What are p50, p95, and p99?

Latency percentiles describe the distribution of response times. p50 is the median, meaning approximately half of requests are faster and half are slower. p95 means approximately 95% of requests are at or below that latency, while the slowest 5% are above it. p99 represents approximately 99% at or below that latency and highlights tail latency more strongly.

Key Takeaways
  • p50 is the median.
  • p95 describes the upper tail more than the median.
  • p99 highlights severe tail latency.
Knowledge Check

What does p95 latency represent?