proft wrote:what kind of overheads should we expect from each eValid instance? I mean, how far off will the timing of internal events be?
Every eValid instance is a separate Windows process, so when there a LOT of
them running as on user -- e.g. a nominal 100-wide BU setup -- you can expect to see 50%+ CPU usage.
In most cases at 100+ BUs you don't run out of memory if you have at least 3 GBytes.
And, we generally see less that 5% input/output channel utilization in such cases.
At those levels our measurements show that the eValid results reported from internal timing commands are accurate to 1%.
eValid LoadTest Support