Information:

Environment

  • Dynatrace Synthetic Classic Platform

 

 

Solution

In brief, Machine Throughput relates to the speed of the machine and Server / Test Throughput relates to the speed of the test.

Machine Throughput

Machine Throughput is an indication of the Bandwidth available during a specific test run. Bandwidth is the frequency at which data can be transmitted. Before and after any test is run, a bandwidth test is executed, which evaluates the throughput of the machine based on a 3rd-party bandwidth URLs, independent from the test run; the results of this test determine your bandwidth speed. Machine Throughput is determined by the average of Pre-Test Bandwidth and Post-Test Bandwidth test results added together.

Bandwidth testing only applies to Last Mile and Private PEER tests. Test data for these kinds of transactions are broken down into three groups, based on connection speed. There are three bandwidth categories:

  • Dial Up: 56Kbps or lower.
  • Low Broadband: 56 to 499Kbps.
  • High Broadband: 500Kbps or higher.


By sorting peer data based on bandwidth, Last Mile and Private PEER tests can give a more precise overview of a web application's performance. If data collected by peers are all grouped together, it then becomes difficult to determine how efficient a web site is because it mixes data from Dial-up users and Broadband users together.


Server/Test Throughput

Server throughput is how many kilobytes are downloaded per second from the target web server during the test run. It is calculated as the total kilobytes downloaded divided by the response time.

Server Throughput is determined by first taking the total bytes downloaded for the page, and dividing this number by the page's total response time.

This number is then broken down into ranges:

  • Low. The lowest third of the results.
  • Medium. The middle third of the results.
  • High. The upper third of the results.


The Portal then calculates the same information for every object by dividing the bytes downloaded for the object by the object's total response time. The result of this is compared with the ranges to find where the object falls - low, medium, or high.

Because Server Throughput is calculated by the test page URL, it can reflect whether anything changed for the tests; it can also be used to quickly scan through the object breakdown and identify the worst-performing objects.