[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: "real" ingres implementation
> for forwarding path, absolutely. For the local destination path, you
> brought up ECN ;-> and so i was just thinking out loudly
Oh, I see ... yes, you have a point there. For bastard qdiscs (*) like
dsmark, this makes sense indeed, ugly as it is ...
(* work-conserving isn't enough, because for local delivery, you don't
have useful backpressure, so it really has to be a qdisc that neither
delays nor depends on backpressure. This leaves ingress, dsmark ...
and maybe sch_atm - now *that* would be dirty ;-)
/ Werner Almesberger, ICA, EPFL, CH Werner.Almesberger@epfl.ch /