Lorenz and Little: How Much Does Your Tail Cost?

A deep dive into quantifying the cost impact of tail latency using the empirical Lorenz Curve and Little's Law. By calculating how much each latency percentile contributes to mean latency, you can directly measure its contribution to system concurrency and therefore infrastructure cost. The post includes a Python implementation and an interactive calculator, showing that in typical services the p99 tail can account for over 50% of mean latency and concurrency.