I again saw Joseph Reynolds speak on web performance and load testing.

Unlike the first talk I saw at last year's Code Camp, this talk, at MeasureUP, had as a tangent discussion a bit on how to cast a web test, which records a series of actions and takes health measurements on how snappy it was, to a Selenium test which is more or less the same thing with a validation of a particular workflow's results in lieu of a gauge of how fast the workflow went. I didn't come away with an understanding of how to do it, but it is possible! I guess this means you can write one set of tests for both things if you put your mind to it. You have to have the highest SKU (stock keeping unit) of Visual Studio to do the load testing stuff and it is called "Enterprise" for Visual Studio 2015 though it is "Ultimate" in Visual Studio 2013. Anyhow, a series of web tests can, as a suite, make a load test which may be run repeated emulating x number of users (and thus simultaneous hits) for y number of hours wherein both x and y are variables you set. Joseph had a formula for whether or not your web app can handle the traffic it needs to handle that utilized the x and the y and also had z for "time spent per user" which itself is driven by a different formula that he didn't delve into. If you divide y by z and then take THAT number and divide x by it you will have your answer! The image shown here shows some sample calculations with the x and the y and the z. If you are making your target, try to go beyond it. You need to know where your app breaks because... What if your Mickey Mouse little web site takes off and starts getting flooded with traffic as you hoped it would? You need to know when it is going to fall down as you yet watch things get "better and better" traffic-wise, right? You may see the talk for yourself here and I know this because I found it here which was advertised to me by an email from MeasureUp's keepers a few hours ago. All or at least most (not sure which) of the talks from Saturday were recorded and are watchable online. Per Joseph: Don't start performance testing at the end of a project. Don't think you can just slip that in in the last few months. By then the little problems have metastasized into big ones and you will have your work cut out for you. That said, you really can't start the performance testing in the first iteration either as there is little to test then. Joseph recommends starting six to eight weeks in. That is the sweet spot.