CalcChef
Search tools (e.g. 'calc', 'converter')... (Press '/' or 'Cmd+K')

Server Response Time Percentile Estimator

Estimate p50, p95, and p99 server response times from a CSV of latency samples. See the percentile distribution, summary statistics, and detect outliers above 2 standard deviations.

Loading Tool...

Applications

SRE and platform engineering for evaluating API and web service performance against SLOs, triaging latency spikes in observability dashboards, comparing response time distributions across deploys, and validating load test results from tools like k6, JMeter, or Locust.

Percentile Estimation

Sorts the input samples ascending and applies the NIST primary method (C=1) with linear interpolation between adjacent ranks. The p-th percentile equals the value at rank (p/100) * (n-1) in the sorted array, interpolated linearly when the rank falls between two samples. Outliers are flagged as any sample more than 2 standard deviations above the arithmetic mean.

Fun Fact: Google's research paper that popularized p99 latencies in SRE culture ('The Tail at Scale', Dean & Barroso, 2013) showed that even tiny fractions of slow requests can dominate user-perceived response time at scale — making tail latencies a far better health signal than averages.

Related Tools

View all tools