t3.smallinstance (T3 Unlimited disabled to avoid bursting) @ 2 vCPU 2 GB and Gigabit network in the
Europe Frankfurtregion. The ping between client location and server location averaged ~ 40ms.
39msto distribute the messages for 250 concurrent users, in addition to a connecting-disconnecting ramp-up amount of between 100 and 250 users. As you can see, soketi easily handles this load scenario.
t4g), the performance delivered was 30% greater using the Docker ARM builds.