[ih] Why was hop by hop flow control eventually abandonded?

[John Kristoff]

On Tue, Jul 16, 2013 at 05:55:00PM +0200, Detlef Bosau wrote:

> The more I think about it, the more I fear, that although the decision
> to abandon hop by hop flow control
[...]

That was well before my time, but arguably versions of independent
per-hop control has been attempted, if not widely implemented since
then.  Isn't a great deal of the CoS (aka, QoS) work, usually done with
those magic few bits in the IP header, later finalized as DiffServ, a
descadent form?

In addition, much of the work on router queue management, usually done
on a packet independent basis such as rudimentary implementations of
RED do this as well, augmented with ECN signallying.

Further, isn't much of the renewed interest in so-called "buffer
bloat" a related area of work.

It should seem like a small miracle, if not a bit scary, that the end
host based flow and congestion control we have works as well as it does
in this hugely autonomous system of networks and hosts.

John