As Voltaire said, “If you wish to converse with me, define your terms.” (He later admitted he was actually quoting John Locke, but that’s another story.) Of the many definitions of VoIP peering, John Longo of Heavy Reading proposed the following: “VoIP Peering is the interconnection between the originating [the one with the calling customer] and the terminating service providers for the completion of a seamless session between subscribers on their networks. Third Party resources — transmission, switching, database or otherwise — may be used to facilitate the session connection.”
When talking about VoIP Peering, we must remember that it is a two-sided coin. One side involves service providers directly connecting with other service providers. The other involves enterpriseto- enterprise peering which reduces service providers to simply providing a dumb pipe over which operate the applications managed by the enterprises themselves.
In both cases, peering eliminates the involvement of PSTN and multiple unnecessary IP-to-TDM and TDM-to- IP conversions. There’s a strong belief that by cutting out multiple parties (“middlemen”) to handle a call passing through the PSTN, you will lower your termination costs. Peering does not mean free, however. That’s a separate commercial discussion, particularly when two service providers are peering with each other.
Potentially, since we’re talking about a flexible IP network, a service provider or carrier could come up with new IPbased services in a peered environment, which can generate new revenue streams, and so on.
Any way you look at it, there are reasons to establish peering between enterprises or between service providers.
Even so, VoIP Peering is not as simple to do as generic IP Peering. The rules of the game for IP Peering are much simpler: An IP address is never ported from one person to another. Instead, an ISP is assigned a range of IP addresses, and no one can pluck a number out of the middle of this range and take it somewhere else. Hence, “addressing” in the Internet is very clear and well-specified. Moreover, the Internet allows anyone to reach anyone. Any security and access control are the responsibility of the user and their browser — they decide whether or not to look at a particular website.
VoIP (define - news - alert) Peering differs in that the address — which is the telephone number — is not necessarily in an “official” range. The number could have been ported, as are many fixed line numbers. Voice service providers aren’t conventionally-regulated entities that are assigned numbers, so there is no official source of information on what telephone numbers a particular service provider operates and maintains for its customers.
Another aspect is that most VoIP end users expect the providers to manage security, which means that they don’t want unwanted calls in the middle of the night from some spammer offering Viagra. Instead, users expect their service providers to somehow manage such a problem out of the system, which obviously is not the case with the Internet in general. This alone makes the establishment of VoIP Peering connections much more complex than they are in the IP Peering world.
For peering among service providers, the following sequence of events usually occurs:
1. Provider A provides its telephone numbers and routing information.
2. Provider B loads that information into an ENUM [Electronic Numbering] transaction server.
3. An outgoing call from a customer of Provider B queries the ENUM server for matches to the phone number.
4. Successful matches provide the routing information.
5. Calls are routed over an IP connection to Provider A, which terminates it.
In this scenario there is an implied need for each provider to supply telephone numbers and routing information to every other provider or carrier with which it wants to peer, and to keep that information up-to-date.
Bye, Bye, Middleman
Major carriers are now getting their feet wet in the VoIP Peering world. For example, Global Crossing (http://www.globalcrossing.com) has hatched some VoIP Peering solutions for both carriers and enterprises.
Anthony D. Christie, Chief Marketing Officer and Executive Vice President of Global Crossing, says: “From our standpoint, we’re very enthusiastic about VoIP Peering, because it plays right into what created our network for — to be disruptive to traditional price models and industry structure models of the big incumbents without having an irrational effect on our balance sheet. We’ve been dealing with VoIP for seven years now, and we have appreciably all of our minutes on our VoIP backbone today.”
Christie elaborates: “Last year we launched two offers: our VoIP Community Peering offer for Enterprises and our VoIP Community Peering offer for Carriers. We went down a path with SunRocket (News - Alert) that we’ve announced. SunRocket is a leading U.S. Internet phone service provider and together we’ve implemented a high-performance VoIP peering solution yielding excellent end-to-end IP connectivity for IP services, and of course it bypasses the usual usage fees associated with completing calls over the PSTN. We’re also doing this with a couple of other players, but we haven’t announced them yet. They all involve what I’ll refer to as a kind of ‘virtual peering fabric’, where we’re taking, say, SunRocket’s actual dial plan and we’re storing it in our softswitch. SunRocket is using our outbound VoIP services and our DID local services, so for virtually any customer that’s calling another customer in the SunRocket universe, as that call traverses the Global Crossing network, it’s an unmetered call — an ‘on-Net’ call, if you will.”
“We’ve also done a deal with Stealth Communications that essentially connects their network to ours,” says Christie, “so that carriers and enterprises can have a much broader footprint to connect to Stealth Communications’ VPF [Voice Peering Fabric, an ENUM-enabled, distributed Ethernet fabric that allows providers to buy, sell and route voice minutes and telephony related-services among themselves by directly connecting to each other, avoiding unnecessary delays and relays.] and through Global Crossing’s network. So we’re interconnected with Stealth but if there’s a player, enterprise or carrier that wants to connect to Stealth but a node is too far away, or there are questions around connectivity, mediums, and things of that nature, then they can come to Global Crossing and we can give them an MPLS connection right into the Stealth VPF. We’re also exploring to see if there are other ways for us to leverage some of the stuff Stealth has done. They’ve got a pretty large ENUM database and they’ve got a pretty fastgrowing business model.”
“In the longer term,” says Christie, “we really don’t see the role of the intermediary being as strong as it is today. The reason that the intermediaries and the VPFs that are there do exist is because there is an absence of a concrete standard around ENUM and we actually see, in the intermediate- to-longer-term, a model very similar to what the Internet peering model is today. We have Tier-1 peers; Tier-1 is categorized based on the breadth and quality of the network and whether or not it pays for peering. There’s a sort of self-enforcing nature, if you will, of the Internet as we know it today, and we see the same thing evolving in the VoIP Peering world. While intermediaries are an important stop right now to get the energy and access around VoIP going, they’ll probably play a lesser role in the longer term. That’s how we view it. We’re extremely bullish on the concept.”
Where is Everybody?
Another major player striding into VoIP Peering arena for service providers is Arbinet (News - Alert) (http://www.arbinet.com). Arbinet (news - alert) provides a huge marketplace through which service providers can buy and sell mainly voice minutes. Their 693 customers include all top ten international carriers, eight of the largest prepaid calling providers, and they trade about 1.2 billion minutes every month, 44% of which are mobile and 30% of which are VoIP. They also convert about 330 million minutes every month between IP and TDM networks. They’re probably one of the most interconnected companies in the world.
Their basic model is that a service provider connects to one of their POPs using TDM or VoIP, and they buy termination from Arbinet, which then turns around and finds the appropriate sellers to provide that termination, routing the call between the originator and the terminator across Arbinet’s platform.
Surprisingly, Arbinet discovered that the whole world was not yet beating a wide path to the doors of VoIP Peering providers. Steve Heap, CTO of Arbinet, says, “Interestingly not a great deal of VoIP Peering is taking place. It’s not yet an enormous opportunity to many service providers. The first reason is that if you look at the U.S., about 0.25 percent of the total traffic is VoIP-to-VoIP, and so if every provider peered with every other provider, at best only 5 percent of a provider’s package would be peered. Second, if you look internationally, it’s even harder to justify investing in the infrastructure for VoIP Peering. What’s the interest in peering between some VoIP customer in Spain and another in Finland? Not a lot. And third, it’s hard work. You’ve got to identify partners and you’ve got to trust them with your numbers, because you don’t want them to phone up your customers and persuade them to move to their service. Do you trust someone who lets customers sign up for free? Because, potentially, that’s where the SPIT would come from.”
“Moreover, the provider still needs to agree on a commercial interconnect with its peers,” says Heap. “It needs to agree about IP addresses, it needs to set up transport, exchange numbers with the peering providers, route test calls, do a query of outgoing calls, and if you may have to be prepared to do bill-and-settle.”
“So, it’s not a slam-dunk,” says Heap. “Providers don’t say, ‘Oh, let’s peer!’ since there is a whole series of things you’ve really got to do behind the scenes.”
“There are alternatives out there. There are peering communities, such as XConnect (News - Alert) [a huge neutral provider of VoIP peering services and operator of the XConnect Global Alliance, which recently launched their DirectRoute service, enabling wholesale carriers to profit from Layer 5 VoIP Peering] or the Stealth Communications service called VPF [Voice Peering Fabric] to some extent. If a provider joins one of those, it gets one connection to the community and it gets access to everyone else. Such peering communities can provide number sharing mechanisms. Some, like XConnect, provide a measure of security.
But, as Heap explains it, VoIP Peering Communities have their Pros and Cons. Here’s his list:
• One connection to the community gives you access to all other members.
• Provide number sharing mechanisms via private ENUM Registry.
• Often provide signaling relay functions and some security.
• Must agree to the rules of the community.
• Zero termination fee for accepting calls to your customers.
• Must accept calls from all other partners.
• VoIP only — so community of interest is still limited.
• No ability to organize direct peering with a partner using private ENUM system.
• No ability to reach members of another community.
• Your numbers are in a privately held company’s systems.
“Arbinet thought it too could create a community,” says Heap, “since everybody trusts Arbinet and we’re a neutral, independent company. But as we thought about it, we asked ourselves if the industry really wants lots of little isolated communities popping up. Instead, we came up with what we think is a better idea. We first asked ourselves, ‘If VoIP Peering has such clear benefits, why restrict it to VoIP?’ After all, VoIP is only a technology, not a lifestyle. So, why not form a peering community between service providers that’s independent of the technology used to reach the end user? When you do that 100% of the peered traffic can travel between service providers.
"There are still no intermediate carriers and no unnecessary conversions, so the quality is at its best. Additionally, each service provider can now evolve at their own pace to VoIP. Moreover, the service provider who ‘owns’ the called customer decides — as now — whether a termination fee needs to be paid — many medium and large service providers get a termination fee now and a change of technology won’t cause their finance group to say, ‘Oh, let’s give up this revenue stream’. So if they decide they want to have a termination fee then obviously they can maintain that. In a peering community that makes sense, you simply pass that termination payment from the far end to the originating party. You wouldn’t add your own margin cost to it.”
Heap continues: “There are other issues as well. How does a provider share numbers? How does it handle the querying of calls? How does it handle VoIP-to-TDM conversion? How about security, routing, transport, commercials and settlement? Thinking about how to deal with this also gave us more ideas when we at Arbinet developed Arbinet PeeringSolutions.”
“To share numbers, for example, we formed, about 12 months ago, a nonprofit, non-stock company called SPIDER Registry, that’s managed by an industry board,” says Heap. “The SPIDER Registry’s sole job is to enable the sharing of numbers and routing information. Once a service provider passes its numbers into SPIDER and decides to share those numbers with someone else, SPIDER then synchronizes the numbers and the routing information into an ENUM server belonging to that other player. There is no charge for creating an account, putting numbers in, setting up peering relationships or downloads, but there is a fee for each successful query against SPIDER-provided numbers which returns a VoIP address, in effect.”
Nuts and Bolts
As VoIP Peering increases among both enterprises and service providers, the infrastructure will have to handle yet another phenomenon, the rise of IMS (IP Multimedia Subsystem (News - Alert)) a common service architecture that’s being built for all the world’s wireless and wireline networks. Can all of the world’s network elements serve both these masters with their usual aplomb?
Over at NexTone Communications (http://www.nextone.com), Nick Turner, VP of Product Marketing, says: “There are specific rules in carrierto- carrier peering which need to be handled by a Session Border Controller on behalf of the IMS network. For enterprise peering, you have a different set of problems and complexities. You’ve got more variances or difficulties in the IP realm. Private IP addresses, public NATs, VLANs, ‘realms’, and so forth can cause problems. Softswitches and IMS networks really don’t have much in their plumbing to handle these things. Fortunately, an SBC’s job is to normalize things into something the IMS infrastructure can understand and relate to. And that’s in addition to the commonly understood security aspects that we all know and love about SBCs.”
At Cantata Technology (News - Alert) (http://www.cantata.com) one of the world’s leading independent providers of enabling communications technology, James Rafferty, Senior Product Manager, says: “At the 10,000 foot level, certainly the driver pushing companies into VoIP Peering is that, as VoIP is taking off, you have various ‘islands’ of VoIP forming. Traditionally, the way you’d connect up those islands would be through the PSTN and a series of exchanges would occur to accomplish that. One of the business drivers for VoIP Peering is to have the different VoIP players start to try and work together more so they can bypass the PSTN connection. In practice, it’s somewhat messier than that, but for a company such as Cantata, it’s a terrific opportunity, because we’re seeing a lot of need to basically tie together different types of signaling and media, and that’s what we do with our products.”
Scott Wieder, Cantata’s Director of Marketing Development, says, “One of the necessary evils in this area is transcoding. There was a theory that went around some time ago to the effect that everyone would settle on using just one or two voice codecs, such as something in the G Series [e.g. G.729] or the newer ILGC [Internet Low Bitrate Codec], and you wouldn’t really need to do transcoding as you sent traffic back and forth between carriers. But that’s not how things are playing out, not at all. There will apparently always be a need for some kind of transcoding. This is one area where peering service providers are calling on a product such as our Cantata IMG to do IP-to-IP transcoding without having to go through a TDM connection such as, say, back-to-back gateways, which is the old way that was done.”
“The wireless guys have their own codecs,” says Wieder, “and codecs such as ILGC are starting to appear and proliferate, and so there’s the need to have some ‘Rosetta Stones’ out there to translate between them. In a similar regard, if you look what organizations such as the IETF are doing in the peering space, most of the focus there is on SIP [Session Initiation Protocol (News - Alert)] and yet there’s still a lot of H.323 networks out there. So there’s a need to make good as far as signaling is concerned, too. This is an area where the SBCs [Session Border Controllers] can definitely come onto the playing field, because if there is a need to arbitrate between, say H.323 signaling and SIP, then that’s most typically being done through an SBC.”
“Even with VoIP Peering occurring, there are still many calls that need to connect to the PSTN,” says Wieder, “so we at Cantata continue to experience customer interest in our SS7 capabilities in particular, to help make those connections. SS7 capability has two aspects that this crowd is paying attention to: One is the basic SS7 connectivity, as can be found in SS7 ISUP [ISDN User Part], which is what we offer directly in our IMG 1010 Gateway (News - Alert). That’s a popular combination. Second, one of the reasons SS7 gets used is for local number portability and TCAP database lookups [TCAP, or Transactional Capabilities Application Part enables deployment of Advanced Intelligent Network services by supporting information exchange between signaling points] and stuff like that. Cantata has another product called the MSP [Multi-Services Platform], an application platform with a multi-layered SS7 stack allowing people to set up their own SS7 lookups and handle local portability lookups that traditionally have been the provenance of SS7 technology.”
“The VoIP Peering guys have a lot of interest in doing these types of lookups on the IP side through something like ENUM,” says Wieder, “so we’re getting interest from companies in finding ways to start managing the transition of that kind of data going from SS7, where it is today, over in the direction of IP, particularly where two conversing parties are both on VoIP networks.”
“But the biggest thing about peering is that, in practice, it’s mostly about business,” says Wieder “You don’t really have peering until you have a business arrangement. So this is one of the reasons why things such as the VoIP Peering Fabric [VPF] have been coming along and been enjoying some popularity, because they offer an ‘umbrella’ or a ‘dating service’ for carriers to get together and then essentially enter into a bilateral or multilateral arrangement called a federation. This sets the rules of engagement as to how the actual business side will work. Once the business side is taken care of, the actual technical connection between carriers can take place.”
VoIP Peering has yet to explode, but it is sizzling. Within a few years it will be as commonplace as IP Communications itself.
Richard “Zippy” Grigonis is Executive Editor of TMC’s (News - Alert) IP Communications Group.