I would like to use the Neotys Cloud platform for my tests and so i'm wondering how many Load Generators i would need.
You can install and use as many Load generators as needed for your test. But if you would like to use the Cloud Neotys platform you may want to use the necessary numbers of load generators and not more.
There are two different load generators type. There are 32 bits and 64 bits load generators. In the Cloud panel it's respectively medium and large. A medium LG instance is a 2GB RAM and around 800MB RAM is allocated to the LG. It is using dual core CPU. A large LG instance is a 8GB RAM and 6GB RAM is allocated to the LG. It is using quad core CPU.
The number of VU's that can handle this configurations depends on multiple factors like script complexity, technologies involved (HTTP, HTTPS, RIA applications ) etc...The best is to run a quick calibration test.
Start an LG instance of your choice (medium or large), run a ramp up load policy test using your real scripts and check the CPU/Memory usage of that LG. It is displayed on the NeoLoad runtime graph.
At some point, the LG will use around 75% memory constantly with maybe some peak. Note the number of VU's and you will have your limit. As an estimation, you can lower a bit that number of VU's to make sure that the LG will correctly handle that load.
For more information, see our “User guides” in the NeoLoad documentation.