Moving from 1 to N regions: an open retrospective
summary
Project turns New Relic from a one-region rollout into an N-region pattern by changing 500 services across ~50 teams to run anywhere, avoiding doubled ops. The stack standardizes on containers + DI service discovery with credentials delivered via URL-like environment variables, uses Terraform and Vault for infra, targets IBM Frankfurt as the first sovereign region, and runs iterative test buildouts and teardowns followed by a limited pilot phase with customers. Result: 95% in container fabric and an order-of-magnitude fewer engineering hours.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:13IBM and New Relic
- 4:21Why was it so painful
- 9:32Strong interfaces
- 11:20Service discovery
- 20:08Execution plan
- 21:50Ramp ups
- 24:29Moving goal posts
- 28:48Communication is hard
- 31:00Local maximums
- 32:57piggyback
- 34:17pilot phase
- 35:21summary
related talks
37mThe Nth Region Project: An Open Retrospective
27mBuilding Multi-region CI/CD Pipelines with a DevOps Mindset
23mDay-3-Operations: New Region and New Services
36mPractical Cloud Native: What Works, What Doesn't
11mBuilding a system that never stops: New Relic at scale
50mEvolution of a Microservice-Infrastructure
01h 02mTransforming IT pros to DevOps gurus: Ssecure your new tech stacks
17mMulti Region Deployments with AWS CI/CD and Terraform