I seem to be getting strange results with eValid.
As we run the load up the performance drops and then it speeds up.
Can you help?
Hi ArleneV. Thanks for posting.
That's not a lot of detail to go on but we think there may be one common problem that we've seen that can cause you to think there is someting going on akin to magic.
Have you turn off the load balancer(s)?
In may websites the page requests are split between multiple servers by a "load balancer" which directs requests whenever possible to the least-loaded server in a "server farm." This often is done with "virtual machines" so that the available hardware resource that is used to deliver pages to requesting browsers grows dynamically...the more you ask for the more you get.
This is a common way, and a very effective way, to get the most possible performance out of several machines.
But for the performance tester this can present a real problem. As the load increases, the load balancer adds in more resources and, as if by magic, performance increases.
How do you get around this kind of issue?
Solution: Turn off the load balancers.
It turns out that this make sense becuase now you can study the performance degradation curve of one machine, when only that one machine is being used to fill page requests.
In the long run, you've partitioned the problem into two parts: testing the capacity of one server, and testing how effective the load balancer is?
-- eValid Support