Analysis Diagram Merging
First, the analysis diagram merging principle
Select View->merge graphs and pop up the dialog box shown in Figure 1
Figure 1 (Setting up a merge diagram)
1, select the diagram to be merged. Select a diagram to merge with the current activity diagram, and note that you can only select the same chart for the X axis Unit of measure.
2, select the merge type.
1) Overlay: View the contents of the two graphs that share the same x-axis. The y-axis on the left of the merge diagram shows the y-axis value of the current diagram, and the y-axis on the right shows the y-axis value of the merged graph, as shown in Figure 2
Figure 2 (superposition of combined analysis diagram)
2) Tile: View in tile layout, share the same x axis, the merged figure appears above the current figure, as shown in Figure 3
Figure 3 (Tile merge analysis diagram)
3) Association: The y-axis of the current active graph is merged into the x axis of the merged graph, and the Y axis of the merged graph is the y axis of the merged graph, as shown in Figure 4
Figure 4 (association merge analysis diagram)
3, Merge Chart title: Set the View after the merged title.
Second, the example explanation
The following is an example to analyze how to combine the data graph, which is the running vusers,hits per second,throughput graph, 3 graphs superimposed combined graph analysis, as shown in Figure 5
Figure 5 (Combined analysis chart)
It can be found that running Vusers and hits per second graphs are almost invisible. This can often happen in the actual testing process, this is because the size of the y-axis is too small to affect analysis, sometimes the size of the x-axis is too small will affect the analysis, it is necessary to adjust the size of the x-axis or the y-axis display ratio, here just adjust the y-axis display ratio can be running vusers and hits per The y-axis of the second graph is magnified 10,000 times times, changed by the View->configure measuerments setting, as shown in Figure 6, as shown in Figure 7
Figure 6 (Adjust the y-axis zoom)
Figure 7 (combined analysis diagram after adjusting the y-axis display scale)
You can sometimes filter settings on a profiling diagram, but you do not have to do this in this instance. The following is an analysis of the 3 steps commonly used to analyze the merge diagram, which is to analyze the trend of the merged diagram and the 3 graphs, mainly to analyze whether the trends of these 3 graphs are correct.
1 Find the determinants that affect the trend of several graphs. The running Vusers diagram is extracted first, because the changes of the other two graphs are related to the VUser user;
2 should be aware of the relationship between the other graphs in the merged graph and the determinant factor. Here hits per second and throughput are proportional to the running Vusers graph, which means that the number of clicks and throughput increases with the increase of vuser users;
3) to analyze the change trend of the decision factor graph. The trend of running Vusers diagram is to load the VUser user first, and when all the load is complete, all the VUser users will run for a period of time and then start releasing the VUser users.
Finally, the relationship between the graphs is used to determine whether the trends of the other graphs are correct. To judge the change trend of hits per second graph and throughput chart, we should judge whether the trend of these two graphs is consistent with the trend of running Vusers chart. If consistent, the result analysis chart is correct, otherwise the result is incorrect, and if an anomaly is found, other analytical methods can be used to determine the real cause.
Parse Diagram Association
Analysis Diagram Association principle
Right-click to select the Auto Correlate Pop-up dialog box in the current analysis diagram, as shown in Figure 8
Figure 8 (Correlation diagram time range, here take running Vusers chart as an example)
1. The Time Range tab is used to set the time range for analyzing the correlation graph metrics
Suggest time rang by there are two ways to:
1) Trend (trend): Select the correlation measure value the relative stability of the period of time range
2 feature (function): Select a time range that is roughly the same as the overall trend in the time when the associated measure changes relatively stable
3 Best (Optimal): Select a period of time when the associated measure has a significant trend of change
You can also manually adjust the time range, there are two ways, one is to manually fill in the specific start and end time; the other is to drag the green and red lines to specify the starting and ending times, where the Green line indicates the start time, and the red line indicates the end time
2. Correlation optins (Auto Association) tab to set graphs, data intervals, and output options to be associated
Select the diagram you want to associate in the Select graphs for correlation. In the Data interval combo box, select the interval between which to calculate the polling for the associated measure, either automatically or in a custom setting. Select the level of display output in the output combo box. As shown in Figure 9
Figure 9 (Auto association tab)
Second, the example explanation
The Analysis Graph Association is through the complex statistical method, precisely locates which factor has the most influence to the transaction response time, the correlation does not pay attention to the concrete data, but focuses on the parameter sample in the certain time range State, the tendency. Only the stitches can use the auto correlation (except the Web page diagnostic The stitches out of the diagram).
Instance: analyzes the case where the ' average transaction response time ' is associated with the ' running VUser ' diagram.
1 The average transaction response time chart is the current activity diagram, and the associated graph is the ' running VUser ' diagram, as shown in Figure 10
Figure 10 (initial correlation diagram)
2) Set the filter condition. After automatic association, it is found that a lot of unnecessary transactions are also associated with it, then you need to filter it. Set Filter/group by, as shown in Figure 11, where the login transaction is selected for analysis.
Figure 11 (Setting filter conditions)
3 Set the time range of the profiling association. Right-click in the Filtered association diagram, select Auto Correlation, as shown in Figure 12, you can manually set the time range of the correlation analysis, but it should be noted that here to select the twists and turns of the analysis, and can not only choose a time range of twists and turns, Select at least one curve with more than two twists and turns. If you select only one twist, after automatic association, you will find that many items have a correlation of 100. This analysis does not make sense, because only one of the twists and turns of the time range is too small.
Figure 12 (set time range)
4 Analysis of correlation degree. Automatic Association will see the following list of all the relevance of information, select the highest degree of correlation to analyze. As shown in Figure 13
Figure 13 (analyzing the degree of correlation)
Here the entire association analysis is over, you can see that there are some similarities between auto correlation and merge, but at the same time there are some differences, the specific difference is as follows:
A.merge can not choose a specific time to slice, so only first use the merge to look at the overall trend, analysis of the global. Find the proper location, then use auto correlation slice, further analysis;
The B.merge output does not have correlation match, and the relationship between two parameters cannot be measured even if the correlate option using the merge does not have correlation match this value.
End.
If you are interested in software testing, want to know more software testing knowledge, please pay attention to the "51Testing Software Test Network" today's headline number.
