High level design of the performance testing

[lpuskas]

• find and decide on the most important metrics to be focused on
• design the approach towards measurements
• find / design the tools needed
• document test cases
• implement test cases (in the order of importance)

[lgecse]

Let's start with measuring the number of messages / second
Use the traffic generator to generate messages: 1.000, 10.000, 100.000, 1.000.000, n ... and so on. After a point we'll see CPU (IO) saturation.

Optionally measure later the CPU, memory usage.

Result should reproducible, and should be a at least a simple report (CSV or md).

This should be the starting point.

Later use different machines / nodes for sender, receiver and slim to make sure traffic generator / sender will not impact the SLIM node itself.
Possible parameters for the next iteration are:

• message size
• number of consumers
• number of producers
• number of CPUs / memory

[lpuskas]

Closing the issue as most of it is contained by the attached PR. #155
Related work can be registered as new tasks