| |
|
I promised in May to expound on the reasons why net neutrality is a red herring. There are a few very vocal proponents
out there who believe that pre-emptive regulation of the Internet is
necessary. On the other hand, we strongly believe that the real problem
lies in having the ability to objectively measure the quality of
broadband connections. I'd like to review the arguments in this post
and allow you to draw your own conclusions.
What is Net Neutrality?
Net neutrality has come to mean many things to many different people
over the past few years since Google and others began pushing the
agenda.
Fortunately, the web provides us with a good historical record of what has occurred. If we go back to the originally proposed Snowe-Dorgan bill,
we can see net neutrality proponents were pushing the notion that no
packet could have priority over another on the network. Gradually,
these same proponents moderated their position with a set of principles
that the FCC and others helped draft. It seems this new "bill of rights" for broadband consumers
attempts to embody fair reasonable and non-discriminitory access for
companies wishing to reach those consumers. Most recently, one of the
chief and very vocal advocates of net neutrality, Larry Lessig, has come to describe net neutrality as the notion that a broadband consumer should be able to connect any device that conforms to the Internet's standards.
Whatever net neutrality really means, most consumers would agree that
they want to have at least the same quality of experience while surfing
the web today as more peer-to-peer and streaming video or download
traffic traverses the network. Unfortunately, we have no way to
objectively measure the quality of experience today for the broadband
subscriber. By doing so, the experience can be guaranteed while either
throttling non-interactive traffic or insuring high quality
transmission of traffic like video streams.
The Internet Video Bandwidth Glut Problem
It is now estimated by some groups that due to Internet video explosion, bandwidth growth on the core of the Internet is exceeding 100% per year .
The rate of growth is so great that some ISPs both see a profit
opportunity as well as a great threat to their existing television
distribution franchises. I've been talking about this with many in the
media industry for several years now. Although, it would be true a few
years ago that many in Hollywood did not believe the video web would
replace TV. Today, only a few key decision makers wishing to lose their jobs might disagree.
How is the Industry Proposing to Solve the Problem?
There are several ways to solve the video bandwidth glut problem. The first is to artificially diminish demand. Andrew Odlyzko, calls this the "punitive tariff" approach. The Australian model for broadband and Time-Warner Cable's recent trial
charge broadband consumers more money for their service if they exceed
using more than some threshold of bytes consumed per month. Its not
clear whether consumers will accept this model. Some have criticized Time-Warner Cable's trial using this approach for impacting at least 12% of their user base
with this "tariff" and this approach may completely confuse consumers
by complicating the pricing model. Telcos and Cablecos are notorious
for creating complex pricing schemes which frustrate consumers. Its
interesting that Time-Warner is trying their model out in a Beaumont
Texas. Apparently, this region has limited or no real competition from
another DSL provider.
Another way to approach the problem is to simply maintain the status
quo. I'll explain a little later why over the long-term this can't
work. However, there are many, especially amongst net neutrality
proponents who simply believe that ever "widening the highways" is the
best way to go for the industry. I find this point of view to be
terribly naive.
The Internet today is not fair or efficient because of this model of ever widening the highway. Services like those of CDN providers exploit this model by effectively garnering more bandwidth at bottlenecks because they have much shorter latency or round-trip times to reach their destination. Similarly, Peer-to-Peer networks (P2P) not
only exploit the short round trip times, they also exploit the
flat-rate pricing most broadband subscribers have today. I find
"punitive tariffs" to be a very blunt instrument by which to solve this
problem. P2P networks are a bit like semi-truck drivers on the freeway
who are getting free diesel fuel. Our highways would be overwhelmed
with semi-trucks if this economic model remained in place. There are
much more elegant instruments we can use to make the network more fair
and efficient...
Mediating the network involves segmenting applications into different
grades of service. We can think of this as creating diamond, "truck",
and best-effort lanes of service. The question, especially in the
context of net neutrality, then becomes how do we insure that
determining which applications get moved into these different service
levels in a fair, reasonable, and non-discriminatory
manner. Network mediation makes it more efficient. Applications
like streaming video need diamond lane services. Sustained flows which
are downloading things like media or computer software can do so in the
background. P2P networks are much more fairly and efficiently treated
on the network by throttling the amount of bandwidth they use during
periods of congestion. This approach is analogous to the transportation
authority insisting that truck lanes are required during rush hour, or
limiting semi-truck access during those congested periods. Today, we
don't have a means by which we can measure best-effort use of the
Internet. There are no hard measures by which we can ascertain the true
performance of a broadband connection. On the highway, the
transportation authority can measure the average speed of the traffic
to determine if its truck lane and diamond lane policies are working
appropriately. Without hard measures, and the equivalent of what we
have with the telephony network's "Erlang model" there is no way in
which we can enforce a fair and consistent set of policies which will
make the network more efficient.
The whole point of the technology provider, or toll keeper for the
InterStream association is to implement such an Erlang model for the
Internet. By having a new measure for the best-effort "lanes" on the
network, we can clearly specify what an adequate or better broadband
surfing experience will be for the consumer. nuMetra
is developing the
control system which enables the industry to insure the quality of the
connection by knowing when to move traffic into their appropriate
"lanes" of service. Thus, we can solve net neutrality's political
problem through the use of the right technology and the objective
measures it provides. We just need to make sure ISPs fairly implement
these policies on their networks and that consumers
can objectively comparison shop for the best quality of their
service through use of the right metrics.
Insuring Fair Reasonable and Non-Discriminatory Access
I'm often frustrated by the "mis-targeted invective", as Esther Dyson calls it,
that often pervades the net neutrality debate. We must understand that
bandwidth on the network today is sold at a wholesale level. A customer
of a wholesale ISP must be able to purchase bandwidth that can reach
all corners of the Internet. Thus, IP Transit has become the primary means by which these customers buy bandwidth whether they have a major Web presence or not. InterStream's Mediated Bandwidth Agreement (MBA)
is a simple extension of this approach. Since it is reciprocal and
transitive in nature, it extends the reach of the existing transit
agreements so that diamond lane service can be offered end-to-end
across the network.
When selling "bits" of bandwidth at a wholesale level, ISPs cannot
discriminate between who can and cannot gain access to those diamond
lanes. Under the MBA, a wholesale ISP simply wants to sell a premium
service in exchange for incremental revenue above their best-effort
service. Thus, fair, reasonable and non-discriminatory access is
built-into the wholesale agreements. On the other hand, it is fair and
reasonable for the media industry to demand that pirates don't get
access to those same diamond lanes.
With the web today, pirates can easily offer content from virtually any website. ICANN, domain name registries and registrars
don't uniformly prohibit use of the sites for distributing pirated
content. The whole way the domain system evolved with its many types of
top level domains couldn't support a regulatory framework to place
these types of restrictions domain name holders. Because InterStream is
a private consortium, such restrictions can be placed
on who does get access to those diamond lanes. Thus, in the InterStream
system, insuring fair reasonable and non-discriminatory access really
comes down to how our terms of service agreement is structured. We're
still working on this and actively seeking input on the appropriate
approach to this agreement. Feel free to contact us if you want to provide us with your own feedback on either the process inside our oversight board or specifics regarding the agreement itself.
Jeff Turner
|
|
|
