(This post was written by Omer Kenan, from the Performance Center Quality Assurance Team)
One of the new and exciting features that was added to Performance Center 12.53—NV Insights—is now available from LoadRunner 12.53 Patch 2.
Network Virtualization Insights analyzes application performance under various load and network conditions according to the scripts in a scenario. Its comprehensive network analysis report comprises information suitable for various audiences—from top-level managers to performance engineers. Starting from the high-level dashboard, the report is divided into multiple sections. This enables customers to drill down and focus on specific interests and gain insight about each and every transaction and virtual location. The report also provides invaluable guidelines on how to optimize an application for improved performance.
In this blog, I want to give a hands-on example of running a simple Web HTTP/HTML script that runs in Controller and generates an NV Insights report at the end of the run.
Here are some preliminary guidelines:
- Network Virtualization must be installed on the Load Generator machine (Windows only). During full standard installation, NV4ControllerSetup.exe and NV4LGSetup.exe are installed automatically. You can find the files for a custom installation under <DVD_folder>\Additional Components\Network Virtualization.
Standalone Load Generator installations install NV4LGSetup.exe, by default.
- Not all LoadRunner protocols are covered, but TruClient and Web-based protocols are included.
- NV Insights is enabled only when the ‘Enable Network Virtualization’ checkbox is selected in the Virtual Location Settings dialog box.
- When adding virtual locations on Controller, use the ‘Per Group’ mode.
- When working with unique parameters, make sure you have enough unique parameter values for an additional 2 Vusers that are added to each group; for which NV Insights is activated (more on this below).
- NV Insights is not supported for WinInet scripts.
NV Insights uses a specialized mechanism that collects data by running two additional Vusers (free of charge) on every group for which "NV Insights" is selected. The first additional Vuser runs without network emulation at the beginning of the script, providing a baseline for later comparison. The second additional Vuser runs with the same network emulation as the one defined for its Vuser group; and is launched together with the last Vuser from the group, i.e. at the peak of the group's load. Each of these additional Vusers runs for one iteration only , and thus their presence does not distort the results of the original load test in any significant manner. At the end of the run, NV Insights compares the results for each transaction and virtual location pair and generates a report.
Let’s dig in with a simple example of a scenario that runs with one group using a WEB/HTTP/HTML script. We’ll run the test on localhost with network emulation , selecting the “NV Insights” for every group.
In the scheduler, we are selecting 4 Vusers in total, each one starts after 15 seconds.
We want to add emulation to the group, so let’s define ‘Emulation_300’ with 300ms latency and 3 percent packet loss.
Finally, to add NV Insights to our run, in the ‘Run’ tab, we’ll select the new ‘Insights’ column check box for our group. By default, this column is un-checked. You can see that NV Insights will run on the group because the following message is displayed:
Here’s what the scenario looks like before starting the run.
When the run starts, two additional groups appear, each with one Vuser. These are the groups added by NV Insights.
At the beginning of the run, the first regular Vuser runs with emulation, and the baseline Vuser added for NV Insights runs without emulation.
After the baseline run finishes, the scenario continues.
When all Vusers are running, the ‘peak’ Vuser is added for NV Insights.
The NV Insights Vuser finishes running after one iteration.
Running the additional Vusers for NV Insights doesn’t affect the scenario run with emulation – as we can see, nothing changes.
The emulation runs with 300ms and the average latency measures ~300ms as expected.
Also, packet loss is ~3 percent as expected.
At the end of the run, data collection is activated and .zip files with data are collected to the results folder. This may take a few minutes.
After the NV Insights data is processed, the extra Vusers are removed from the ‘Run’ tab.
When there are files to be shown and no error is reported, the ‘NV Insights’ button is enabled.
You can also activate it from the Results menu.
After clicking ‘NV Insights’, a message tells us that the report is being generated.
Then a new window opens with the report. Here we see part of the dashboard.
Only thing left to do is dig in…
NV Insights is an exciting feature that reveals problems and provides optimizations for the scenario being run.
- NV Insights Vusers do not consume a license – they’re free of charge.
- Unique parameters that are part of the script should take into account the additional 2 Vusers in your scenario design. For example, if you defined 10 parameters and there are 10 Vusers, the first baseline Vuser consumes a parameter, so when you arrive at the 9th regular Vuser, it will fail.
Hope this blog was clear enough. NV Insights is definitely worth using. It has lots of benefits for performance engineers.
Interested in more information about NV Insights report in LoadRunner?, Visit the LoadRunner Help Center.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.