Most of the respondents to my last week’s challenge got it almost right. The minor (common) error was the assumption that police rate percent 50 would result in a TCP session getting 50% of the bandwidth. Eyal got that right: the TCP throughput is always significantly lower than that due to frequent drops caused by low burst sizes assumed by the police command and resulting TCP restarts (the most I was able to push through was around 90 kbps; half of the bandwidth would be 128 kbps).

Many respondents got the third case (bandwidth class, police class and default-class all active at the same time) wrong. Vaidotas was guessing in the right direction and Petr knows the correct answer, but did not want to spoil the fun. Here’s the surprising result: the bandwidth class gets almost all the bandwidth. Sometimes the TCP sessions in other classes wouldn’t even start.

To understand that behavior, we’d need to go deep into the bowels of Weighted Fair Queuing. Topic for another blog post… in the meantime I found an old article by Petr Lapukhov on Internet Archive.

Recommendation

If a single class in an outbound service-policy uses the bandwidth action, all the other classes should use the bandwidth action as well. The classes without the bandwidth action and the default class might get starved during congestions.