A Working Performance Test ~ ASAP!

A Working Performance Test

A Working Performance Test ~ ASAP!

Over the years I’ve learned that when starting performance testing on a new, unfamiliar application, it is invaluable to create a working test as soon as possible.

BAKE A LITTLE CAKE!

Let me illustrate with a story from Downton Abbey. The cooks needed to bake a hugely important wedding cake, but they have some unknown ingredients from an unusual source. Instead of baking the big cake, they make a small “test” cake to see how the flour will perform. After cooling overnight, the mini test cake looks great but tastes awful. They go back to the bag of flour and find that it is mostly plaster dust. (Good ingredients were hard to come by in post WWI England).

A WORKING PERFORMANCE TEST

In the same way, when we need to performance test a new application (made up of unfamiliar ingredients from an unfamiliar source), wisdom tells us to start small and create a mini performance test. Yes, we need to get the big picture such as scope, schedule, cost, and quality. As well as to establish that a performance test engagement is a multi-threaded effort. From Workload Analysis, Test Harness Creation, Monitoring Strategy, and so on. That all can happen while we are “baking that little test cake”.

RECIPE: BAKING A SMALL PERFORMANCE TEST “CAKE”

  • Create an initial script ~ perhaps just a login process. Don’t sweat having lots of different login IDs. Simply get two or three at a minimum.
  • Get the script working in a scenario that will run in your load test tool ~ e.g. LoadRunner, StormRunner Load, Blazemeter, NeoLoad
  • Run a test that does a little something ~ Try running a bunch of logins and logouts in rapid succession. This workload isn’t necessarily valid, but it makes the application work.
  • Collect your data from your test tool, and any relevant monitoring data that is in place.
  • Create a simple report for your IT colleagues and Clients to review.

There you have it, from soup to nuts, or maybe more accurately from flour to cake.

WHY IS THIS A GOOD IDEA?

There are many reasons why this is a good approach. Here are just a few:

  • Scripting! This is a great way to discover how to script this new application. We need to know how easy or how hard scripts are to create, which will also serve in creating informed test script creation estimates.
  • Client Satisfaction! The client gets to see us in action. We create load, we create a test report. They get a feel for the process, and how to interpret the information we provide.
  • Avoiding Paralysis Analysis! It is easy to get in the weeds on a test engagement, particularly large ones. This “test cake” creates momentum and can set the tone for the rest of the engagement.

SUMMARY

To recap, create working performance testware as soon as possible and you will serve your clients and your team well.



RSS
Follow by Email
Facebook
Facebook
LinkedIn