The graphs allow the user to see how the statistics evolve over the test's duration.
Important
Please refer to the section called “General Information” for details on various topics affecting this subject, such as filtering by Virtual User.
Select the item to be plotted in the tree: web page, container or HTTP request. Select
Allto obtain an average for all the HTTP requests.Select the statistic to plot: average time, maximum time, minimum time, hit rate, errors, throughput (Kb/s)...
Use the "
Add to Graph" button or drag and drop the item or statistic onto the graph pane on the right.
Click on the Monitors tab.
Select the monitor to plot in the tree.
Use the "
Add to Graph" button or drag and drop the monitor or group of monitors onto the graph pane on the right.
To delete a curve, either select the curve name in the caption bar
and click on the "Delete" button, or right-click on the
curve name and delete. You may also delete a curve using the pop-up menu.
Click and hold on the mouse button to select the rectangular area to zoom in on. The zoom takes effect when the mouse button is released.
Use "Ctrl" to obtain a zoom area that automatically takes in the entire vertical space. Click on the graph to return to normal view.
Right-click on the graph for the following options:
save the graph as a PNG image file: "
Save as image ..."print the graph
delete a graph curve
Each value is multiplied by its scale factor before being plotted. Selecting a curve name in the caption bar will:
set the scale factor for the curve to one, and change the scale factors of the other curves accordingly.
display the actual values in the scale for the selected curve.
View - displays or hides the curve. This does not change the other curves or the scale factors.
Color - curve color. Can be edited: double-click to display the color picker.
Factor - scale factor. All the curve values are multiplied by this factor before being plotted.
Name - name of the item being plotted (web page, HTTP request, monitor...)
Population - Population name. Empty if N/A. (See Filtering).
Virtual User - Virtual User name. Empty if N/A. (See Filtering).
Statistic - statistic name. Empty if N/A. (See Filtering).
Unit - statistic unit. Empty if N/A.
Click on the graph's "Options" button to display the options window:
Graphs may be plotted using different data for the horizontal axis:
time data (by default)
user load data: the statistics are presented according to the number of Virtual Users running, the existing curves being converted using the new horizontal axis. This plotted curve provides an easy way to study a statistic's behaviour under different loads. This axis data is less useful if the load is constant or varies little.
monitor data: a monitor may be selected among the monitors run during the test.
Here is an example of a curve showing the number of hits/sec. for a ramp-up load, using the number of Virtual Users as the horizontal axis data:
To compare the results of two tests, select the first test and plot the statistic for an item. Then select the second test, and plot the same statistic on the same graph. Thus, you can easily compare the impact of any modifications over time. The same method may be used to compare the results of several tests.
Several graphs may be created simultaneously for plottng statistics:
Use the tabs under the caption bars to switch from one graph to the other.
A percentile graph may be created to analyze the spread of a statistic's values during a test. This type of graph is used to determine the percentage of a test statistic at a set value. This may reveal, for example, that 80% of the response times for web page P were under 5 seconds.
Procedure 9.1. Creating a percentile graph
Click on the "
New Graph" button.Select
Percentile GraphDrag and drop the curve to be plotted
The selected graph's data may be exported using the Export button.
The Export wizard provides the following options:
Export data in CSV format
Export the image in PNG format
Copy the image to the clipboard (for pasting to any document: mail, MS Word,...)
Note: Not all test data may be exported at one time.