[171902] in North American Network Operators' Group
Re: Observations of an Internet Middleman (Level3) (was: RIP
daemon@ATHENA.MIT.EDU (Owen DeLong)
Fri May 16 10:00:45 2014
X-Original-To: nanog@nanog.org
From: Owen DeLong <owen@delong.com>
In-Reply-To: <5375ADFB.3000303@vaxination.ca>
Date: Fri, 16 May 2014 06:54:33 -0700
To: Jean-Francois Mezei <jfmezei_nanog@vaxination.ca>
Cc: nanog@nanog.org
Errors-To: nanog-bounces@nanog.org
All the talk about ratios is a red herring=85 The real issue boils down =
to this:
1. The access (eyeball) networks don=92t want to bear the cost of =
delivering what they promised to their customers.
2. This is because when they built their business models, they =
didn=92t expect their customers to use nearly as much
of their promised bandwidth as they are now using. Most of the =
models were constructed around the idea that
a customer receiving, say 27mbps down/7mbps up would use all of =
that bandwidth in short bursts and mostly
use less than a megabit.
3. New services have been developed (streaming video, et al.) which =
have created an increasing demand from
customers for more of the bandwidth they were sold.
4. Instead of raising the prices to the access network customers or =
accepting that the lavish profits that they eyeball
networks had been pocketing were no more, the access networks =
are trying to slough off the costs of delivering
that higher fraction of what they sold onto someone else.
5. The content providers looked like an easy target with the =
advantage that:
A. Some of them appear to have deep pockets.
B. They are the competition for many of the access =
network=92s other lines of business, so increasing their costs
helps make them less competitive.
C. Consumers are emotional about price increases. Content =
providers look at it as a business problem and
perform a mathematical analysis. If their customer =
satisfaction impact costs more than paying the extortion
from the access networks, they=92ll pay it.
In reality, if the $ACCESS_PROVIDERS wanted to satisfy their customers, =
they=92d be aggressively seeking to peer with content providers in as =
many locations as possible. They might (reasonably) require content =
providers to build out to additional locations to keep their long-haul =
costs down (It=92s reasonable, IMHO, for a content provider not to want =
to carry multiple gigabits of traffic from a content provider clear =
across the country for free. If $CONTENT_PROVIDER wants to access =
California customers of $ACCESS_PROVIDER, then it=92s reasonable for =
$ACCESS_PROVIDER to insist that $CONTENT_PROVIDER peer in California for =
delivering those bits.)
Neither side of this issue has completely clean hands. Both have been =
trying to take as much of the money on the table for themselves with =
limited regard for serving the consumer. The Access Networks have done a =
far worse job of serving the consumer than the content providers and =
that=92s a big part of what is driving the current backlash. As a =
general rule, access customers don=92t select the provider they love the =
most, they select the one they think sucks the least.
I think the recent FCC NPRM is a bit optimistic in that it expects the =
$ACCESS_PROVIDERS to act in good faith. If they do, it will likely turn =
out to be a limited victory for the $ACCESS_PROVIDERS. However, I don=92t =
expect the $ACCESS_PROVIDERS to live within that limited victory. =
Assuming the NRPM becomes rule and then withstands the likely legal =
challenges, I expect they will, as usual, play in the gray areas of the =
ruling as much as they think they legally can and push the edges as far =
as possible to try and extort every dollar they can from =
$CONTENT_PROVIDERS with this so-called fast-lane (which we all know is =
just preferential peering and/or QoS[1] tuning). I suspect they will =
likely push this far enough that over the next several years, things =
will get progressively worse until the FCC finally decides that they =
have to move from section 706 to Title II.
OTOH, if I=92m wrong and the $ACCESS_PROVIDERS suddenly start behaving =
like civilized companies, develop a sudden concern for their customers=92 =
experiences, and start unimaginably acting in good faith, the proposed =
rule wouldn=92t be so bad for $CONTENT_PROVIDERS, $CONSUMERS, or =
$ACCESS_PROVIDERS.=20
Of course, you can already see the $ACCESS_PROVIDERS laying the =
groundwork to try and mount a legal challenge against the FCC=92s =
authority to use rule 706. Sadly, some of this groundwork is being laid =
by FCC commissioners. Said commissioners clearly have no interest in =
representing the people=92s interest and are strictly there as =
mouth-pieces for some of the big players in the industry.
Owen
[1] QoS =97 A deceptive name if ever there was one. QoS is not about =
Quality of Service, it=92s about screwing over network users by choice =
rather than by chance when you haven=92t built an adequate network.
