Gatling maxConnectionsPerHost: Determining the Optimal Value for Production Workloads

MaxConnectionsPerHost in Gatling

The maxConnectionsPerHost setting controls the number of concurrent TCP connections Gatling will open to a single target host. The default value of 20 is adequate for low‑concurrency local tests but can become a bottleneck in high‑traffic production simulations.

Uncertainty in Production Tuning

Gatling’s documentation does not specify a recommended threshold or scaling strategy for this parameter. While increasing the limit can reduce request queuing, it also risks exhausting OS‑level socket resources, especially on constrained production networks.

Open Questions

• What is the optimal maxConnectionsPerHost value for a given production load profile?
• Should the limit be scaled proportionally to the number of virtual users, and if so, how?
• How can we detect when the chosen value leads to OS‑level connection exhaustion without explicit error messages from Gatling?