We often get the question: If we are using a Virtual Service Environment, instead of the live environment, are we still effectively testing the system?
When some people first hear about replacing constrained systems, mainframes, incomplete and under-development services and the like with Virtual Services, their first reaction is immediately: "That's not a real performance / load / regression test bed then."
I was at a meeting in Europe last week -- and chalk it up to perspective but one of our partner's engineering counterparts used a really effective analogy for how LISA VSE works.
Picture a car. When you want to test a car's aerodynamics, you can't get in it and start driving around and expect any degree of certainty in your results. So you put the car in a wind tunnel, and then do the testing. So when you simulate the wind hitting the car, you are certainly still testing the car, but you have the ability to turn the wind up and down and do the testing with more control.
I really liked the simple message and hope it helps you understand the concept of how a virtual service environment can actually work better and provide a better range of performance results than the expense and effort of always having to replicate or use "the real thing."
The same concept also applies for a stable set of test data to do your regression testing - the external data around your system under test is configurable and controllable, so if there's a problem, it is your app failing, not the test scenario becoming invalid due to unexpected changes in the data.
Comments