softqa wrote:What happens when two or more eValid's runing in parallel
expand the time it takes to download the JavaScript passages?
You've hit on the essential idea of eValid's approach to server loading: run multiple eValid copies to simulate multiple eValid users hitting/exercising the web application at the same time.
The key measure of success in server loading is, of course, scaling. When you are using the Thin version of eValid for script playback the scaling factor for eValid is at least 100:1. On machines images "in the cloud" we often can obtain scalings of 250:1.
The limiting resource, by the way, for this kind of instance scaling, is the amount of available RAM.
So the first part of the answer to your question is: no problem running multiple eValid instances...they are all separate processes and they don't intefere with each other at all.
The second part of you questions relates to the possible slowdown due to multiple eValid copies EACH having to load large JavaScript files. Here the situation is different because the limiting resource is the input/output capacity of the machine. In turn, that means that unless your connection to the internet can handle a 100:1 or more scaleup in input/output rates then you'll not be getting a true picture.
On the other hand, we have found that the actual input/output time is usually only a fraction of the total time a test playback takes, so the JavaScript download times usually -- but not always -- are not a factor.
If you run your multiple eValid's from the cloud, the input/output usually ceases to be an issue because most cloud instances have enormous (here, ready GByte/second) transfer rates.
As we say, however, "your mileage will vary" and a lot depends on the script, how long it runs, how much memory it uses, what it's duty cycle is, etc. In most cases, however, the scale-up issue is not a show stopper.
The eValid Team
by softqa » Fri Aug 28, 2009 4:09 pm