Good to hear the results so far make sense.
Re the instability, I was finding that nodes were dropping peers until they became disconnected. CPU utilisation seemed ok, so I don’t think it was too much time signing. I believe that it’s more related to volume of network connections.
We are working on improving the scalability and stability of the simulations in Warnet. I will report back when we can push the tx rate higher.
In the meantime, could you help me understand the following a bit better?