Building and Managing Kubernetes with Kubernetes
summary
eBay runs a fleet management system that models the entire data center as Kubernetes CRDs and operates it with controllers that actually provision bare metal, allocate IPs, manage DNS and remediation, and build clusters via Salt using git commit IDs for Salt masters. The estate includes 50+ production clusters, ~160,000 pods on ~30,000 hosts and 15 edge clusters. The system supports declarative rolling transactions that stage upgrades in buckets and a VM provider (virtlet) so non‑container workloads run atop Kubernetes.
AI-written summary · about
chapters (11)
- 0:00Intro
- 0:29Agenda
- 1:11eBay's k8s deployments
- 3:20Fleet management system
- 6:04Architecture View (single Availability Zone)
- 10:08Bootstrap and Model the Datacenter
- 11:56BM Provisioning System
- 18:30Build & manage k8s with k8s
- 20:10Building & managing k8s with salt operators
- 22:59GitOps for devops
- 24:46Features inspired by k8s Schedule
related talks
28mBuilding and Managing Kubernetes with Kubernetes
38mModel and Operate Datacenter by Kubernetes at eBay
32mRunning eBay’s High-Performance Workloads with Kubernetes
01h 01mDeploying Kubernetes, a Container Cluster Manager
36mOperator Framework to Manage Production Grade Stateful Workloads
32mRunning Kubernetes in a Manufacturing Line – What Could Possibly...
37mRunning Kubernetes in a Manufacturing Line
27mContinuous Integration at Scale on Kubernetes [B]