We reform how you...
Business and technical success when load and performance testing depends on three main things:
These things can only be achieved when the parties involved understand the challenges, ensure the right capabilities are in-place and recognise who is responsible for each element. As a precursor to load and performance testing you should consider the fourteen points below to assess whether you are set for success.
Who will work out what patterns and levels of load will provide you with meaningful performance predictions? How will they do it?
How far should the system be pushed to see if it breaks? In which direction should it be pushed? Who can determine this?
Do you know what response times and other characteristics really impact the experience, effectiveness and value of the system? How can these be identified? How are the tolerable ranges to be worked out?
What is important performance, reliability or are they equally as important?
Where are the weak points of the system likely to be? Who is going to sit down and find out what these are?
To what extent will tests be tailored accurately target business and technical risk as opposed to being simplistic usage scenarios? Will testing simulate with internally generated activity? Will testing simulate activity triggered by external systems as well as users? Will the testing be layered?
How much data do you need to get into the system before you can start to trust the results? Where does it come from? Who is going to get it there? How long will that take?
How representative does the test platform need to be? What about scale? Where are you going to test? What about other systems interworking with yours? Will any simulators be required?
How much time should be allowed for getting the tests to work? How long for building the test rig? How long for testing? How long for fixing? What is the latest that can possibly start? When should it really start?
Where does the kit used to generate the load come from? Who is going to set it up? How does it connect to the system being tested? Is there enough bandwidth?
Who will maintain and operate the system during between test runs? Who will do it out of hours?
Who is monitoring what when the system is under test? Who decides what needs to be monitored? Who will set it up?
When thing go wrong what is going to happen? Who will ensure issues are tracked and resolved? Who will reproduce the problem? Who will diagnose the problem? What happens when a possible fix has been prepared?
Once it as gone live and we need to change things what do we do? How is that safe? Do we have to start from scratch again?