The Fixed Load Generator simulates the behavior of thread-based LoadTesting tools, needs to be inter-connected with a Runner component, which it will trigger until the configured number of threads is running. In other words, this Generator will make sure that a certain number of simultaneous connections are always open.
An important note regarding load-based testing: Simulating real-world load by specifying a fixed load does not always make any sense, which is why all Generators in loadUI but this is rate-based as opposed to load-based. Real world users can not and will not take current server load or response times into consideration before visiting a website — but load based load generation does. Apart from being unrealistic this means that a server that handles requests really slow can easily pass a load-based load test. This means that an easy way for a developer to pass any SLA (Service Level Agreement) is to just delay the server handling of the requests (actually making the server slower), making it able to easily handle thousands of "Virtual Users". This is why we in most cases recommend using the rate-based generators (any other generator than the Fixed Load Generator). Read more about the concepts of load testing here.
The image below shows how the Fixed Load Generator is supposed to be connected. It is important that you connect it exactly as below (The Requests Currently Running output on the Runner to the Sample Count input on the Fixed Load Generator) in order for it to work as intended.