:00Start
- 2:02Use Cases
- 4:28Defining Edge Computing
- 5:37Why Not Run Everything in the Public Cloud
- 26:58Multi Cluster Load Balancing
- 28:10Proximity Routing
- 28:34Load Balancing
- 31:39Resource Scarcity and Auto Scaling
- 32:25The Difference between Cloud and Edge
related talks
36mEdge Computing in Kubernetes: Best Practices for the Future
01h 16mIntro + Deep Dive: Kubernetes IoT Edge WG
23mMigrating 5G Core to the Edge, Wisely and Confidently with Kubernetes
39mIoT Application Running on KubeEdge + ARM Platform
34mManaging Edge Nodes With “KubeEdge”
39mIoT Application Running on KubeEdge + ARM Platform
20mKubernetes at the Edge for Telcos
34mKubernetes IoT and Edge Working Group