From the combined RBUDP and TCP bandwidth results, presented in the "Bandwidth Interval Plots" and "Data Size Plots" sections, the following conclusions can be drawn:

• Besides the statement from the developers that RBUDP is not TCP friendly, there still has been considerable adjustment from the RBUDP flow to the TCP flows, especially when the # TCP flows and  / or the shaping bandwidth is increasing.
• Probably due to the almost zero round-trip time there are no differences between the situations where the RBUDP flow is delayed (left plots) and where the multiple TCP flows are delayed (right plots).
• When five parallel flows between node pairs have been used instead of one, but with the constraint that the sum of the shaping bandwidth over the parallel flows is the same in both situations, the adjustment of the RBUDP flow to the TCP flows is much larger in the situation with five parallel flows.
• When the RBUDP data size has been enlarged the adjustment from the RBUDP flow to the TCP flows is decreasing.