Migrating 5G Core to the Edge, Wisely and Confidently with Kubernetes
summary
Compares three migration stages and then measures real effects: etcd sync latency hurts large-edge clusters and increases cloud egress costs; suggests using CRDTs or operation-based replication to avoid strong-consistency bottlenecks. Benchmarks four edge Kubernetes distributions and finds K3s advantage while KubeEdge shows the highest response times. Demo runs IP-camera load on a 32-core EC2, 100GB RAM node and scales from 10→100 replicas, recording pod startup and API latency to show persistent-storage and scheduler limits and where proactive ML-based autoscaling is needed.
AI-written summary · about
chapters (7)
- 0:00Intro
- 0:55Migrating SG Core to the Edge, Wisely and Confidently with Kubernetes
- 6:58ETCD Performance
- 10:56Pod-scheduling strategies
- 14:035G Core Open-Source Projects
- 16:21Demo
- 20:42Takeaways
related talks
36mEdge Computing in Kubernetes: Best Practices for the Future
34mManaging Edge Nodes With “KubeEdge”
33mKubeEdge: From Fixed Location to Movable Edge, Latest Updates and...
36mEdge Computing and Kubernetes BOF
01h 16mIntro + Deep Dive: Kubernetes IoT Edge WG
20mKubernetes at the Edge
22mKubeEdge: From Fixed Location To Movable Edge, Latest Updates And Future
20mKubernetes at the Edge for Telcos
