Network bandwidth impact on loadtest


We are running a load test on a website with an arrival rate of 4000 transactions per hour. We are seeing Read Time out errors happening during the test.

We are running the tests from our local machine which doesn’t have too good a bandwidth speed (7-10 Mbps download). Is there a possibility of the bandwidth causing the read timeout’s or is it just that the site under test is not able to handle the load?

In the test report -> Agent section, we see that the test didn’t put much stress on the memory.

Please check the network tab and see your used bandwidth. For the agents, missing bandwidth is not stress at all, hence the cpu is no indicator.

When you test from a local machine, you might run into firewall or system limitations especially when running from Windows. In terms of timeouts… does everything timeout or just certain requests?

Thanks for the quick reply @rschwietzke !

In the network tab I see that received (download) was 178 Kbps and Sent (upload) was 14.7 Kbps. So I guess not much network bandwidth was needed for the test.

Timeouts were happening for all the requests. We see a spike in timeouts at the end of the ramp up period. Which indicates an issue with the site performance I guess.

1735 saw an event aka some state change and something went toast quickly. Because it later was better, I guess there is something wrong… maybe scaling pain or a second box coming online… looked great till 173450 or so… try that load at max and run it for 30 min to see if it is stable.