cancel
Showing results for 
Search instead for 
Did you mean: 

Analytics: Understanding Time Charts and Metrics

This article is part of a series, beginning with Analytics Application - Concepts and Metrics Explained

 

The Analytics Application included with Services Director displays a number of different types of metrics. The process by which metrics are generated and displayed varies between different types of graph, but the metric definition itself remains constant.

A timechart is any graph where the x-axis is time-based. The purpose of a time chart is to show how a metric varies over a time range. The start and end points for the x-axis are determined by the selected time range. The following sections describe each graph types, and how the various metrics are generated for each.

Time buckets

The time range represented in the timechart is subdivided into smaller time buckets (or 'spans'), such that the chart shows an appropriate level of detail for the sample. 

For example, if the timechart is showing values over an hour, it is not useful to subdivide the chart into 30 minute buckets, as the chart would consist of only two points. Nor is it useful to subdivide the chart into millisecond buckets, as the resulting graph would have too many data points to plot sensibly on most displays.

The Analytics Application uses an algorithm to choose the bucket size, which uses the smallest time bucket available that does not result in more than 100 buckets. The smallest time bucket available is 1 second, the largest is 10 days, with 18 incremental bucket sizes between.

When timechart calculations are performed, results are split by time bucket. This diagram shows of an example how transactions being mapped to four time buckets:

an2-buckets.png

It is important to note that the duration of the time bucket is signficant, as it determines how many (and which) transactions fall into a given time bucket. This can give rise to apparent fluctuations in timechart values as you zoom into smaller and smaller time periods, particularly where data is sparse or the metric is "split" by a field with numerous categories (for example, when the Location category is set to display Countries). These fluctuations are artefacts of the changing resolution of the time buckets, and are notcalculation errors.

 

Requests / second

In the context of a timechart, Requests / second is calculated as follows:

 

  • Filter away all connection-based records.
  • Apply any further filters (from the Component Filter and/or the Extended Filter).
  • Count the number of requests in each time bucket, also applying any selected splits at this point.
  • Divide the count of requests in each time bucket by the time bucket size (in seconds).

 

Connections / second

In the context of a timechart, Connections / second is calculated as follows:

  • Filter away all records outside the time range.
  • Filter away all request-based records.
  • Apply any further filters (from the Component Filter and/or the Extended Filter).
  • Count the number of connections in each time bucket, also applying any selected splits at this point.
  • Divide the count of connections in each time bucket by the time bucket size (in seconds).

 

Throughput (Mbps)

In the context of a timechart, Throughput (Mbps) is calculated as follows;

 

  • Filter away all records outside the time range.
  • Apply any further filters (from the Component Filter and/or the Extended Filter).
  • Calculate the size of each transaction in Megabits (Mb): 
    • Add together "vserver bytes in" and "vserver bytes out"
    • Multiply the result by 0.000008 to convert into Megabits. 
  • Sum the transaction sizes in each time bucket, also applying any selected splits at this point.
  • Divide the total transaction sizes by the time bucket size (in seconds).

 

NOTE - This process does not filter away either request-basedor connection-based metrics; throughput is a metric that is equally applicable to both connection and request metrics, and Throughput (Mbps) shows the total throughput for all metrics. 

Avg. Request Duration (ms)

In the context of a timechart, Avg. Request Duration (ms) is calculated as follows:

  • Filter away all records outside the time range.
  • Filter away all connection-based records.
  • Generate a request duration field for each analytics record in milliseconds (ms):
    • Subtract "timeline.crqs" from "timeline.crse". 
    • Multiply the result by 1000 to convert into milliseconds. 
  • Apply any further filters (from the Component Filter and/or the Extended Filter).
  • Calculate the average duration of the requests for each time bucket, also applying any selected splits at this point. 

 

Avg. Connection Duration (ms)

In the context of a timechart, Avg. Connection Duration (ms)is calculated as follows:

  • Filter away all records outside the time range.
  • Filter away all request-based records.
  • Apply any further filters (from the Component Filter and/or the Extended Filter).
  • Calculate the average duration (in milliseconds) of the connections for each time bucket (average of "duration" field in the transaction record), also applying any selected splits at this point.  

 

This article is part of a series, beginning with Analytics Application - Concepts and Metrics Explained

Prev: The Connection Dataset

Next: Exploring Table Views

 

Version history
Revision #:
6 of 6
Last update:
Thursday
Updated by:
 
Labels (3)
Contributors