I've read but not in depth all the various traffic shaping docs and seen the long question on HSFC scheduling
Does anyone really understand how HFSC scheduling in Linux/BSD works?
looked at the other questions but everything seems to be solving hard problems such as how do you divide up a given amount of bandwidth fairly, maintain good interactive sessions, etc.
But all I want to do is to divide up a known total maximum theoretical throughput, say, 1gig, into (say) 10 chunks of 100 each, and apply some magic so that any one TCP session is capped at 100 maximum.
It seems like a simple request, or a pointless one?, perhaps so simple that examples never discuss it instead go straight for more complex scenarios. But it would be very useful in a controlled environment where clients are not competing.
Edit: nginx has a limit_rate command that does exactly what I want (limit rate for an individual connection to an individual url), but I would still prefer to do it at the iptables level with tc or some other module so the question is open.
