← ⚙️ Infrastructure at Scale
Core
Saturation and utilisation targets: why systems fall over before they are full
A service can run at 70 percent CPU with a p99 four times its median and no component reporting a fault. Little's law and the queueing curve explain why, and they turn a headroom target from a habit into a number you can defend.
a free account opens the core tier · no card
RELATED CONCEPTS
PRACTICE THIS IN REAL QUESTIONS
Cloud Platforms & ArchitectureYour service autoscales on CPU and still falls over during traffic spikes. Why, and what would you scale on instead?→AI & GPU InfrastructureProduct wants to raise the context limit from 8k to 128k tokens. What happens to your infrastructure?→DevOps System Design & ArchitectureDesign a disaster recovery plan for a 5PB data platform with a four-hour recovery objective.→DevOps System Design & ArchitectureAn e-commerce platform takes 10x traffic during flash sales and the database locks up. Fix it.→Incident Response & Production DebuggingYour p99 latency is three seconds while the average is two hundred milliseconds. Where do you look?→AI & GPU InfrastructureYour cluster has three GPU generations because that is what you could buy. How do you schedule against it?→