How to calculate the 95th percentile

How to calculate the 95th percentile

The 95th percentile is a standard statistical measurement used to interpret performance data, particularly in traffic generation on an interface. It represents the value below which 95% of the data points fall, effectively ignoring the highest 5% of values to reduce the impact of short-lived spikes. This ensures that occasional high-usage moments don’t unfairly impact overall calculations, providing a more accurate and fairer representation of sustained usage compared to other statistical metrics like the average or median. 

Steps to calculate the 95th percentile

Follow the sequence of steps below to calculate the 95th percentile:
  1. Gather the dataset with numerical values recorded over a defined period.
    Suppose we have the following 20 data points (e.g., the CPU usage recorded at 20 different time intervals):
    90, 33, 83, 6, 93, 41, 49, 24, 53, 63, 81, 41, 33, 49, 87, 36, 46, 29, 119, 116
  2. Arrange all the values in ascending order.
    6, 24, 29, 33, 33, 36, 41, 41, 46, 49, 49, 53, 63, 81, 83, 87, 90, 93, 116, 119
  3. Calculate the index:
    1. P is the position in the sorted list.
    2. N is the total number of data points.
      P=N×0.95
      P=20×0.95=19
  4. If P is not a whole number, round it up to the nearest integer.
    Here, P=19 which is a whole number.
  5. Locate the value at the Pth position in the sorted dataset.
    The 19th value in the sorted dataset is 116.
So, the 95th percentile of CPU usage is 116%.
The unit of the 95th percentile depends on the measurement unit of the monitored parameter. For resource utilization metrics like CPU, memory, and disk usage, it is represented as a percentage (%). 

    • Related Articles

    • How does Site24x7 calculate downtime of a website?

      When a website availability check from its primary monitoring location fails, Site24x7 waits up to 15 seconds to get a response from the website (15 seconds is the default connection timeout period). If there is no response within 15 seconds, ...

    • How to calculate Mobile APM screen load times

      Mobile APM provides load time metrics to help you understand how long it takes for your mobile app screens to load. These load times are calculated by capturing start and end timestamps during specific lifecycle callbacks. Load times for native ...

    • How does Site24x7 Kubernetes monitor calculate the memory and CPU utilization for pods?

      Site24x7 Kubernetes Monitor uses the kubelet API and kube-state-metrics API to collect the metrics related to memory and CPU utilization. For memory utilization, the calculation works as follows: Working set size (WSS) memory usage ...

    • How does poll strategy setting work?

      When configuring thresholds for performance metrics, you can set strategies depending on which the monitoring agent will send alerts during a threshold breach. Here are the available options: Poll count Poll average Time duration Average time Poll ...

    • Global Benchmark reporting with multiple monitor types in monitor groups

      Global Benchmark Report lets you quickly analyze and benchmark availability and response time trends of Websites, REST API endpoints, and REST API Transaction monitors for a selected time period. You can use this report to measure your various ...