Faultline: The rise of the P4P architecture to turn the Internet into a big TV set Mar 19, 2008 – Rethink Research
From an ISP’s point of view, P2P traffic can appear to be exceptionally
daunting. If they choose to block it, as some have accused almost all of the
major US ISPs of doing, then their networks would become ghost networks, with
virtually no traffic in sight. But if they embrace it, their networks are
fast moving crazy places, where suppliers have to sprint to keep their
network surviving.
So what’s it to be? Well Verizon appears, at least to be considering a
middle road, one where instead of working against P2P, or just putting up
with its traffic costs, it will offer protocols to help co-operate with P2P
networks to deliver entertainment, by better understanding the conditions of
the network it is traveling over. That really IS open.
The initiative began last July and is through the auspices of a Distributed
Computing Industry Association (DCIA) working group called P4P, which stands
for Proactive network Provider Participation for P2P. The two founder members
and chairs come from Pando Networks and Verizon Communications. Pando is one
of the new breed of P2P companies trying to eek out a living in legal P2P
file delivery.
This is really a club for ISPs and P2P suppliers in which they can work out
there differences and it is so much more of a positive approach than whining
about network traffic and investing purely in “traffic shaping.”
Statements from this workgroup, published by the gadget loving Ars Technica
web site and others, claim that software that is already being tested which
can improve download speed between 200% and 600%, purely by offering up a set
of network APIs, which let a P2P application know which parts of a network
are busy, and using this to intelligently decide which P2P nodes should be
uploading in support of a file or stream delivery. It’s not rocket
science, and if a CompSci grad student had been given the problem he could
have come up with the same answer, but it is how to phrase that question
which is interesting.
If the question was “How do we get traffic zingin around the internet,
for nothing, without the help of the ISP and despite its best efforts to stop
us,” then that definitively is the Wrong question. If it were simply
told “you have a network and multiple copies of large files distributed
around that network, how do you build a rapid file delivery mechanism,”
then naturally you reach the DCIA answer. It is the history of ISPs and P2P
suppliers being at each other’s throats for so long, that makes it hard
to see how this might ever have come about.
In fact what needed to happen was that the livelihood of ISPs needed to be
threatened, where the average customer was expecting more and more from the
ISP, while the average monthly price for ISP service went down and down, and
traffic on their networks went up and up, forcing more and more investment.
At that point, P2P traffic is taken as a fact of life, not something that the
ISP looks to the US Supreme Court to make illegal.
ISPs cannot block all P2P activity because Verisign’s Kontiki P2P
client, which is now used to deliver millions of hours of TV services around
the world from respectable broadcasters, Skype, as well as Joost and
Babelgum, are not breaking any laws. Even Kazaa and BitTorrent may now be
carrying more legal than illegal traffic, or if not yet, they should lean
that way over time.
If we look beyond this simple set of proposals we see more and more which
might be done. By bringing ISPs and P2P suppliers closer, perhaps the
handshakes for this type of co-operative routing might also include some form
of legitimate traffic audit. So we perhaps reach a point where if P2P traffic
from your software passes some kind of “threshold” test of mostly
sending legitimate files (something that deep packet inspection might still
be needed for) then the APIs to sense the condition of the network are open
to your client software, and it is pushed higher up the food chain in terms
of the priority attached to the traffic. If mostly copyrighted material
appears to be traveling across the network, then perhaps that API
co-operation is refused by the network nodes and the resulting traffic
packets will be treated as low priority. That would create an underclass and
upperclass of P2P clients, each with a signature which would trigger the
various treatments by ISPs.
Now that all sounds fine and dandy, except that much of this traffic is
encrypted, and one P2P client can be made to emulate any other, and can
re-establish itself in different ports once it is identified and slowed, so
there would be technical hurdles, but we believe that there would remain a
class of P2P players that are working with the ISPs, and class that is not.
What that creates politically is an accelerate acceptance of P2P activity for
the average ISP customer.
Regardless, it is encouraging to know that Verizon at least is looking to a
future when the FCC might make it illegal to indiscriminately block or slow
P2P traffic, and instead is thinking about how to make the internet turn into
one huge TV set, sending Gigabyte class video files to every home. While so
little of Verizon’s revenues currently rely on video delivery compared
to say Comcast, its sworn enemies the major cable operators perhaps may not
feel able to embrace this approach, and this will accelerate a drift towards
using the RBOCs as ISPs rather than the more expensive and more restrictive
cable cos.
In the end we would expect that protocols and APIs that comes out of this
work will have to be a standard, and one that EVERY ISP, regardless of what
their main business is (cable or telco) will be forced to offer it. Either
customers will begin to leave in droves as word gets out that P2P goes faster
on other networks or the simple fact that those networks that don’t
wish to co-operate will still be faced with a day by day war trying to keep
P2P at bay, and will be still suffering the traffic consequences of badly
saturated networks. Comcast is supposedly supporting this new P4P activity,
despite the accusations about its excessive “traffic shaping,”
activities.
There is the feeling that the ISPs are the companies that need the technical
help to make this happen, not that they are being begged by P2P software
suppliers, because the P2P players seem to be winning the technology war
here. Perhaps it is the ISPs that need the P2P players’ help, not the
other way around.
CourtesyRethink Research, publisher
of Faultline, a weekly feature on technology and innovation.
Faultline: The rise of the P4P architecture to turn the Internet into a big TV set
