| |
|
InterStream
has frequently been asked to articulate the differences in our approach
to creating a new wholesale bandwidth interconnection agreement over
the "traditional approaches". Unfortunately, there really haven't been
any traditional methods that have gained wide industry traction for
offering "diamond lane" services at exchange points. About 4 or 5 years
ago, one small and relatively isolated ISP in Asia began offering this
kind of service using MPLS-based VPNs (Multi-Protocol Label Switched
Virtual Private Networks). This "traditional approach" certainly worked
and as I understand it, now, a select few other carriers have picked up
on this method to deliver premium bandwidth to their broadband
consumers as well.
Unfortunately, MPLS-based approaches haven't delivered on the promise. At first, these "stateful" approaches
to quality of service (QoS) appeared to offer a great deal of promise.
Then, the network equipment vendor community and coordination
requirements amongst ISPs created an overly onerous proposition. For
those of you in the industry, you probably remember the consternation about Martini, Kompella, for layer 3 VPNs and now Atom
as an interconnect method for layer 2. Remembering which network
equipment vendors were going to support which standard and insure
interoperability significantly delayed market adoption.
While much of that is now behind the industry, there are still a number
of interoperability challenges which include how new standard protocols
should work transparently between ISPs. In the InterStream framework,
this would include standards for demand specifications of ISTP and the
mediation policies used on the downstream networks. In other words, due
to proprietary implementation differences, and the lack of coordination
amongst ISPs, the MPLS-based solutions haven't met the need to work
ubiquitously across the Internet -- particularly in the enterprise.
In addition, since the existing "diamond lane" agreements, as sparse as
they are, only exist between two parties, they don't satisfy the
requirement for a pervasive broadband television network service. Here
in the United States, we have two primary automatic toll collection
services: E-ZPass and FasTrak.
These services allow you to place a transponder in your car and breeze
through toll booths without stopping. In some cases, you can use them
to pay for "diamond lane" access to avoid congested highways.
The video Internet today doesn't have a general purpose "FasTrak or
E-ZPass service." We don't even have general "diamond lane" access.
Two-party agreements at the exchanges can't possibly enable an
end-to-end system for "diamond lane" or other services. There has to be
a single party collecting those tolls for access to the services and
acting as a neutral metering service for the exchange of traffic
between the ISPs. This is only business arrangement by which a true
end-to-end Internet service can evolve. InterStream provides the
framework by which this service can be enabled through nuMetra. The Mediated Bandwidth Agreement (MBA)
forms the basis for the legal and business framework of the association
and a business model which ensures mediation technology will be
deployed pervasively through the network. In other words, InterStream
enables nuMetra to act as the "FasTrak or E-ZPass for the video
Internet."
In this system, anyone who pays the toll and conforms with the association's anti-piracy policy,
gets their required bandwidth to stream video anywhere on the Internet
just like today's best-effort model. I can load a web page from
Australia, India, South America, or Africa today because the existing
wholesale peering and transit model scales globally. The same will be
true under the MBA. Because it reciprocal and transitive, any
InterStream provider in the system will automatically be connected
globally with all downstream providers. Therefore, customers of the
media grid can buy service from these ISPs with confidence that they'll
be able to reach all corners of the network.
There are number of other advantages to use of a MBA as well. Specifically, the MBA is:
Non-Proprietary - open, not vendor or service provider specific
Transparent - open to all legitimate parties who meet the InterStream association's quality standards and anti-piracy policies
Ubiquitous - works with virtually all broadband connections
Extensible - open source allows any party to extend the system, agreement, or network specific technology
Supports Media Industry Anti-Piracy Initiatives - pirates don't get access to diamond lanes, and illegitimate content will be throttled via mediation policies
Audit-able - legal and business framework allows association to audit and evolve anti-piracy policy quickly to meet market conditions
General - supports all mediation policies - interactive, streaming, best-effort, throttled, and other new services
Enables a General End-to-End Protocol - ISTP can work across entire Internet
In essence, the interconnect approach using the MBA doesn't fan the
fear-mongering flames of the net neutrality advocates either. Private
interconnect agreements would, in essence, allow any two parties to be
exclusive in a "diamond lane" service agreement. Anyone who pays to
tolls in the InterStream approach gets service. The proprietary and
private wholesale approaches create a great deal of risk for ISPs and
the media industry in that they invite regulatory oversight. Keeping
interconnection open and transparent is not only the best thing for the
industry, it also serves the self-interest of the ISP and media
industries because it will demonstrate that self-regulation can work
and the industry doesn't require formal governmental oversight.
In an upcoming blog, I'll discuss the why the real problem with net
neutrality is a technical problem, or the lack of an "Erlang model" for
the Internet. Until then, feel free to send us your feedback to <info> <at> <interstream> <dot> <com>.
Jeff Turner
|
|
|
