(This article originally appeared in the November 2007 issue of Internet Telephony magazine.)
peering — there’s a lot of talk, but relatively little action. That’s probably okay, as VoIP peering is a telephony concept. In the world of Internet applications, we care about things like communities, endpoints and mashups. But to understand the real issues, we first need to resolve some confusing terminology and discuss some economics.
VoIP Peering vs. IP Peering
The term “peering” by itself refers to IP peering — the voluntary interconnection of two Internet networks for the purpose of exchanging their customer’s IP traffic. Operator A wants his customers to be able to reach Operator B’s (News - Alert) customers so they strike up a deal to exchange traffic, just for those customers, usually not for other Internet addresses. This differs from “IP Transit” where A pays B to deliver A’s customer’s traffic to any or all Internet addresses.
IP peering is about layer 3 packets — nothing specific to telephony. Peering between ISPs is almost completely unregulated and most peering arrangements are subject to non-disclosure agreements, so information on the industry must be derived indirectly, for example, by examining Internet backbone router tables. Nonetheless, the market seems to work extremely well. In 12 years of commercial Internet experience, no backbone monopolists have emerged and new backbones (both public and private) continue to be built.
VoIP peering (or voice peering) is a newer term. It refers to the direct exchange of VoIP telephony traffic without PSTN involvement (or translation). VoIP peering is all about telephony addresses (phone numbers or SIP
URIs). In the PSTN, this is called “interconnection” and it’s usually regulated, with mandated settlement rates.
Economics and Network Effects
A network is only valuable when it connects you to what you want, thus network value depends upon network scale — the more members, the more valuable the network. And when two networks interconnect, the smaller network derives the most benefit as its customers get proportionately more reach. Barring regulation, if payments are made, the smaller network typically pays the larger for connectivity.
Of course the first interconnection any new VoIP provider needs is connection to the biggest network in the world, the PSTN
. Except for regulatory loopholes, this has real cost, as the PSTN is large (many interconnection points) and subject to complex, per-country rules.
Once you have PSTN connectivity, the decision to do voice peering with another VoIP provider depends entirely on cost. Do my subscribers make enough calls to subscribers of the other VoIP provider to justify the expense of setting up interconnection? In most cases the answer is no. Any one VoIP service provider is so small that it’s hard to justify the interconnect expense.
But suppose both VoIP services actually did more than PSTN voice telephony? Wouldn’t that justify VoIP peering? Maybe. . .
Digital POTS vs. IP Communications
Today there are two very different kinds of VoIP players. Arbitrage services, like Vonage, AT&T CallVantage and, more recently, the cable companies, use VoIP technology to produce a PSTN experience at lower cost - think of this as digital POTS (Plain Old Telephone Service).
IP communications services leverage IP to provide new capabilities, like instant messaging, wideband audio and video. Skype is a leader here, but AOL (News - Alert), Yahoo, Google, MSN and many others are in play.
Doesn’t it make sense for two services like Skype and Google (News - Alert) Talk to interconnect? Both services provide better than telephony voice quality through the use of wideband coders (at least on Skype-to-Skype or Google-to-Google calls). Interoperation between Skype and Google Talk could allow better audio on such calls.
But the Internet is not the PSTN; new rules apply.
IP Communication is Different
Most IP endpoints are intelligent devices. The Internet’s end-to-end design principle encourages this. Today, most are PCs, but even mobile phones and other mobile Internet devices incorporate relatively powerful processors. Furthermore, Moore’s law ensures a stream of ever-more intelligent devices.
IP communications involve relatively sophisticated applications running on connected devices (PCs, mobile phones, etc.), which themselves are platforms. As a result, it’s feasible to run two or more IP communications applications at the same time. Certainly that’s the case with PCs, where it’s not unusual for a teen to use three different instant messenger programs at once and be active on MySpace (News - Alert), Facebook and the like.
These are separate networks. They don’t peer, but equivalent functionality happens at the edge. Each user runs client software for each of the communities they participate in. Thus all their contacts are visible and reachable, despite the lack of peering relationships between the services.
Enabling Mashups More Important than Peering?
Today, the competitive edge for Internet applications is how open they are for third parties to extend and incorporate in mashups. When Facebook opened up to third parties earlier this year, numerous third party extensions appeared and membership soared. This is what counts.
VoIP peering is for digital POTS. IP Communications providers need to focus on open APIs and enabling mashups.
Brough Turner (News - Alert) is Senior VP of Technology, CTO and Co-Founder of NMS Communications.