Two sides of the platform coin with operator's and customer's voices
summary
New York Times platform engineers build a multi-tenant Kubernetes cluster that delivers namespace-as-a-service and automatic GitHub repos plus CI/CD so teams get a production-ready deployment in minutes not weeks. A content team moves a GraphQL subgraph serving 400 to 1000 requests per second onto the platform, uses hosted previews and autoscaling, discovers a networking hop that adds latency, flips to internal URLs to remove the bottleneck, and rolls that network policy out to other teams.
AI-written summary · about
chapters (16)
- 0:00Platform story introduction
- 1:05NYT mission and products
- 2:33Localized API challenges
- 5:22Design goals and standards
- 7:28Developer needs vs platform focus
- 9:52Visibility and data modeling
- 11:58Runtime choices and multi-tenant K8s
- 14:47Service creation workflow
- 16:33GraphQL subgraph migration
- 18:17Internal platform benefits
- 20:03Preview environments iteration
- 21:05Load testing and observability
- 22:33Feedback loops in development
- 23:14Evolving tech standards
- 23:55Iteration as product mindset
- 24:57Platform success principles
related talks
44mIs Platform Engineering a hype? Let's build a platform together!
37mBuilding a Managed Platform While Maintaining a Good Developer Experience
33mBlueprints of Innovation: Engineering Paved Paths for a User-friendly Developer Platform
31mThe 20 Year Platform – bringing together Kubernetes, 12-Factor and Functions
23mMore Power, Less Pain: Building an Internal Platform with CNCF Tools
39mDesigning & Securing Multi-Tenant Runtime Environment at the New York Times
40mThe Elephant in the Kubernetes Room: Team Interactions
23mScaling Kubernetes in Regulated Environments Through Platform APIs