For a new software release, when should the performance testing take place? At the end? Performance testing is all too often assumed to mean load testing which can only really happen when the infrastructure and code is in place, so carrying out this activity at the end seems to make sense.
The trouble with this approach is that at such a late stage in a release, the options open for fixing a problem are usually limited to optimising back-end performance – or buying greater server capacity.
It’s now 10 years since Steve Souders was hired by Yahoo! to improve the performance of their products and he too assumed that his job was going to primarily be about making back-end systems work better. But he found something surprising. Most of the time users spend waiting for a page to load was after the html had already arrived, in getting, assembling and executing the other resources.
And so he set to work on developing a set of best practices or rules and weighting them according to importance. YSlow was developed to allow any webpage to be analysed scored and issues raised with possible solutions.
By using tools like YSlow or Google’s PageSpeed early in the development process, performance can be designed into the solution. moreover, they can be incorporated into your continuous integration jobs to give timely feedback for code changes.
So let’s go back to the question at the top:
For a new software release, when should the performance testing take place?
At the beginning.
e-testing have assisted a number of organisations with their performance testing. We offer a range of services to companies both large and small. If you would like a quote for performance testing at your company, please use our online performance testing quotation tool.