Wednesday, November 11, 2009

PathFinder Business Model Flaw

"PathFinder is the GSMA’s multi-tiered, fully extensible number resolution system. It is designed to facilitate the efficient global interoperability of new and existing IP services using telephone numbers – all within a business framework consistent with operator needs. It provides a comprehensive suite of services designed to enable IP-to-IP interconnections between interconnect partners for voice, messaging, video, and other IP services."

is a certified PathFinder partner. The TransNexus OSPrey server can launch a ENUM query to the PathFinder service as an alternative to least cost routing. We believe the efficiencies from IP to IP routing made possible by PathFinder will generate new profits for carriers and better services for their customers.

As a PathFinder partner, I have been promoting the service to carrier customers. Some of the feedback I have received is very clear. Carriers do not like the Called Party Pays business model of PathFinder. Here is a simplified description of the PathFinder business model. Service providers provision their telephone numbers and corresponding Service Provider Number (SPN) to the PathFinder database. When a service provider originates a call they send a query with the called number to PathFinder. PathFinder returns the SPN of the service provider that serves the called number. This type of service has become critical since number porting between service providers is widespread. All of this is good. Now here is the problem - the carrier serving the called party pays PathFinder for the query. This simple flaw in the PathFinder business model is a serious obstacle to its success.

There is some sound logic to the PathFinder business model and here are some reasons why it makes sense:
1) The called network may benefit from the PathFinder service because it may receive the call directly over an IP connection which could be less expensive than a TDM interconnect.
2) Also a direct IP interconnection will enable end to end IP services so the called party can benefit from the advantages of VoIP such as wideband codecs and better sound quality.
3) The carriers who terminate the most calls will be the biggest carriers since they serve the most telephone numbers. Therefore, the biggest carriers with the most financial resources will make the largest revenue contribution for funding PathFinder. This is an egalitarian concept that regulators would endorse.

Now here are the counter arguments:
1) The called party pays model means that the terminating carrier is forced to pay an expense that it has incurred on behalf of the calling network. Everyone fears an expense liability that is out of their control.
2) The called party network has no record of PathFinder dips that point to its network and no way to audit its PathFinder bill. Conversely, the Calling party network has complete control over its number of Pathfinder dips and perfect audit capability of its PathFinder expenses.
3) Why should large carriers, who will be the net terminators for PathFinder queries, subsidize smaller carriers for PathFinder expenses?

And here is the Partner's appeal:
Put logic and intellectual arguments aside when defining the PathFinder business model. Go with what works in the market. PathFinder, like so many network initiatives, has a chicken and egg problem. The idea makes total sense, but is worthless until there is a big database that offers value. Getting the world's largest carriers to participate and provision their numbers to PathFinder is goal number one. Any other focus is folly. If the big carriers want a calling party pays business model - the argument is over. PathFinder needs to implement a calling party pays business model. TransNexus is a big supporter of the PathFinder concept and looking forward to its success.

If you think I have this wrong, please post your thoughts here. This is all about getting it right, you won't hurt my feelings. This post is all about getting the economic incentives right so PathFinder and the promise of IP to IP communications become a reality for everyone.


Anonymous said...

Jim, I totally agree! If they leave it as a Called Party pays model then I can see a lot of people trying to become an agent (if they have such a program) to bring business to Pathfinder. Those people will probably also have some involvement in a IXC, or LEC of some type that thet will ten originate millions of calls to Pathfinder from. Hey, why not? They're not paying for it - the big carries are! Then the agent get a commission from Pathfinder for running up a big bill that the big carrier has to pay.

I have seen this play out in several different ways in the voice business. Recip Comp and the Free Conference Call services are a perfect example. Pick an NPA in an expensive LATA and give away a conference calling service after you have your recip comp agreements in place and watch the calls come in! Most end users have flat-rate all-you-can-eat plans at home/work today, but the LEC in LATA XYZ still gets paid per minute from the carrier handing off the call - usually the RBOC. It is no wonder the RBOC's want to kill Recip Comp.

The bottom line is that spending someone else's money is easy, so if the model works that way it won't work. True multi-lateral (free exchange) works best in the IP world as proven by the ISP's - otherwise we would have had a pay-per-packet model for the Internet which would have killed its adoption.

The issue is that the "numbers" and registration of them have been separated from the databases that can facilitate the interconnection of those IP endpoints - and both service providers want to get paid. One for the "number" and the other for the "lookup".

Aside from Calling Party pays, the end resolution of end point resolution will be to have true multi-lateral VoIP peering which is just like a DNS lookup. You don't pay every time you type in a URL and go to a website, so you should not have to pay to look up an IP endpoint in the VoIP world either. Anything else is and will be fragmented, disparate and create loopholes, like the current Pathfinder model, that will be found and exploited at someone's expense.

Hunter Newby

Anonymous said...

Some great thoughts here, thanks for taking the time to discuss Pathfinder. While the called/calling party paradigm is one interesting way of looking at this, I believe Jim is right in that it debates a business issue prematurely. It’s a little like debating if the northbound or southbound traffic on a proposed road should be charged the toll before there’s a good understanding of where the road will be built and where its on-ramps and exits exist.

My observation to the issue is the tail trying to wag the dog. Traffic transport and interconnect economics are the dog and publishing registries and query economics is the tail . The latter cannot alter the negotiation of the former. However, transport and interconnect agreement must occur before any talk of registry and query becomes relevant and will, in most cases, dictate its economic outcomes as well.

Perhaps the debate of whether calling or called party pays is nothing more than a manifestation of the disparity in philosophy of the peering concept itself. I have seen 3 models under debate, each very different in their business intent and likelihood of reaching acceptance:

1. Bi-Lateral Contracts: Two willing parties agree to exchange traffic between each other and similar to the ancient practice of PSTN interconnect, they mutually negotiate a contract to trade traffic. In some cases, it may be goodwill where neither party charges the other. In some cases, there are rules and guidelines to balance traffic loads (and other items that attempts to create “equality” in sharing the cost and load). Sometimes, they are very profit driven, particularly when one carrier is far more dependent on the other, thus north may have to cover a heavier share of the burden than south and thus costs are aligned to power of negotiation. Some of these agreements can be agreed to in a few minutes and others are quite complex, but in all cases terms and an agreement is set prior to peering taking place. The terms can be all over the place, but the point here is that there is no discussion of routing registry and query techniques until after the structure of the bi-lateral agreement is in place. In this world, routing registries and query methods follow the same process in their own discreet negotiations and are not bundled into actual peering arrangement.

2. Public Exchange: A far more utopian quest of peering is the concept that carriers (companies in the business of profiting from the investment of building a network) would open their network for public consumption without any prior bi-lateral terms or contracts being established. This would be the same as a hotel opening its rooms to all interested parties and enabling them to stay there for as long as they want without any validation or prior terms for stay. In this world, a carrier publishes a public entry point or gateway and all takers are welcome and no prior traffic terms need apply. I have a hard time understanding why this concept continues to get such heavy discussion in the routing registry and query portion of the discussion when it would be considered almost inconceivable in the transport and exchange business itself. I think it stems from so many people really not understanding the true economics of publishing data in a registry and doing queries against it.

3. Multi-Lateral Agreements: In this case, “a higher power” enforces a practice for exchange and all participants must agree to those terms prior to participation. This scheme works very well when imposed by government regulation or small communities or federations with like needs come together and agree to common terms. Like a bi-lateral agreement however, all participants must agree to these terms ahead of participating or they cannot gain access. (cont)

Anonymous said...

(cont) So when we talk about routing registry and query, the economics unfold as a result of the 3 scenarios above. Here is my opinion (and would love to hear yours) of the likely outcome:

Bi-Lateral Agreements are where virtually all of the real peering marketplace exists today and will be the bulk of the next 5 or so , whether it’s just within the US or on a worldwide basis. Under bi-lateral peering agreements, both parties agree to the types of traffic and the “footprint” of what type of traffic each party accepts from the other. For the most part, the parties really don’t care what registry the other party queries to decide their own routing choices, they just need to make sure that both parties have the means to connect traffic with one another and that what is being exchanged lives up to the rules. Evidence of this is in the existing use of the PSTN, where each network uses many different databases, services or routing schemes to decide where to send the traffic that they originate. In the peering world, this does not change and so far it hasn’t. Companies conducting peering arrangements today, use many different self-defined means to conduct their own queries and routing schemes, often built within their LCR practices and such.

Those in the business of helping carriers with registry, routing and query negotiate their own bi-lateral agreements with each of the players on how that carrier gains access and uses to the database. So two peering parties interconnected to each other’s networks may in fact use very different query means, each with their own unique practice and agreement. In this world, it will always be calling party pays for registry/query because it is in fact the bi-lateral agreement between the query originator and the query provider. Unless a major carrier comes forward and consciously desires to provide their peering service where they provide “free query” to the originators who they have prior bi-lateral agreements with, then calling party will always wind up paying. In the peering registry world, there are early signs of this possibility, where some carriers have very lucrative low cost means to publish their footprint to their bi-lateral agreements and provide an additional incentive/term for striking peering agreements. One possible outcome of this is that carriers will “publish” their footprints through different peering platforms as a means to advertise and sweeten their termination services and in some cases they will pay for the call attempts and in others they will expect callers to pay for their own attempts.

The opposing technique of making the carrier of the called party pay (against his will) is counter intuitive to the flow of commerce and most likely could only evolve in the utopian public exchange where perhaps eventually the cost and risks of interconnect become so irrelevant that no one is really being imposed of anything meaningful. This could potentially happen where the public exchange becomes the final stage after the emergence of a very large multi-lateral agreement scenario. If a government were to step forward and alter the economics of the business and offer incentives or penalties to alter natural commerce, we could then see where carriers adhere to the imposed scheme. This is basically the same argument that has stalled public ENUM for quite some time.