Re: [conex] An interesting video

[John Leslie <[e-mail]>]

Mikael Abrahamsson <[e-mail]> wrote:
> 
> So let's take the analogy to this group. How much would it cost to bring 
> in an infrastructure to create the incentives to make people adapt to move 
> some of their traffic to another part of the day, or not run the traffic 
> at all (which is what a traffic congestion system tries to achieve).

   Umm... those aren't the only two possibilities, actually...

> I don't believe I have seen a system that actually reduces cost by trying 
> to intelligently manage congestion or moving traffic spatially, compared 
> to the cost of building out the network.

   There are already _many_ multiple paths from sender to receiver: we
don't have to build-out for these to exist. Our routing system, however,
ignores them because it's optimized for something else.

   Rather than change the routing system (IMHO), it's more likely we'd
build an overlay upon it able to "avoid" congestion by not following the
"direct" path. With a system of settlements, ISPs would have a reason to
_use_ the overlay system for some portion of traffic not marked for
"priority". The "cost" of doing so would be tiny.

> A lot of the success of the Internet has been flat-rate billing so one
> basically doesn't have to have a lot of system and staff to handle
> usage-based billing.

   Correct! A lot of plain-old-telephone-service companies are suffering
big-time from trying to maintain their usage-based billing systems. We
don't expect to charge individual end-users here.

> I don't really see how a congestion management system would fare any
> better. It adds complexity to the Network and I don't see how it would
> really help. Basically I don't see the incentives for the user to change.

   I'm not sure my views are considered "mainstream" anymore, but at first
our drafts were written in terms of a "congestion allowance" to the user
where the user would see no change in price, merely a limit on how much
of his/her traffic would be marked for priority at congestion points.
(The ISPs would resolve any imbalance with settlements, or merely drop
some congestion-expected traffic in the absence of settlement agreements).

   Toby and I being told to shut up, we have mostly done so; and now the
drafts concentrate on how to catch cheaters, without even discussion of
_why_ anybody could be bothered to cheat.

> But I'd gladly be proven wrong, that's why I follow this mailing list.

   :^)

   It's certainly possible that some actual protocol will come out of
this WG yet...

--
John Leslie <[e-mail]>