Categories: Workload Analysis | Author: Scott | Posted: 25/09/2015 | Views: 472

Once upon a time, when I was younger and new to the world of performance testing a much more experienced colleague said to me “Scott, you will never be good at this job until you understand Little’s Law”. I pooh poohed the suggestion at the time after looking at the Wikipedia article (which is actually quite good).

Read More...


Categories: Problem Analysis | Author: Scott | Posted: 23/09/2015 | Views: 359

In a previous post I talked about the benefits of taking a break when trying to deal with a nasty little problem that you just can’t quite figure out. In this post I’m going to talk about one of those problems that kept me awake at night, when I should have taken a break and what the solution to the problem actually was.

Read More...


Categories: Problem Analysis | Author: Scott | Posted: 18/09/2015 | Views: 313

As tempting as it might be at times to plug away with something until it's beaten in performance testing, you run the very real risk of getting into a position of not being able to see the wood for the trees. Sometimes the issue is one that in hindsight is so simple that you wonder why it took you three days and a brief flash of insight at 2 o’clock in the morning to come up with a working hypothesis.

Read More...


Categories: Workload Analysis | Author: Scott | Posted: 08/09/2015 | Views: 366

Everyone loves averages. We see them on the News almost every night, promotions and competitions tend to tout averages (1 In 6 chance of winning!), and of course, in performance testing there is always the average response time – that critical measure that everyone focuses on. The reason we love averages is because they are reasonably intuitive, it gives a single metric that describes the dataset that people can conceptualise. However, they can also be highly deceptive. The problem is that outliers can throw an average off to the extent that the dataset becomes useless. Or, in the case of performance testing and workload modelling if an average is aggregated over to coarse a time period it can seriously ‘flatten’ out spikes in load that we are interested in.

Read More...


Page 2 of 2First   Previous   1  [2]  Next   Last