Stop Rate Limiting - Capacity Management Done Right

Explore capacity management techniques in cloud computing and microservices environments through this 42-minute conference talk from Strange Loop. Learn why rate limiting can be problematic and discover a superior alternative: concurrency control. Dive into queuing theory and Little's Law to understand the limitations of rate limiting and the benefits of controlling concurrency. Examine an elegant implementation that's easy for clients to grasp, and investigate an extension allowing for decentralized enforcement and adaptive algorithms. Gain insights into managing elastic origin capacity, fluctuating client populations, and varying usage patterns. Equip yourself with essential knowledge for protecting your services and ensuring fair resource allocation among clients in distributed systems.

Introduction
API Management Gateway
Why Rate Limiting Doesnt Work
Queueing Theory
Demos
Queue Theory
Limit Concurrency
throughput vs capacity
TCP congestion avoidance
Adaptive capacity management
How capacity gets managed
When to admit a request
When to borrow quota
Summary