Validation during load testing – trust but verify

A few days ago I went to pick up a package at the post office. To my surprise, everything went smooth – the place was open, the line moved fast and customer service was excellent. Only one small problem. The package I received was not mine. Which made me think, you guessed, about load testing.

What are we looking for in load testing? We’re aiming to validate ALL the following attributes:

  • Availability
  • Performance
  • Capacity
  • Validation, or delivery of the service/functionality that was asked for

The last item, validation, is often neglected in load testing. Most load testing articles or best practices discuss availability, performance and capacity. But the validation of the data received is hardly ever mentioned. Maybe we all think it’s obvious?

A few years ago, the company I worked for was using a cloud service of a well-known company. Part of the service included reports, and in one case, one of the reports included data of different customer instead of ours. The explanation was that “something went wrong with the multi tenant implementation under load”. I have no doubt that the company QAed their system under load, as it always performed flawlessly. Just that they never verified that the services were doing the right thing.

While you may trust the system, you must use validation points in load tests.

While you may trust the system, enter verification points into your load tests.

In short, to ensure your system does not deliver the wrong goods:

  • Make sure your load testing product provides an easy and efficient validation functionality.
  • Add to your test scenarios enough validation points that will validate that the services you are testing are executed successfully.

The short video below demonstrates how WebLOAD handles validation. Or you can download WebLOAD and see for yourself.