Workday's Next Generation Private Cloud
summary
Workday runs an OpenStack private cloud at ~1M+ physical cores on ~16,000 nodes across 5 data centers (87 clusters ≈300 nodes each) and migrates from Mitaka→Victoria, OpenContrail→pure L3 Calico with BGP, Chef→Ansible, custom images→Kolla, Sensu/Wavefront→Prometheus/Grafana, Jenkins→Zuul and CentOS7→CentOS8 Stream. They rebuild pipelines as a gated Zuul flow (source→build→90‑minute nested VM tests), add per‑cluster DNS, IPAM, a VM‑boot health checker, Prometheus 2 OpenStack SD and singleton container coordination; forklifting clusters reduces toil and initial upgrades complete.
AI-written summary · about
chapters (8)
- 0:00Introduction to WPC
- 1:47Architecture and operations
- 5:25Managing maintenance windows
- 7:53WPC5 and gated development
- 11:18Deployment strategies
- 14:05Internal infrastructure tools
- 18:15Technical improvements in WPC5
- 27:33Q&A and closing remarks
related talks
19mZuul CI for Kubernetes on the Workday Private Cloud
36mAdventures in Running a Private Cloud
52mLet’s build a private cloud – how hard can it be?
21mBack to the original mission: an open cloud operating system
31mServers Are Not the Hardest Thing to Orchestrate
29mLessons Learnt Operating Multiple Cloud Foundry on OpenStack Deployments
41mOpenShift: a Kubernetes Story in an Enterprise Environment
59mCloud Native Home Lab