When preparing to run a load test using On-Prem Load Injectors in LoadView, it’s important to install the appropriate number of injectors to ensure accurate test execution. The number of On-Prem injectors required depends on the capacity of your servers, total number of concurrent users in your load test and the complexity of your test scenarios.

In LoadView, to run the cloud Load Injectors, we use the following configuration:

  • Operating System: Windows Server 2022
  • CPU: 8 cores
  • RAM: 16 GB

For these machines, the recommended numbers of concurrent users per machine are the following:

These payload values are just default numbers and can vary significantly based on the target website performance and test scenario settings.

  • HTTP(S)-based Test: 500 virtual users.
  • Web Page / Streaming Media / Web Application / Selenium Test: 8 virtual users.
  • Postman Collection Test: 20 virtual users.

Optimal On-Prem injector payload for your load test can differ from the values provided above. However, you can use these default payload values as a reference for assessing the number of the required On-Prem injectors for your test scenario.

Calculation

To calculate the number of On-Prem injectors, divide your maximum desired number of concurrent users by the payload the Load Injector can handle. For example, if your machine can handle 8 concurrent real-browser users, and you want to simulate 100 users, you will need:

100 ÷ 8 ≈ 13 On-Prem Load Injectors

Although the calculation returns 12.5 On-Prem Load Injectors, LoadView users should always round up to the nearest whole number—in this case, 13. Rounding up helps to avoid the risk of overloading individual On-Prem injectors, which can lead to inaccurate test results.

Adjusting Default Payload for Higher-Capacity On-Prem Server

Let’s say, you want to run a load test with the maximum of 100 concurrent users in a real browser. You plan to use your own server with the following configuration:

  • CPU: 16 cores
  • RAM: 32 GB
  • OS: Windows Server 2022

This server has twice the capacity of LoadView Load Injectors in terms of both CPU and memory. Assuming similar system efficiency and no network bottlenecks, your server should be able to handle roughly double the load.

Calculation

  • Default payload for LoadView Load Injector: 8 concurrent users/per injector
  • Client server (is 2× more powerful) payload = 8 × 2 = 16 concurrent users
  • Maximum number of concurrent users: 100 users

Estimated number of required On-Prem Load Injectors:

100 total users ÷ 16 users per server ≈ 7 On-Prem Load Injectors

Running a Validation Load Test

To avoid overloading your On-Prem server, consider running a small-scale validation test after estimation to fine-tune the actual payload per injector under your specific network conditions.

Monitor CPU utilization during the test to identify the number of users at which CPU usage reaches and stays at about 80%. Use this result to ensure On-Prem injectors run your full-size test without exceeding CPU limits, providing more stable and accurate results.