High response times and connection timeouts during simulation run

Gatling (Open-Source)
1

I have been running a Gatling simulation to evaluate how a content platform behaves under moderate load, specifically testing endpoints similar to the structure on this site. During the test execution, I noticed that once the virtual user injection reaches around fifty concurrent requests, the average response times increase substantially and several requests begin throwing connection timeout exceptions.

I checked the server logs to identify whether the issue originates from connection pool exhaustion or unoptimized database queries, but the metrics seem within normal ranges before the drop occurs. Has anyone experienced a similar bottleneck when simulating traffic patterns against this type of architecture, and are there specific Gatling engine or HTTP protocol configurations recommended to isolate the root cause?

3

At such a low load, there’s absolutely zero chance that the bottleneck is Gatling itself.

Typical culprits:

  • you’re hosting Gatling on a host with an anti-virus that’s kicking in
  • what you’re monitoring with your logs is actually behind the bottleneck: a load balancer, a reverse proxy, your HTTP server