ENUM's Alternate Route

In the communications business, almost everything is always about the money. Sometimes that isn’t such a good thing. But it’s a pretty good rule to follow, at least from the standpoint of trying to understand why things move the way they do.

So consider ENUM (electronic numbering). Up to this point, voice industry executives have been noodling on the reasons why it does or does not make sense to join any of the existing voice peering fabrics.

Business model issues, more than anything else, have driven most of the thinking, says Brian Cafferty, Vero Systems CEO. Carriers that make serious money providing access might not be so interested in giving up that revenue stream by “peering” with other providers (a different matter from “interconnection”).

Intriguingly though, it may well be business model issues that drive lots of industry players to embrace ENUM, though possibly in ways ENUM backers haven’t considered so far.

And to the extent there is a business model difference between “peering” and “transit,” as opposed to the mere matter of network interconnection, some providers might not feel any particular pressure to “peer” at all.

AT&T’s I program, where all AT&T plain old telephone service customers have free access to all other AT&T telephone numbers, free and on network, creates a potential peering fabric of 100 million numbers, notes Hunter Newby, telx chief marketing officer. Throw in Verizon and Qwest numbers and nearly the entire U.S. telephone market instantly is “peered.” The business corollary might be that everybody else is a “transit” customer.

Conversely, VoIP providers have lots of incentive to cut their costs precisely by avoiding payment of those fees. But some carriers still are uncertain they can save money by joining a peering fabric, says Gary Coben, deltathree director. Others are just as adamant that independent VoIP providers have no hope of challenging the incumbents unless they do so, as VoX Communications president Mark Richards argues.

And that’s just the bearer traffic. Not much yet as been done in the area of application peering, notes Natan Tiefenbrun, XConnect chief operating officer.

In the time division multiplex world, of course, there are other ways to reduce operating cost, including but not limited to least cost routing and other optimization techniques. And you might think that when calls originate IP and terminate IP, that there’s no need to pay special attention to routing.

Vero Systems and VoEX disagree. VoEX runs the SuperRegistry, a peering registry that today includes 20 percent of U.S. wireless destinations and more than 40 million numbers, the company says. Some “tens of millions” of additional numbers are housed in registries with which VoEX interoperates, the company says.

So where does routing enter the picture? The SuperRegistry provides interconnection and transport services, including call signaling and media translation, as well as ENUM registry capabilities.

So think of advanced routing capabilities as the front end to an ENUM registry, the same way access and transport infrastructures are the front end to voice services and features. That’s where Vero Systems comes in. Again, you might think there is no need for routing when an ENUM registry is used to connect IP end points.

“The challenge with hybrid networks is how you decide which network you are peering with, and what the paths are to that telephone number,” says Charles Studt, VoEX director.

Consider calls that originate using a dual-mode phone, for example. “Wi-Fi network routes might really change from call to call,” says Cyril Matthews, VoEX director.

The routing has “almost a mobility manager kind of function,” he says. “Things aren’t so static anymore.” A call from a handset at a hotspot might route one way, then another on subsequent calls. They obviously will route differently as soon as the phone flips back to mobile network service. “It might be cheaper to route one way compared to another,” says Matthews.

But in a world where multiple ENUM registries might exist, that is not the case. There are several problems. “Registries are popping up everywhere,” says Cafferty. “How many registries can you afford to dip and not exceed the timers of the switch?”

So latency is going to be an issue if multiple ENUM registries must be interrogated before a call delivery decision is made. There might in fact not be time to do a complete dip of every registry a calling number is a member of, in other words. Then there remains the matter of routing packets over the originating number’s network, to the recipient number’s network, and any other networks that must be traversed in order to complete the session.

So there still remains an issue of what is the most efficient way to terminate any single ENUM-supported voice session, because business rules remain in place even for private IP networks. Not all deals are going to be settlement free, says John Fitzpatrick, Vero Systems senior vice president.

Cable or mobile operators might have different business rules about terminating traffic to members of their peering fabrics, for example, says Fitzpatrick. So think about that for a moment.

Those among you who spend way more time than I do following such things may have a better grasp of the matter, but I do not recall that access charge obligations apply to cable or wireless providers, though other mechanisms are in place for universal service, 911 system support and other social obligations.

So might optimal cost routing and ENUM fabrics help large tier one wireline and wireless companies create new revenue streams? Specifically, might they be able to create some form of compensation for call termination on wireless or cable domains?

One might simply note that wireless voice average revenue per unit is dropping, wireless terminations are growing, and wireless and cable providers, though major players in the voice business, do not benefit from some of the termination mechanisms that contribute to wireline voice revenues.

Whether or not that ever is the case, ENUM does increase routing granularity, at least in principle. And the issue is whether service providers and end-user entities want to take advantage of the new levels of granularity to reduce costs, improve quality, provide new services or create additional revenue streams.

Traditionally, the question asked was “what is the cheapest way to terminate to that end office?” says Fitzpatrick. “Now it’s about how you get to an actual 10-digit phone number.”

“The unique thing is we can have a public edge server that multiple carriers can hit, or a private one only some carriers or one carrier can hit based on your payment of special rates,” says Studt.

In essence, this allows a carrier, service provider or enterprise to build a custom ENUM data base. That might be useful for “a lot of providers who can’t route to NPA-NXX levels,” says Matthews. “This cuts your routing function costs.”

That is sort of the whole point. Enterprises can use each other as competitive local exchange carriers, in terms of function. Calls go number to number, from my IP exchange to your IP exchange. Who terminates a call, in such cases? An enterprise originates, and another enterprise terminates.

Granular control of IP endpoint to IP endpoint routing will be even more important when video bits start to augment voice bits. “You need to do many of the same things for quality IPTV as you now do for voice,” says Cafferty.

In fact, Vero Systems is “working on a draft for ENUM standards bodies for some sort of cost-based augmentation of the protocol to support a cost factor,” says Fitzpatrick.

In fact, there may be other angles as well. “There’s a big play for TDM networks using AIN to dip an ENUM server and route using the TDM network,” says Fitzpatrick. “Lots of people focus on ENUM primarily for IP players, but it has implications for TDM providers as well.”

That might sound crazy, but Fitzpatrick argues that where calls originate in TDM format, a service provider might want to avoid protocol conversion to preserve quality. How it might work is that if a neutral peering provider has access to enough information on available routes, quality and cost over any particular path to any particular dialed number, it can make a decision on whether using a gateway makes sense or not.

In some cases it might make sense not to convert a TDM originated call to IP. In other cases it might. The parameters are not simply cost, but also quality and other business considerations.

One carrier might have a volume commitment with another carrier that needs to be fulfilled in a short amount of time. In that case it might make sense to route traffic over one network on a temporary basis, just to top off the quota, even if the cost is not the absolute lowest or the quality the absolute best.

“This is disruptive to the Vonages of the world,” says Cafferty. “Really, this is 10-digit bypass.”

On the other hand, keep in mind the difference between “interconnection” and “peering” as a business model. They aren’t the same thing. Networks must interconnect to be valuable. But business arrangements can vary, since the market power, assets and business drivers for various players are diverse.

AT&T needs to interconnect with other networks. It might not benefit financially from a “peering” arrangement. The reason is the words we no longer use but are implied when we use the word “peer.” Originally, the full phrase was “settlement free peering.” Today, we drop the “settlement free” clause, but peering in that sense is a business arrangement, not simply another word for interconnection.

Everybody benefits from interconnection. Not everybody benefits or benefits to the same degree, from settlement free peering. “Why wouldn’t cable companies want to make money for terminating traffic from a wholesaler?” Cafferty rhetorically asks.

That’s an interesting angle. There might be some service providers that do not today qualify for the equivalent of reciprocal compensation. But under the right set of business relationships, they might be in a position to get equivalent payments.

“It’s going to take time, but the routing becomes more complicated and sophisticated over time,” says Cafferty. “Theoretically, every 10-digit call can be routed separately.”

The reason is that there may be a business reason, or a cost driver, to do so. “Just because a number is in a registry doesn’t mean the cheapest or ‘best’ way (assuming there are other business considerations) to deliver the call is to use the first settlement free peering exchange where a match is found, he says.

In cases where a single service provider is a member of multiple exchanges, for example, there might be matches at more than one exchange. So what’s the best way to route a particular call? “Say you have a number in two exchanges, one with 5 million numbers and one with 50,000 numbers,” Cafferty says.

Assume the cost of terminating a call is literally the same at both exchanges. Which way do you want to route the call? “You send the call to the smaller registry because sheer traffic volume is less,” says Cafferty. There should therefore be some processing speed and free port advantages.

“You always can overflow to the larger registry if you have to,” he says. Other factors also could be significant. Perhaps protocol conversion is required. In that case, it is conceivable that conversion costs are higher at one of the two sites.

In any case, there are some subtle complexities to 10-digit number routing. “Exchanges don’t have ranges of NXX numbers, so all the routing is far more sophisticated and more real time,” he notes.

And Cafferty believes it ultimately will be necessary for service providers, enterprises and other users to “interface to all the registries.” Why would this be so? “ENUM without the ability to route has no value,” says Cafferty. “In TDM, you have to ‘peer’ your entire numbering plan,” he notes. “The minute you get into selective peering, traditional switching and routing methods no longer work.

“All the networks fall down when there is a mix of TDM and VoIP numbers to be looked up and routed,” he says. And all this before we even get to the matter of passing through an ENUM fabric all the applications and data required for those applications to work.

Oh, and money. Almost forgot about the money. IP