Building for Global Engineering Platforms - Let's make an Operator
summary
Shows a concrete platform pattern: define API-first CRDs and implement them as operators (using Go, Operator SDK, Helm, Kubernetes) so cloud identities and permissions are abstracted by an IM operator pattern and reused by resource operators like an DynamoDB operator and S3 operator. Team built controllers that create Kubernetes service accounts, sync provider identities in status, and enforce regional write routing via Istio plus AWS Global Accelerator and DynamoDB multi‑write. On one client this cuts workload‑to‑prod time from months to hours across 12–100 clusters and multiple clouds.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 3:14What is platform engineering
- 10:58GitOps and missing API layer
- 18:29Global platform challenges
- 21:22Latency and DynamoDB multi-region writes
- 30:42Q&A
- 35:03Workshop setup
- 41:02Building IM operator 1
- 48:34Building IM operator 2
- 52:25Building IM operator 3
- 1:06:35Building IM operator 4
- 1:18:24Extending operator for cloud providers
- 1:23:13Q&A on maintenance
- 1:30:49Final Q&A
related talks
44mPlatform Engineering: Raise the Level of Abstraction using Kubernetes Operators
45mKubernetes Operators: Operating Cloud Native services at scale
36mEssential Patterns For Designing And Implementing Your Operator
33mAutomation with Kubernetes – Creating an operator - Johann Gnaucke
31mKubernetes Operators within Reach of Any Developer
24mAutomating Stateful Apps with Kubernetes Operators
35mPanel: State of the Operators: Hubs, Frameworks, SDKs, and Beyond
29mRage Against the API-Machinery: Writing an Operator for Production