Favorites (Re: UUNET peering policy)

Which side of the debate do you want to take?

The traditional arguement is a network composed mostly of a few large data
centers, with lots of servers sending traffic is getting a "free ride" on
the network which built out nationwide and has POPs in every LATA.

UUNET deserves a return on its investment on all those wholesale dialup
POPs and circuits to underserved rural areas. Abovenet is just cream
skimming in a few large metro areas, while UUNET does the hard work of
carrying that extra traffic imbalance. Abovenet selling "cheap" bandwidth
because it doesn't have the cost of delievering the traffic that UUNET
has to pay.

The opposite side is Abovenet has invested a lot into its sites and MFN
into its networks. It just choose to do it in a different way than UUNET.
Its more expensive to lay fiber in metro areas than rural areas. It costs
a lot of money to operate the centers. Whether the traffic is being paid
by the millions of $19.95 dialup users on UUNET's wholesale ports or
by the hundreds of hosters in Abovenet's sites, the traffic is paid.

The argument that "outbound is bad" was particularly sound when most
national backbones were 1) leased and 2) small. That's not really the case
anymore. With lots of bandwidth, this paradigm should really start to
shift - who cares if someone if dumping outbound on you, and you have to
backhaul it across the country? There would seem to be no actual economic
reason to justify traffic ratios - except, of course the desire to
discourage peering and encourage transit, but not removing a barrier that
once was meaningful, and now is deprecated.

I don't feel sympathy for the example of many dialup POPs and a few large
data centers. There is a value for both parties in the exchange - content
is delivered to customers faster. With the cost of long-haul taking a
dive, one might think this would have more impact than a traffic ratio.

- Daniel Golding

Warning: the following is oversimplified:

I think what it comes down to, and what it has come down to at least since
the inception of hosting companies who spew large amounts of traffic back
at the access networks, is who gets paid twice for carrying the
traffic? Does UUNET get paid twice for carrying the traffic, once by
their customer that pays for dial or leased line access and once by the
hosting company that pays for peering because their traffic ratio is
off? This seems to be the status quo. The other way around would mean
that the hosting company got paid twice for carrying the traffic.

-travis

... and if we bring the chicken and the egg syndrome into all of this,
then who has most to gain that the internet is fast and efficient, the
people who make content, the people who live off of the people who make
content, the people who live off of the people consuming content or the
people who consume the content? ... and where is the growth? ... and who
gains from that?

One might think that UUNET in this case would be happy that someone would
provide content so that more users would want to buy even more bandwidth
so that UUNET would make more money. On the other hand one might also
think that UUNET would want their investment to last as long as possible
and that upgrading is more expensive than making it possible for people to
buy more bandwidth.

Whatever is most important, I do not know. I know it's easy to argue for
both ways.

If I am a generic business, and I connect to UUNET twice, in different
locations, and the ONLY traffic I send is between my two offices, should I
expect a discount on transit?

No. I would expect to pay it at both ends.

What is the difference?

Actually, if you are a generic business, and you connect to UUNET twice,
in different locations, and the ONLY traffic you send is between your two
offices, you should talk to your salesperson about VPN services on their
backbone vs purchasing transit from them. And YES -- you SHOULD expect
it to be less expensive.

sean@donelan.com (Sean Donelan) writes:

> "Favor"? What, precisely, connotes "favor" in this regard?
> Sending more, or receiving more? And: why?

Which side of the debate do you want to take?

Neither. The argument had better not be about last mile costs, since those
are actually comparable when you count them per bit-mile rather than per
customer. Nor is it about last mile margins, since if someone is charging
$19.95/mo for a service that costs $25.00/mo to provision, then expecting
companies who are less stupid than that to go ahead and do some bizarre kind
of "profit sharing" anyway is just insulting.

I don't see a "favor" being granted here, either by the side who transmits
more, or by the side who receives more. Both sides had better be charging
what it costs them to provide the service, plus some kind of margin to keep
the shareholders in the game. Failing to do this, even if for competitive
reasons, does not create liability against those who manage to do it right.

After all, in an actual "media", the content providers collect money from
the networks who distributed their wares, and those networks then sell the
eyeballs to the highest bidder. Fortunately, the internet isn't a "media".

The traditional arguement is a network composed mostly of a few large data
centers, with lots of servers sending traffic is getting a "free ride" on
the network which built out nationwide and has POPs in every LATA.

  Note, one of the ways AboveNet addresses this issue is by
honoring meds. This does not work with all networks, but I will
put forth an example:

                            LAX NYC
Net 1 (eyeballs): 1---------------2-----user
                             > >
                             > >
                         PeeringA PeeringB
                             > >
                             > >
Net 2: (content): server----3---------------4

  In a traditional (closest exit) setup the user will make a request
of the server, and traffic will flow as follows:

user->server: 2->4->3
server->user: 3->1->2

  Since there is generally much more traffic (bits) server->user
the "eyeball" network incurs a higher bit/mile cost (the long haul from
1->2) in this case.

  In the pure case (100% eyeball network peering with 100% content
network, closest exit) it is safe to say the eyeball network carries
a higher cost.

  Consider what AboveNet does. By honoring meds the following
paths occur (abovenet as net 2 in the picture):

user->server: 2->4->3
server->user: 3->4->2

  The bit/mile cost just shifted to the 3->4 link, paid for
by the "server" network. At this point the server network is actually
incuring more cost (assuming the pure case, again).

  This all gets muddied very quickly in the real world, as no
network is "100% eyeballs" or "100% content", and there are many other
geographical issues involved etc.

  This is but one tool that can be used to fix the problem of
one network incuring more cost than the other to interconnect. To
refuse flatly based on ratio, or bit/mile cost without looking for
relatively easy alternatives to balance the cost is a bit short sighted.

  Peering is about "equal value". Consider the pure content provider
co-located witht he pure eyeball provider, where neither have a long haul
network (eg local providers, similar bit/mile cost on each side). The
costs are equally distributed, both get similar "value" from the
interconnection (and indeed, can't exist without each other in the pure
case), but the ratio may well be 10:1.

This assumes that the network you are peering with sends correct MEDs.
This also assumes that one always honors MEDs. What happens during fiber
cuts or when you are waiting for an upgrade on your circuits? Do you
continue to honor MEDs and hurt the customer?

Honoring MEDs is not Black and White. I dont know one large network that
honors all MEDs from every peer. For the most part, I would say that most
providers do 'warm' routing. it just depends on what degree of warm you
can live with.

Christian

According to what I've heard from Qwest folks, they honor MEDS. (Taken
with a grain of salt, it was part of a sales pitch).

It has begun. Welcome to the 3rd world.

http://www.sfgate.com/cgi-bin/article.cgi?file=/news/archive/2001/01/17/state1503EST0161.DTL

Blackouts hit California as utility financial woes deepen

Associated Press, SF Gate Wednesday, January 17, 2001
Breaking News Sections

Even third world countries aren't as idiotic as California. There when
electricity goes off, it usually is because there is none to be had. In our
situation, it is manipulation and poor planning to the extreme.

My power was out for at least one hour, it is back now.

Bora