Extending Cloud Foundry at Scale - Addressing the...
summary
Comcast builds CFCR clusters to host non-12-factor, stateful workloads (Kafka, CockroachDB) using BOSH and Concourse pipelines, Vault and GitHub for secrets and manifests, deployed across on-prem vSphere and AWS. Each statefulset uses 300 GB PVC per replica so data survives pod restarts; clusters follow a 3–4 week release cadence. Proof it works: production-ready UX is identical across clouds, Kafka clusters run in the platform, and a Kubernetes patch was rolled out 24 hours before its public release.
AI-written summary · about
chapters (11)
- 0:00Introduction and team background
- 3:12Cloud Foundry at Comcast scale
- 7:26Gap in 12-factor app support
- 11:26Persistent volumes in CF CR
- 15:19Toolset for CF CR deployment
- 19:35Deployment pipeline overview
- 21:17CF CR on-prem and cloud
- 23:03Team benefits and challenges
- 25:34Turnkey services for developers
- 27:01Open source contributions
- 28:27Q&A session
related talks
06mHow CFCR tackles Day 1& 2 Ops for Production Kubernetes
30mLiberate Your Cloud Foundry Installation From Being Constrained to a Single Cloud
19mPaaS on Kubernetes: Journey of Making Cloud Foundry Kubernetes Native
01h 02mCloudFoundry or Kubernetes/Kubo: Choosing the Right Platform for the Job
46mWhat's Cloud Foundry and why you want to use it
30mCF^3 - Using Kubernetes as the Cloud Foundry IaaS
01h 03mDeep Dive into Pivotal Cloud Foundry 2.0
31mCloud Foundry Container Runtime: BOSH-Powered Kubernetes