Re: [RRG] some musings on PI v. PA, and assumptions, requirements, and tradeoffs

[jnc@mercury.lcs.mit.edu (Noel Chiappa)]

    > From: David Meyer <dmm@1-4-5.net>

    > in the case of a large multihomed content provider, the core (network)
    > still bears the burden while the site gains the benefit.

Ooops, I just realized that your comment is just about the costs and benefits
of the multihoming, not about the costs and benefits of the content provider
in its entirety (which is what I replied to). Sorry!

To that point, I would say that yes, the network providers are providing
routing table entries. However, again, I would say that the providers'
customers, who are the ones using the content provider, are the ones who are
paying the price for those routing table entries, through their payments to
the service provider (or they should be, if the service cost structure was
rational).

And because the site is so widely used, the distribution of the costs (to the
customers) is roughly congruent to the distribution of the benefits (to the
customers).


To come at it another way, think about a small provider in some other
country. At the moment, all other providers (worldwide) have to have a
routing table entry for that provider, even though that provider doesn't
directly pay them for that cost.

Yes, I know there are inter-provider settlements (which cover mostly traffic,
I thought, but could be construed to cover routing overhead as well), and so
perhaps some money is coming through in an indirect way.

But it's really the same basic thing: providers everywhere are bearing a cost
(providing the routing table entry), but 'only' one site is getting a
benefit. In reality, of course, the *users* are also getting a benefit - but
at least in the case of a large content provider with PI-space, users
*everywhere* are getting a benefit, whereas a service clients of small
provider may be only benefitting a small number of users across the network.


I suggested long ago that people ought to have to pay to rent routing table
entries, to encourage people to do aggregation so that they could advertise
their particular piece of the address-space over a minimal scope.  Alas, I
don't think they world is ready for that (at least, yet :-).

(Now that I think about it, though, you'd think the providers would jump on
this idea - they are hurting for income, and this would be a new income
source... :-)

Certainly, if this were done, it would be useful for multi-homing as well (in
addition to other cases, such as small providers with non-connectivity-based
address space), because the costs would be directly visible, and would be
directly factored in to chosing among different engineering options.

To the extent that costs got shifted, though, expect people whose oxen got
gored to complain - even if they would now only be paying for something they
used to get 'for free'...

	Noel

--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg