APIs beyond datacenters: Architecture and Best Practices
summary
Built a centralized, customer-facing API that provides a single endpoint to manage PSK across three regions, using a stateless API server, MongoDB for persistent state and RabbitMQ for messaging to agents that translate API models into Gardener calls. Deployment runs API, MongoDB and RabbitMQ in separate clusters; MongoDB uses the MongoDB Enterprise operator to stretch data, and agents consume RabbitMQ queues to sync and report status. Demo shows project and shoot creation works; upkeeper failover and cross-region IP failover are still being hardened.
AI-written summary · about
chapters (7)
- 0:00Introduction to APIs beyond datacenters
- 1:46Current system limitations and goals
- 4:58Architecture design of the new API
- 8:49Evaluating stretch vs multicluster setups
- 13:03Current multicluster setup details
- 15:30Demo of the API in action
- 21:03Q&A session with audience
related talks
34mAPIs Without Borders: Exploring the world of Locationless API Management
37mAdvanced API Management & Operations, A Developer & Customer-Focused Approach
48mTransforming a monolith to an API platform
26mLessons learned from Re-architecting an API platform
18mThe architecture of an API platform
23mDecentralized API Management for Agile Enterprises
59mBuilding resilient event-driven architectures
38mAdvanced API Design for Scalable Distributed Systems