Following a second round of real-world network tests in the past year, the P4P Working Group has formalized its structure as a non-profit corporate alliance, laying the groundwork for commercializing a technology that participating telco, cable and Internet firms say reverses the perception that peer-to-peer (P2P) media traffic necessarily damages broadband network efficiency.
Informed sources say the formation of P4P Working Group Ltd. may reshape Net neutrality debates, but it won’t necessarily slow momentum toward another Congressional attempt to write new rules as the Obama administration formulates its policies governing network operators’ abilities to manage traffic and network congestion on their broadband access networks.
P2P traffic, both legal and illegal, has presented a particular challenge for broadband access network operators, because a handful of end consumers using BitTorrent or other P2P client software to up- or download music or video content can generate so much traffic as to degrade performance for most or all other users in the same serving area. Comcast found itself in trouble with Net neutrality advocates and with the Federal Communications Commission a year ago when it constrained P2P media traffic that it said was hurting performance for a majority of its end users
The evolving P4P protocols, initially developed by Yale University researchers, transform the way P2P content delivery works by enabling a video or other stream to be assembled from peer devices on a single, rather than multiple random, networks. P4P advocates say this process can significantly alleviate P2P traffic congestion for any cable, telco or wireless network operator.
“The second round of P4P trials again illustrated how ISPs can operate more efficiently if traffic remains on their own networks,” said Doug Pasko, co-chair of P4PWG and senior technologist for Verizon. “The P4P Working Group’s formal structuring will facilitate the completion of work on a final specification for P4P, which will represent its next major step towards commercial deployment.”
Companies represented on the board of directors of P4PWG, which now claims more than 100 participants, include network operators AT%26T, Verizon and Telefonica and P2P-assisted content delivery network (CDN) operators Abacast, Akamai, Pando Networks and Solid State Networks. Yale University’s Richard Yang also is on the board. Comcast, which participated in the latest field trials along with AT%26T, Verizon and an unnamed European operator, also is involved in P4P working groups, as are additional vendors, including Cisco Systems and Microsoft.
A next wave of field trials involving more P2P and other players will come as early as March 2009, according to managed P2P-assisted content delivery platform provider Pando Networks, which first collaborated with Yang and other Yale researchers and has helped drive the trials and formation of P4PWG.
“In the random P2P world, people viewing a show or downloading a game via P2P protocols may get data across all kinds of networks, but P4P at its core localizes delivery, so Verizon people get it from people in the Verizon network and in the same metro area,” explains Pando CEO Robert Levitan. “The reason this is so important is, one, it obviously reduces the distance the data travels, and two, that in turn increases delivery speed for the consumer and also decreases operating costs for network operators. You end up with faster downloads and lower costs for the operator network, so everybody wins. This is kind of the exact opposite of what P2P was in past: longer, random distances and connections and higher costs.”
Noting that the Internet is fast becoming the media distribution platform of the future, Levitan adds that users of Web media services like Netflix “want everything available all the time, and to make that happen, you need innovations like this. That’s why it’s so important. We’re also debunking a common perception that P2P was bad for network operators. It turns out P2P can be extremely beneficial for network operators. In the past, operators have said it’s swamping their networks and causing costs and we need to block it. Now they’ve found a way to see it is efficient and good for network performance.”
According to Levitan, the second round of trials in late 2008 validated P4P effectiveness in cable, as well as telco, infrastructure, “and a significant improvement in P4P algorithms produced a big jump in efficiency, in particular around optimizing transfer of data between ISPs. The first trial was limited to single cities. The second test added traffic from adjacent ISPs, which also tended to improve efficiency,” he said.
Perhaps most important in the Net neutrality context, “P4P doesn’t treat P2P differently,” says Laird Popkin, chief technology officer for Pando and P4PWG co-chair. “It just makes smarter network routing choices, but it’s not treating P2P better or worse” than other application or traffic types.
In the meantime, network operators continue to address performance damaging network congestion in various ways. On one side, Verizon itself has partnered with P2P-assisted CDN operator Velocix, which is building a CDN with P2P servers concentrated on Verizon’s own network – an approach that is in line with P4P practices. Velocix is participating in a P4PWG task force on interoperability.
On another side, cable operator Cox Communications announced that it planned on February 9 to initiate a traffic management method in several of its markets that assigns top priority during peak usage periods to activities that are time sensitive, such as Web pages and streaming video, and lower priority to less time-sensitive operations, such as software updates.
Net neutrality advocates immediately suggested that such practices as Cox’s raise policy issues. “As a general rule, we’re concerned about any cable or phone company picking winners and losers online,” say Ben Scott, policy director of Free Press. “These kinds of practices cut against the fundamental neutrality of the open Internet. We urge the FCC to subject this practice to close scrutiny and call on Cox to provide its customers with more technical details about exactly what it’s doing.”
In the case of Verizon and Velocix possibly leveraging P4P to offer CDN services that finally gain a revenue stream from over-the-top P2P traffic, FreePress offers no objections, because the P4P approach does not discriminate among traffic types. “Ultimately, P4P would be analogous to a smarter highway system where traffic communicates with the road ahead and tells the driver the best route to his destination,” says Robb Topolski, chief technology consultant for FreePress. “It isn’t the road behaving differently, but the driver making better decisions based on better knowledge. It’s not that P4P traffic gets to go first.”
By contrast, FreePress questions Cox’s approach because “we object to the premise that you can tell which traffic is time-sensitive,” Topolski says. “For example, supposedly one non-time-sensitive protocol is FTP (File Transfer Protocol), but real-time police and other systems communicate in FTP. It’s been a convention of Internet design that protocol is separate from priority, and we have an operator ignoring that convention and not participating in IETF [Internet Engineering Task Force], which we’re fine with; just don’t do it on the Internet.”
At the same time, Net neutrality advocates and the FCC itself also have objected to revised network management practices initiated by fellow cable operator Comcast – practices that do not discriminate between P2P or any other traffic types, but do discriminate among heavy and light users. Comcast now resets specific heavy users’ priority assignments only when they exceed normal consumption at times of overall serving area congestion. In January the outgoing FCC of the Bush administration sent a letter to Comcast questioning whether it is using that management system to favor its own voice over IP (VoIP) service. Comcast responded that its VoIP service does not travel over the public Internet, as do P2P and other Web services.
Judging by support from such large operators as AT%26T, Comcast, Telefonica and Verizon, P4P does appear to hold promise for moving renegade P2P traffic down the list of traffic congestion sources. Whether applying P4P algorithms to more and more Internet traffic would largely moot the festering arguments over Net neutrality versus operator traffic management rights remains an open question.
For one thing, a Comcast or Verizon would be positioned to provide P4P-based CDN services to an emerging market of legitimate P2P video and music distributors because there would be an identifiable video, music or game distribution company with which to negotiate. However, in the case of millions of extra-legal BitTorrent P2P distributors who do not enjoy legal contracts with the content owners themselves, network operators could be left in a legal conundrum. If the operator fails to manage the illegal content traffic, notes one source, it could be libel to litigation brought by the content rights holders for allowing illegal content distribution.
Stressing that he is not, himself, a lawyer, FreePress’s Topolski notes that, at a technical level, any attempts to build a network that “has to know what content it is transmitting” would create “an awful lot of overhead. You’d need a list of what is legitimate, keeping in mind that we create new content every day. So it seems like a recipe for disaster in engineering.”
Ultimately, he says, the P2P protocol alone is not a valid measure of what’s legitimate, and from the operators’ point of view, “It’s in the network’s best interest to offer P4P services. If they offer it and users find it a faster means to their ends then both sides win.”