As our BLOG readers may know, we have been working recently with developing better understanding of the scalability of the eValid solution. "Scalability," simply put, describes the degree to which the solution can be ramped up to provide for large numbers of Browser Users (BUs) to drive load into a server stack.
Using a single user on a modest Windows machine, we regularly achieve 100 BUs in parallel. Using multiple users and a fairly large cloud-based machine image we have hit 10,000 BUs.
But a different way of doing things, which involves having one parent eValid BU launch multiple sub-windows, seems to be capable of besting even these very good scalability numbers. The advantage of using sub-windows is that each one takes far less memory footprint than a complete browser instance. But a disadvantage is that all of the sub-windows can't be active at the same time; so this solution only works when the tests you're running have a low-enough duty cycle (percentage of time active) so that cumulatively they don't exceed 100% of the main browser's time-line.
Within those constraints, however, this approach does appear to work quite well. Here is a screen shot of a typical Windows desktop with a total of 50 eValid browsers, each running 10 sub-windows, for a total of 500 + 50 separate activities: 550 Browser Screen Desktop Image. (Look closely at the bottom of the images and you'll see the 550 browser + window count.)
Isn't one picture worth at least 550 words?