eValid™ -- The Web Quality Suite Browser-Based Client-Side Functional Testing and Validation Page Timing/Tuning Transaction Monitoring. WebSite Spidering & Analysis and Realistic Server Loading.

Introduction
In eValid LoadTest executions the amount of data downloaded during a test varies depending on how detailed you wish the data to be. eValid supports three modes of download. These modes are selectable in the eValid script file with the three types of Server ... command.

• Serve FULL Mode. In this mode the entire page is downloaded. This includes the base page and all LINKed pages, plus all images referred to by either the base page or within LINKed pages. This is the equivalent of a single browser GoTo link.
• Serve TEXT Mode. This includes the base page and all LINKed pages, but excludes any images.
• Serve URL Mode. This includes only the base page and does not include any LINKed pages or images. This is the equivalent of a single HTTP Get Page request.

Illustration
To illustrate the differences we recorded a simple eValid script that downloads a single public available page: www.boardroom.com.

# Recorded by eValid V9
# (c) Copyright 2012 by Software Research, Inc.
# Recording made on Microsoft Windows 2000
#

ProjectID "Project"
GroupID "Group"
Testid "Test"

Serve FULL 
InitLink "www.boardroom.com"

We then edited the script to include each of the three playback variations described above, and then played the scripts back individually. To assure accuracy of the data, the eValid cache was turned off but detailed timings were turned on. We saved the three eValid Event Logs (which can be viewed below).

Comparison
The table below shows the results of this comparison. Remember, this data is for just one page, i.e. one URL.

What Does This Mean
Load testing could be done at any one of the three levels above. There will be big differences in the load measured at the server. Note that the above test always representes exactly one user, even though significantly different amounts of work are imposed on the server the this one user being served at the three different levels of detail.

Here are the pro's and con's of each level of detail:

• The FULL is the most realistic but imposes the most work for the server. If your LoadTest script uses this mode then the number of users served will appear to be small compared with the URL version. The advantage of this option is that you won't lull yourself into a false sense of security -- the data is 100% real.
• The TEXT is is fairly realistic and imposes much lighter load on the server. This is a middle ground, because essentially all of the page is actually downloaded and served -- only the images are left out. The number of server-side page hits will be low, but the number of users supported could be fairly high.
• The URL is is also realistic and imposes very light load on the server. There is some risk you could deceive yourself into thinking that your server has higher simultaneous user capacity because your simulated users are rather light weight.