Five Years of Cloud Foundry at Rakuten
summary
Five years running Cloud Foundry on 5,000 VMs with a team of seven, Rakuten describes concrete solutions: placement pools and a three DEA/router groups trick to expose dev/stage/prod as logical environments inside one CF, a single CF deployed across vSphere and OpenStack by splitting manifests, and a log/metrics pipeline that decouples producers and consumers using BOSH, syslog, Kafka with per-application Kafka topics, ELK/Graphite/Riemann consumers, and archive to blob storage. Future work targets VM autoscaling, multi-datacenter deployment, TLS/HTTP2 and automated certificate provisioning including Let’s Encrypt.
AI-written summary · about
chapters (12)
- 0:00Rakuten company overview
- 0:37Platform history and lessons
- 2:26Operational best practices
- 5:31Designing for failure
- 6:26The need to update
- 7:00Infrastructure provisioning
- 9:27Managing environments
- 11:36Testing and validation
- 13:47Log aggregation and monitoring
- 17:54Internal tools
- 18:25Future roadmap
- 21:01Contributions and feedback
related talks
19mKeynote: 5 Years of Cloud Foundry at Rakuten
33mCloud Foundry Weekly: Live at CF Day Germany
38mWhat Does It Take to Run Cloud Foundry at Enterprise Scale?
17mDeployments We Can Believe In
48mCloud Foundry - your PaaS on OpenStack
21mKeynote: Cloud Foundry at Rabobank after 10 months of real use
31mFive Years Delivering World-Class Business Critical Apps on Cloud Foundry
23mCloud Foundry and OpenStack