Peer-to-Peer VoIP

By Richard Grigonis

The are three basic levels of “peering” in the telecom universe: 1) VoIP Peering (also called Voice Peering) wherein one VoIP provider forwards calls to another directly without crossing over to the PSTN; 2) peering at a middle level where an enterprise peers and “federates” data with its partners and/or customers; and finally 3) “peer-to-peer” IP phones that plug into a corporate LAN and coordinate amongst themselves to become a voice system. A fourth “semi-rogue” level is where one finds gamers and peer-to-peer file sharing by collectors of music and other files, a process that eats bandwidth and can interfere with the first three if done in an inopportune setting (e.g. the office).

At the highest level of peering, carriers enjoy use of the Voice Peering Fabric, established in New York in 2003 by Stealth Communications (http://www.stealth.net). It’s basically both a telephony traffic exchange environment (via the VPF Minutes Market and VPF ASP Market) and a distributed private Ethernet network for service providers and enterprises. The VPF enables members to directly connect to each other, avoiding needless delays.

VoIP Peering is facilitated by such technologies as ENUM (TELephone NUmber Mapping, also known as “Electronic Numbering”) a protocol suite that unifies E.164 format telephone numbering system with the Internet’s DNS addressing system so that VoIP users can be called from both IP and PSTN networks. (See the article on ENUM elsewhere in this issue.) In April 2004, the VPF ENUM Registry was introduced to route phone calls among one another directly across the VPF without traversing the PSTN.

The VoIP managed-service provider VoEX (http://www.voex.com) also uses ENUM to map the world’s phone numbers to resolvable IP addresses, combining a carrier-grade VoIP peering infrastructure with what they claim is the world’s biggest Carrier ENUM registry (over 250 million phone numbers) that interconnects IP, TDM and hybrid-network service providers at very low prices.

Community Congestion?

BlueNote Networks (http://www.bluenotenetworks.com) is known for its SessionSuite Business Communications Platform that delivers SIP-based voice, video and other real-time interactive communication services to extended enterprise users as an integral part of an enterprise IT applications architecture or Service Oriented Architecture (SOA).

Hal Clark, Senior Product Manager for the telephony side of BlueNote’s product offering, says, “Ultimately, given the technology of IP networks, conceptually at least, direct communications on a peer basis makes a lot of sense, in particular for individuals working at a distance who don’t need a lot of services or if the services they need are sufficiently compartmentalized for easy distribution. Where the peering approach begins to get congested is when it’s applied to the idea of ‘communities’, in particular enterprises that need many more services than just a raw, basic peer-to-peer system found running over a corporate LAN. These peering communities are generally served by a collection of services. It’s like network voicemail. Peer-to-peer is nice as long as you’re online all the time and your device can take messages, or people can IM you in the middle of the night.”

“We’ve had solutions for those discontinuities in the past but they’ve been related to service providers or PBXs or something like that,” says Clark. “Those types of communities where some service or some social affiliation defines what that community is, is something that I believe will survive. But how those communities interoperate becomes an interesting issue. To me it’s very much like the way the American telephone system operated before the consent decree decades ago that established AT&T as a monopoly so that people could literally talk to each other. In earlier days you signed up with a telephone company and you could talk to people as long as they also subscribed to phone service with that particular phone company. The individual companies didn’t know or didn’t want to interconnect their networks. To a certain extent that situation exists now with many peer-to-peer communities. They’re either using proprietary protocols or just a service provider-managed VoIP type of service, which may or may not actually be strictly peer-to-peer in nature. In any case, the key issue is about how those communities talk to other communities.”

“Ultimately, that situation can ‘flatten out’ given that all of these peering communities are on the same network, so it should be a lot easier technologically than it was with the early circuit switched voice network to get them to interoperate or interconnect,” says Clark. “There still has to be some kind of infrastructure in place to bring that about through a transition period until normal network mechanisms such as DNS and directory services can help make a true peer-to-peer federation actually appear among these communities.”

Greg Pisano, Blue Note’s Director of Market Development, says, “We think of an enterprise as a community largely built around certain services. The enterprise has its own ideas as to how they want to enable their employees to work productively within an IP type of environment and give them the ability to then federate with these peer-to-peer communities directly as opposed to having everybody communicate by going out through a PSTN gateway and come back in via old legacy telephony approaches. There are of course cost implications of doing it that way.”

“An interesting option is for the peering exchange between communities - or even enterprises - to actually occur at the uppermost carrier peering level,” says Pisano. “Today, the carriers are basically collecting all of this traffic ‘within their walls’ and they try to control it and force it to flow through their ‘metering points’ or ‘toll booths’, as it were. In reality, these are just switches sitting at Internet peering points. Some VoIP carriers are already beginning to exchange traffic across those particular points. The technology allowing that to happen in theory should also be accessible by other communities to exchange traffic as well. That would reduce PSTN conversions and change the cash distribution among the service provider community. That’s exactly what the lower-end peer-to-peer market is trying to do to. After all, why does it cost so much money to make a phone call, when it now has a fairly low value when you consider how much the call really costs?”

Forging Quality Connections

Ubicom (http://www.ubicom.com) is a technology supplier that provides both chips and software for various home applications. They use a highly optimized architecture in terms of both hardware and software. They employ a multithreaded processor and some very compact, fast code for doing things like gateways and endpoint devices such as audio and video players, not to mention VoIP products such as ATA (Analog Telephone Adapters).

Keith Morris, Vice President of Marketing, says, “We’ve been trying to improve the quality of experience for all of these next-gen applications in the digital home. VoIP is one such app, video and online gaming being another. All of these applications are very sensitive to packet latency and loss. An early problem we solved was to look at WAN uplinks and to figure out how you guarantee that things such as VoIP and online games don’t get mangled by many other applications such as email, web browsing and also peer-to-peer file sharing. We found that there were a lot of issues in the market. One was that a lot of equipment out there didn’t scale very well as you increased the number of connections through the boxes, so we could see a huge drop-off in the performance of the overall box once you started opening lots of connections. Fortunately, our equipment can now support tens of thousands simultaneous connections without any degradation in performance.”

“We also looked at how all of these different applications interact with each other,” says Morris, “and how you completely, automatically, without any user intervention, find those applications that are latency-sensitive and give them high traffic priority, then find those interfering applications such as peer-to-peer file sharing and move them into the background.”

“All of this technology can be found in our StreamEngine portfolio brand,” says Morris. “ So, we’re about putting intelligence into the customer premise equipment [CPE] in the home such that all of these next-gen applications will work seamlessly without any user configuration, and, for that matter, without the need for a service provider to come in and try and manage a user’s application in the home. Service providers are good at providing end-to-end connectivity and deliver services end-to-end. But we’re now in an interesting environment where content isn’t just going to be emanating from the service provider - there will be many sources of content both from within and without the home. The next generation of CPE equipment needs to adroitly manage traffic that’s both service provider-related and user-related. And that’s what we do.”

Peering will be a fact of life for businesses both large and small as time goes on. But whether this will lead to a healthy business environment or a congested network of “walled gardens” remains to be seen. IT

Richard Grigonis is Executive Editor of TMC’s IP Communications Group.

» Internet Telephony Magazine Table of Contents