Why is the "Max Latency" for “Throughput Threshold” in “Throughput Parameters” of RFC 2544 wizard, by default 30us?
There is no specific technical reason for selecting this value
when this RFC wizard was being designed, the DUT had probably given a worst value of just under 30us over the whole development/test cycles.
Hence they integrated that value as the default value when the wizard was implemented with STC firmware/software.
As for the designed usage of the parameter:
The basic idea is that this parameter allows for shorter iteration durations.
The main reason you have to run for longer times is to catch a situation where the DUT can nearly handle the tested rate.
In this case, buffers on the device can disguise the lack of forwarding capacity for a while.
If you make the duration long enough, the buffer fills and can no longer cover for the forwarding performance.
A properly defined latency number (perhaps 30us for a DUT but perhaps not for another DUT – And that is why the value can be tuned manually) will catch the buffering earlier in the test as the latency will increase as the buffer fills.
This can make the shorter test runs better at accurately discovering the true forwarding ability of the device.