KubeFM Super-Scaling Open Policy Agent with Batch Queries, with Nicholaos Mouzourakis
May 13, 2025
Nicholaos Mouzourakis, a Staff Product Security Engineer at Gusto, dives into the intricacies of scaling authorization within Kubernetes using Open Policy Agent (OPA). He explains how traditional approaches fall short in microservices and shares his team's journey optimizing OPA performance through batch queries for impressive efficiency gains. Nicholaos also highlights surprising interactions between Kubernetes CPU limits and Go's performance, alongside deployment strategies that ensure smooth operations in production. His unique transition from the gaming industry enriches his insights.
AI Snips
Chapters
Books
Transcript
Episode notes
Nicholaos' Tech Journey
- Nicholaos Mouzourakis shared his journey from game development to cloud native engineering.
- He worked on titles like Deus Ex Mankind Divided and Hitman Sniper before moving into Kubernetes authorization.
Limits of Traditional Authorization
- Traditional authorization often fails in microservices due to fragmentation and complexity.
- Code-driven approaches tightly couple logic to services; data-driven approaches like RBAC lack needed flexibility.
OPA’s Kubernetes Fit
- OPA’s declarative, logic-based policies fit naturally with Kubernetes’ JSON/YAML data flow.
- Its lack of side effects and guaranteed termination suits efficient policy evaluation at scale.