Highly Available Cloud Foundry on Kubernetes - Vlad Iovanov, SUSE
summary
SUSE Cloud Foundry is containerized and run on Kubernetes using a custom fissile tool that turns BOSH releases into Docker images and Helm charts. High availability relies on Kubernetes primitives: stateful sets for clustered services, readiness and liveness probes for active/passive promotion, and services for load balancing. A chaos script kills pods every three minutes over three hours (total 54 pod kills) while the system records 99.5% application availability and 9.8% API failures.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 0:45Cloud Foundry Components
- 1:50Critical Components
- 3:20Containerized Cloud Foundry
- 5:33High Availability
- 7:53Kubernetes
- 8:55Kubernetes Services
- 10:20Stateful Sets
- 13:00Probes
- 15:23Help
- 18:40Overview
- 22:25Conclusion
- 23:58Loss of entire host
- 24:36Deployment speed
- 25:48Kubernetes as a host
- 27:31Upgrades
- 28:10Secret
- 28:52Why Kubernetes
- 29:58Containers in Containers
- 31:37How to keep helm on par with Bosch
related talks
38mLet's Run Cloud Foundry on Kubernetes
31mCloud Foundry Container Runtime: BOSH-Powered Kubernetes
06mLightning Talk: The Latest on How SUSE is Bringing Cloud Foundry and Kubernetes...
04mSymbiosis Update: The Latest on How SUSE is Bringing...
23mCloud Foundry and Kubernetes: A Peanut Butter & Jelly Application
01h 02mCloudFoundry or Kubernetes/Kubo: Choosing the Right Platform for the Job
05mFour Levels of High Availability in Cloud Foundry
48mCloud Foundry - your PaaS on OpenStack