What is the difference between load testing and performance testing?

Replies to This Discussion

Permalink Reply by Scott Price on January 14, 2010 at 8:39pm

Hi Vanita,

Thank you for your reply. I went to your blog and enjoyed the post you wrote about non-functional software testing. With your permission, I will include an extract quote from your writing into my blog post and link to your full blog. Please review it and let me know if you approve. Notice that I have placed pictures of the authors I quote, however I cannot find a photo of you. Would you like to send me one that I can include?

Thanks again,
Scott

• ▶ Reply

Permalink Reply by KevinTMartin on January 14, 2010 at 10:46pm

I believe the common usage of the terms, without concern for distinctions, as being equivalent, is the key, particularly when the domain shifts the factors influencing performance.

The qualification of web applications, as the domain, is important. Too often, the definitions circulate without qualifying their appropriateness is scoped to a particular software domain, ie the website user response time perspective can be from a classic dot com site, or a regulated industry operating under strict methodologies may define a performance measure in the requirements, and therefore need to spell out distinctions in the flavors of testing that show the requirement is satisfied.

QA professionals should understand how terminology varies and is appropriate to a specific domain, recognizing a definition of a term is merely a label, and definitions are not universally correct (or the accepted/common terminology) for all the domains where software exists.

When the performance/load/volume context shifts from dot-com to an operations monitoring domain (J2EE or .Net application with only a few users, and those users are passively watching a dynamic data presentation, where the data is updating every X seconds in response to server-side processing), then the relevant factors on system performance are not user initiated transactions but the data processing variables on the server side. In such contexts, the objective is to discover the bend and factors impacting in the performance curves (N server resources defining performance), independent of ideas about a requirement or what is a failure. In this domain, the factors behind the processing load, and therefore performance, are the data volumes, and server side processing parameters. If the parties involved in this domain think performance=volume=load, then any of the related labels (volume/load/performance/stress) assigned to the investigation will adequately distinguish the activity from other QA activities. Definitions in certification test sources or wikipedia, are irrelevant.

I encourage any person who interviews testers, and the for-profit-certification-businesses, to not focus too much on memorized definitions, but rather to focus on the candidates' understanding of the art of such testing by asking about their particular experiences in those tests and lessons learned in such testing, with an open-mind to testing outside the domains of dot-com or regulated industries.

• ▶ Reply

Permalink Reply by Scott Price on January 14, 2010 at 11:24pm

Well done Kevin! I'm impressed with the profundity. Thanks for the insight. I understand that context changes the meaning, and thus the way we perceive the differences. Not sure what portion of your thoughts I can boil down into a pithy excerpt - would you like me to include any of your comments in my blog?

Did you learn that as a Razorback? :-) I lived in Little Rock for about 8 years, and many of my friends are still there. I miss calling the Hawgs, but I love the mountains.

• ▶ Reply