Comments on: Verizon touts smart P2P software

Company will reveal test results indicating it could reduce the impact of peer-to-peer traffic on its network by more than 50 percent by "intelligently" routing such traffic.

[xnowimcoolx]

advancement

with HD movie and tv downloads as well as streaming video and online gaming these internet providers are going to have to suck it up and find ways to accommodate high usage users.

And as a high bandwidth user I'll be the first to leave any company that throttles my bandwidth.

[umbrae]

But what if...

you want to get that packet from Singapore because the New York location is actually an illegal FBI dragnet. Users should have control over location management to avoid illegal practices of the FBI and other agencies.

[bluemist9999]

Sounds like a win-win to me

Dramatically lessen the impact on the network and improve performance for the users of the service.

Much better than Comcast's heavy-handed "block P2P traffic" approach.

[Rants&Raves]

Not the ISP's business

The ISP should not mess with protocols; developing reliable fault-tolerant trust-based protocols is hard enough as it is without having to deal with a myriad other custom tools messing with the data flow, each with their own bugs and limitations.

The best way to deal with this is to share the details with software developers so that they can improve their own code, AND allow for network topology discovery at the subnet level. This way, every tool can optimize its own traffic (a speed-up is a significant goal for P2P software developers) and every competing application can come up with its own enhancements and optimisation ideas.

[Maggie Reardon]

Not ISPs business

Actually, that is what the ISPs will be doing. They are not doing anything to the protocol. All they are doing is providing more information about their customers and their networks. Companies like Pando will then use that information to enhance their content distribution software.

[aryachaty]

0.89 hops?

How can the number of hops be less than 1? Hasn't it got be be a minimum of 1 hops to get to another computer?

[Maggie Reardon]

0.89 hops

To answer the question about the 0.89 hops, I asked Doug Pasko of Verizon to clarify. He said that in the study they counted the first hop as the hop that routed the traffic outside the local or regional network. I hope this helps clarify.

[laird666]

The 'hops' in the analysis are long distance links

The 'hops' that we analyzed are long distance links between
metro areas, not router hops. The long distance links are the
scarce/expensive resource - router hops aren't a good measure
of delivery cost, and the field test's goal was to measure the
impact of P4P on data delivery cost.

The amazing statistic to me was that with P4P same-metro data
went from 6% to 58%! The fact that data delivery can be localized
this well is what leads the the delivery cost reductions and
performance improvements that we saw.

There's more information on P4P at
http://www.pandonetworks.com/p4p and at
http://www.dcia.info/activities. Participation in the Working
Group is open to all p2p companies, ISP's, and researchers, so if
you're interested you can contact co-chairs [e-mail] or
doug.pasko.com, or the head of the DCIA, the host
organization, [e-mail].