I have a very simple load question.
I am running a load test using VSTS 2008, and I have a test setup with a controller + 10 agents. This is a load test against the SharePoint farm that I have. My goal of the load test is to find out resource utilization at the web + app + db levels of my farm for any load scenario. An example load script is
Usage Profile: Medium Collaboration (as defined by SCCP)
User load: 500 (using a step load pattern = step 50 every 2 minutes and a warm-up time of 2 minutes for each step)
Thought Time: 0
Load duration: 8 hours
Now the question arises: is it fair to expect that indicators such as Requests / sec,% processor time on the web interface / App / DB, Test / sec, etc., will become flat or will enter a stable state at one point in time load test time. As I said, the goal is not to create a bottleneck, but only to measure resource utilization using the above load profile.
I ask this question because I see something else. At one point in the load test, requests / sec become more or less flat. But processor usage on network / database servers continues to grow. Rummaging through the data, I see that the counter of "tests works" also increases over time. So, if I run a load test for more than 8 hours,% processor can go further. Thus, I do not know what to consider as load derived from the load profile.
What does this counter "test running" mean? How is this different from tests / sec?
Another question: how can I find out why the “test works” counter shows an increase in overtime?
thank you for your time
source
share