AT&T has proposed to the FCC that there be a handful of major switching hubs created in the country. Every carrier would then send their voice traffic to these hubs to be sorted and handed back to the various terminating carriers. Their argument is that the whole idea behind the IP transition is that the network be made as efficient as possible; they are promoting this idea as the simplest network to implement.
But is it the most efficient? Over the years I’ve done a lot of traffic engineering, meaning that I’ve helped companies analyze where their voice traffic comes from and goes to. What I’ve seen is that approximately 80% of voice traffic for most companies stays in a relatively small circle of perhaps 60 miles. This distance can vary a bit by company depending on how far away they might be from a major metropolitan area, but this basic traffic rule seems to apply pretty much everywhere I’ve looked.
So let me first look at the practical application of what AT&T is proposing. One would have to assume that if there was only a handful of nationwide interconnection points that they would be put in the same places as the major internet hubs – Atlanta, Chicago, Washington DC, New York City, Dallas, etc. What this idea means is that states not near to those hubs—say Montana, Nebraska, Maine, etc. would have to ship all the voice traffic from their state to the nearest big hub and back again.
While it might be more efficient to have only a few hubs, it certainly would not be efficient from a transport perspective. Somebody has to pay for all of that transport to and from the main hubs, and that is the real purpose behind the AT&T proposal. Under their proposal carriers other than them would pay to have all traffic brought to these main hubs. That is a major change from the way the industry works today.
Today there are two different sets of transport arrangements—one for regulated telcos and one for competitive CLECs and other kinds of carriers. Regulated companies today provide joint trunking between each other. For instance, if there is a fiber route between AT&T and another telco, AT&T generally owns the part that is within their territory and the other telco owns the part in their own territory. Sometimes this is negotiated differently, but under this arrangement both sides bear some of the cost of carrying voice traffic.
CLECs and other competitive carriers have a different situation. CLECs are allowed by the Telecommunications Act of 1996 to establish a point of interface (POI) at any technically feasible spot within a LATA (or region). Once that point is established, the CLEC is responsible for all costs on their side of the POI and AT&T (or Verizon or CenturyLink) is responsible for the costs on the other side of the POI.
AT&T’s suggested new network gets rid of both of these arrangements and instead moves all of the points of interconnection to the small handful of locations, and in doing so shifts 100% of the cost of the transport onto other carriers.
In a further money grab, AT&T would (as the assumed owner of these large hubs) charge a fee to other carriers for handing traffic from carrier to another. These fees exist today and are called transit fees. But today transit fees are charged on a relatively small percentage of voice calls since there are no fees charged for all of the jointly-owned arrangements I described above. Instead, under this new arrangement there would be a transit fee charged for every call that is handed from one carrier to another.
AT&T’s proposal is ridiculous for several reasons. First, the transit fees are not cheap and they cost more today than the traditional access charges that the FCC has been trying to eliminate. So AT&T’s proposal would increase the cost of making voice calls. The proposal is also a blatant attempt to shove all of the cost of carrying voice traffic to somebody other than AT&T. And finally, it forces companies to carry calls a much greater distance than they go today. This likely will lower call quality and increases the danger of the voice network going down due to a fiber cut or other network problem.
There is a much simpler alternative to what AT&T is suggesting, which is to let carriers negotiate how and where they hand off traffic. There are already huge numbers of local and regional interconnection points in existence, most established to hand-off local traffic in the local market. Carriers should be free to continue to use arrangements that work and with which they are happy. Think of these local arrangements as voice peering arrangements and they quickly make technical sense. Nobody is going to be unhappy if local connections transition to IP instead of TDM. But making the IP transition doesn’t mean that the whole nationwide network has to be massively reorganized. That is far more inefficient and costly than letting carriers find their own solutions.