Did Internet Founders Actually Anticipate Paid, Prioritized Traffic?

Michael Dillon wavetossed at googlemail.com
Tue Sep 14 13:39:19 CDT 2010


>> Would you object to an ISP model where a content provider could pay to get an ISP subscriber's package upgraded on a dynamic basis?
>>
> Yes... Because the reality is that it wouldn't be an upgrade. It would be a euphemism for downgrading the subscriber's experience with other content providers.

A lot of people hear the term "quality of service" and think that it
refers to some mechanism that makes some packets go faster like a JATO
rocket pack made late 40's, early 50's airplanes go faster. But that
is not how the mechanism works. QOS mechanisms are based on making
some packets go slower, either by delaying them or deleting them so
that they have to be resent. This creates the illusion of speed for
the remaining untouched packets if the QOS is successful in preventing
congestion at network bottlenecks. QOS does not always prevent
congestion; it just reduces the likelihood that congestion will occur.
If the delayed/deleted packets belong to the same organization as the
so-called boosted packets, then this works OK because this
organization will have reasons for preferring that certain packets be
delayed/deleted.

The problem begins when the delayed/deleted packets belong to a
different organization than the boosted ones. That is not net
neutrality even if the packets have different diffserv markings. It is
even worse when the network operator selectively remarks packets from
one organization to cause them to be delayed/deleted. In this second
scenario both organizations inject packets into the network with the
same IETF diffserv markings but another network operator degrades the
service for one organization.

--Michael Dillon




More information about the NANOG mailing list