lupine25 wrote:If I run a bunch of tests from eValid copies that are executed in the compute cloud (I'm thinking Amazon), are they realistic?
-Lupine
Thanks for asking.
As you know, eValid instances all run separate instances of the eValid browser. Ususally this is the "thin" version, which uses a smaller footprint than the full version, but behaves identically with respect to the activity it produces on the server.
So, given that all of these eValid instances -- we call them "Browser Users" to distinguish them from the "virtual users" that eVlite runs -- are all working as if they were real users, then yes, the load they generate is 100% realistic.
We see a lot of people doing loading with HTTP-based driver software and that is a good way to impose activity on a server. This is what eVlite does. But the big "but" here is that doing HTTP-based page GETs is not at all the same as having a simulation of a user filling in form elements, clicking into JavaScript functions, and all of the other things that AJAX applications do as they interact with the server.
So caution is advised when you hear folks claim that their HTTP-based load is the same as an Browser User...because it's not.
The eValid Team