Hello friends!  Today we will be discussing the histogram chart and the value that it provides for network engineers when looking to understand how network bandwidth is being utilized in their network’s estate.

We will be covering:

• What is a histogram?


•  What insights does it provide?


• How and when can you use it in capacity planning?


• What best practices can be applied to its use?

When viewing network performance data from the perspective of the capacity planner, your goal is to first identify the highest (or lowest) utilized links in order to “right-size” the amount of link capacity that you require to support the application traffic used by your business.  If you over-provision, your applications will have plenty of bandwidth available to them and less congestion but you will be spending money on bandwidth that doesn’t get used.  If you under-provision, your monthly bill will be less, but your applications and their users will probably suffer from the lack of necessary bandwidth, causing congestion (think traffic jam) and the associated slowness that it inflicts on application response time.

What is a histogram?

Histogram charts help you quickly understand how your network bandwidth utilization is distributed over the course of the time period you are considering.  For simplicity’s sake, let’s consider “yesterday’s data”.  Looking at the trend chart for yesterday, we see time-series data represented using a line graph.  In this case, we are measuring bandwidth at 60 second intervals.  While we can see a lot of useful information from the trend chart – average, min/max values – it is also useful to see how much time is spent in various ranges or “buckets”.  That’s where histogram charts come in.

Standard trend chart showing network bandwidth utilization over time

What insights does it provide?

When we represent the same trend data using a histogram chart, we must first take all of the individual data points and place them into buckets, based on ranges of values.  SevOne allows you to specify how many buckets you want to use.  If you select 5 buckets, it will subtract the minimum value from the maximum value and then divide by 5 in order to determine the distribution bucket sizes.  This method allows SevOne to show the distribution ranges quickly for any indicator’s metric value set.  Before the chart is rendered, SevOne will sort all of the time-series data points into the range buckets allowing the user to understand what percentage of values were spent in each bucket.  With bandwidth utilization metrics this equates to how much time during the day application traffic was consuming the available bandwidth within the range of the various buckets.

Looking at the chart below, we can see a right-skewing chart, where most of the values (and time) is spent in the left-most bucket – 60% of the network bandwidth utilized was within 25.21 – 36.7 percent of available capacity.  To the right of the chart, we see 8.33% of the traffic (and time) was spent in the highest utilization range between 71.19 – 82.69%. 

Histogram output of the same data, showing a bucketized distribution of the datapoints during the same timeframe

How and when can you use it in capacity planning?

Histogram charts provide a quick at-a-glance view of how your network bandwidth is being used over time.  The capacity planner can quickly view the highest bucket range to see the percentage of time that network bandwidth was consumed during the business day providing “time in the red zone” insights.  If your red-zone is 90 – 100 %, then you should consider upgrading your network bandwidth capacity so that your application performance (and users) suffer due to network congestion.  If your red-zone is 60 – 70 %, then you may want to consider reducing your network bandwidth capacity to save on monthly bandwidth costs.

What best practices can be applied to its use?

SevOne provides the report author with the ability to filter out non-work hours or non-interesting times so that the data that is distributed into your histogram buckets is the most interesting data to the capacity planner and that the distribution considered is most impactful in understanding trends in your network’s bandwidth usage.  As a best practice, you should consider using this feature, defining your device work hours (or custom hour ranges) so that you filter out “noise” data points that aren’t useful.

I hope you have found this article useful in understanding the value that histogram charts can provide when planning for capacity with SevOne.