Skip to content

Resources

Want to read our latest whitepapers or Tech Tips? Or check out our library of data sheets and case studies? Netcordia's Resource Center is constantly updated with new materials, so you can always stay on top of our products, solutions, and services.

Tech Tips

Network Analysis Tip # 49 - 95th Percentile

Why is this important?

Network administrators need to have some way to identify high link utilization. Average and peak link utilization either under state or overstate the real link utilization. How should true link utilization be determined? Is more bandwidth is needed or should the traffic be subject to bandwidth shaping? The 95th percentile calculation provides a good measure of utilization.

The algorithm is as follows: collect all the data samples for a period of time (commonly a day, a week, or a month), then sort the data set by value from highest to lowest and discard the highest 5% of the sorted samples. The next highest sample is the 95th percentile value for the data set. The link is running at that utilization value or higher for 5% of the time over which the samples were collected. On a minute-by-minute basis 5% over 24 hours is 72 minutes out of 1440 minutes. Therefore, the link was running for 72 minutes (depending on the sampling period) at, or above, the 95th percentile utilization value.

In the charts below for a theoretical T1 link, we see peak and average utilization of 1.32Mbps and 0.27Mbps, respectively. There are 24 samples, so 5% would be one sample (rounding to the nearest integer value) to ignore. The 95th percentile utilization is the value of the next highest value, or 0.75Mbps.

Hourly Link Measurments GraphSorted Hourly Link Measurments Graph

Manual determination

Manual data collection isn’t really an option, even for a single interface. The collection should be on a regular basis for the entire period (day, week, or month) and that’s not a job for a person.

Automatic determination

At a minimum, you’ll need an automated mechanism to regularly collect interface performance statistics and record them. Several open source programs, such as MRTG, can automate the data collection. You’ll then need to perform the calculation and identify links that are nearing full capacity for long periods of time. These are the candidate links on which to configure QoS for bandwidth shaping, depending on the applications in use, or for increasing the bandwidth of the link if mission critical applications are not getting enough bandwidth. NetMRI automatically collects the necessary interface statistics and performs the 95th percentile calculation.

Further reference:

Other definitions and use of 95th percentile:
http://www.sitespecific.com/about/95th_percent.asp
http://www2.arnes.si/~gljsentvid10/pct.html

Back to Tech Tips