Highly Available Cloud Foundry Deployment for Allstate
summary
Allstate builds an on‑prem multi‑region Cloud Foundry with separate Cloud Foundry foundations per AZ, running on VMware across three AZs per region (37 servers, 756 CPUs, ~14,000 GB RAM) and using Concourse plus an open‑sourced ops‑manager CLI. Applications require two instances per AZ (internet‑auth patterns deploy 16 copies) and use automated blue‑green deploys via conveyor (deployaDoctor) that handle auto‑routing and rollback. The platform targets 99.9% uptime, supports live maintenance (PDU swaps) with zero user impact.
AI-written summary · about
chapters (9)
- 0:00Introduction and goals
- 2:00Architectural building blocks
- 3:36Design principles
- 5:49Building the infrastructure
- 9:21Operational benefits
- 13:09Application architecture
- 17:12Automated deployments
- 20:16Successes and resources
- 23:40Q&A session
related talks
40mThe Rocket Science of Deploying Cloud Foundry in Multi-Cloud Environments
38mWhat Does It Take to Run Cloud Foundry at Enterprise Scale?
05mFour Levels of High Availability in Cloud Foundry
16mDeployments We Can Believe In
17mDeployments We Can Believe In
31mPut Your Money Where Your CF Is
29m#BuildingTheFuture: Keeping our Dutch Feet Dry: Setup and Operation of a...
35mCloud Foundry for Everyone!