Access control in message-driven systems
summary
Claims that a token-based, OPA-backed model enforces authorization across asynchronous flows by creating per-message tokens with unique message ID and token bound to payload hash so consumers can verify non‑tampering and reject replays. Enforcement occurs at command and event handlers via an OPA sidecar decision point or colocated auth service, and high volumes push auth responsibility toward each pod or self‑signed tokens. Event sourcing needs long or permanent expiry and system roles for deleted users.
AI-written summary · about
chapters (14)
- 0:00Access control intro
- 1:06Zero trust for messages
- 5:00Zero trust principles
- 8:13JWT tokens in sync
- 11:04Identity loss in async
- 13:12Message types overview
- 14:38Command/event enforcement
- 17:30Open Policy Agent
- 21:07Zero trust for events
- 26:34Scaling tokens
- 27:39Token expiration in events
- 29:46User identity in events
- 32:36Summary and implementation recommendations
- 35:25Q&A
related talks
40mAccess control in message-driven systems
24mExpanding beyond authentication with emerging authorization standards
29mAuthz as a dev workflow
23mAuthentication and Authorization in Micro Services
27mSecurity-by-construction: How to weave authorization into modern app stacks using Open Policy Agent
56mDistributed Authorization for Microservices Powered by Kubernetes, Istio and Open Policy Agent
14mAPI access control strategies beyond JWT bearer tokens
19mDynamic Authorization and Policy Control for Your Kubernetes Cluster