Profiling RMB performance

[sameh-farouk]

In a round-trip these steps happens for each way

• peer querying rpc node for twin public key
• peer encrypt message
• peer send the message to a relay
• relay federate the message if needed
  • relay querying rpc node for relay info
  • relay send the message to another relay
  • relay push the message to connected peer
• peer decrypt the message

Then there is the time the node (the handler service) takes for handling the command in the message and push a response.

We should gain a deeper understanding of the performance of the over-RMB communication and identify any bottlenecks or areas for improvement.

[xmonader]

The relay should be able to handle 10k requests/s from, And for the client it should be able to communicate with 10k nodes still in under 1min