.NET Architecture & System Design Interview Question #21
What are p50, p95, and p99 latency?
Architecture Fundamentals, Scalability, Availability & Performance Junior Beginner
Quick Interview Answer
Percentile latency shows how response times are distributed. p50 is the median: 50% of observations are at or below that value. p95 and p99 describe the response time below which 95% and 99% of observations fall.
Detailed Explanation
Percentile latency shows how response times are distributed. p50 is the median: 50% of observations are at or below that value. p95 and p99 describe the response time below which 95% and 99% of observations fall.
High percentiles expose tail latency that an average can hide. Service-level objectives often use a percentile, for example requiring p95 latency below a defined threshold.
Code Example
Example:
p50 = 80 ms
p95 = 240 ms
p99 = 700 ms
99% of measured requests
completed in 700 ms or less.