How Long Is A Mile On The Internet?

  Oktober 09, 2015

If you’ve got a website, chances are good you want to not only know how it’s performing, but optimize it as well. Both of those objectives can be achieved by monitoring your website synthetically using one of the many tools available, but with so many options you might not know which type of synthetic monitoring is right for you.

Synthetic monitoring basically comes in two varieties: backbone and last mile.

Backbone testing utilizes synthetic nodes (which are basically small computers with scheduling programs that run tests at specific intervals) located at data centers all over the world. Backbone testing allows users to run specific transactional tests using prerecorded scripts from the geographic locations where they are most concerned about performance.

Last Mile testing utilizes actual end user computers running a program that they’ve downloaded from the testing provider to run transactional tests using prerecorded scripts. Testers can select which bandwidths, browsers, and geographic region from which to run their tests.

With backbone testing, the metrics provided after tests are run are clean, accurate, and free of the “noise” caused by slow internet connections, multitasking, and other interference.  This data is obtained by testing under optimal conditions in a controlled environment, using dedicated machines that are owned, maintained, and operated by the testing provider.

Backbone Testing versus Last Mile Testing

Conversely, the metrics provided by last mile tests are often cluttered with extraneous data, misrepresented results, and can be unreliable due to the fact that often the testing provider does not own or control the computers performing the tests (or the programs installed on them), computers that are often being used for a variety of other tasks. Furthermore, the selection of bandwidth is often determined by the provider sending a small file to the computer running the test and measuring the time it takes to download. This makes it possible for bandwidth to be miscalculated when the actual test is run. Due to the ability of users to disguise their IP addresses, even the geographic region selected may be misrepresented.

For these reasons, backbone testing is more able to provide clean, unobscured results as to the slowest performing content, any issues or failures, and overall metrics under optimal conditions. Last mile testing may be used to obtain an overall picture of performance; however there should be an understanding that the data is not completely reliable, and that a good amount of data mining may be necessary to obtain useful information.