First there is the tyranny of numbers. A Least Cost Routing application typically will use the Local Exchange Routing Guide (LERG) to determine all possible dialed destinations. The LERG defines roughly 450,000 destinations. For each destination you might have a cost proposal from several IEC. We tried evaluating 8 IEC.

So there are several million tests to run simply to derive a single list of the cheapest routes to each possible destination. But this is only the beginning.

Once the list of cheapest routes is compiled, there are several more steps:

• Collapse consecutive destinations served by the same IEC into a single route.
• Interleave “special” routes, such as 411, 611, 911, 976 area codes, specific toll-free destinations, etc. Each of these routes may be dedicated to a particular IEC or service provider. This is necessary when the rules of the switch require routes to be lexically sequential. Our Nortel MTX has such rules.
• Generate a routing file suitable for uploading to your switch from the list of cheap routes

Collapsing consecutive destinations is not difficult, nor is generating a translations file for your switch. Interleaving exceptional routes has proven particularly vexing. There are several conditions that might make this so:

• Individual LERG destinations can be entire NPA-NXX blocks of 10,000 MDN, or they might be NPA-NXX-Y blocks of 1,000 MDN. The code must determine this and adjust as needed.
• If the current exception is lexically less than the next destination then it must be interleaved between destinations.
• If the next destination is a block of 10,000 MDN, then it might be necessary to split it to interleave the current exception.
• There might be several exceptions back-to-back.

Of course the whole exercise could be attempted using Excel. To ably handle any list having 450,000 members Excel 2007 would be required, as it can manage worksheets having 1.2 million rows, and earlier versions maxed out at 65,000 or so rows. I know of one operator that has taken this approach and has found that it does not work well when it works at all. Apparently the high-end PC frequently crashes during runs.

We continue working on our Perl and SQL approach. Our ultimate goal is to have a solution that can be run frequently, perhaps as often as every month. But certainly whenever we get new lower pricing from the IEC.

What we saw initially was increasing traffic volume to destinations that seemed to be rather out-of-the-way, not a destination one would expect to be in our top ten. We use an application called CDR Collection Manager (CCM, which is highly recommend!) from Quantum Systems Integrators that collects all Call Detail Records from our Nortel switch and saves them to a UNIX server in a searchable format. Further Analysis of these CDR showed both high number of calls and extremely long call durations. Searching Google for the called number revealed it to be a free chat room.

We spent several days wondering about these businesses before we saw the light. It wasn’t at first clear to us how they would be making money. No for-fee services were evident on the web site, there were no obvious advertisements that might’ve supported their service. Then my Regulatory Director found this link at the Heartland Institute. The article calls these businesses “traffic pumpers.”

So the business model of the traffic pumpers now becomes clear. The companies offering these services partner with small Local Exchange Carriers (LEC). The LEC are chosen based upon them having an abnormally high access charge. Access charges are fees, laid down in law, which long distance service providers, aka “Inter-Exchange Carriers (IEC), must pay to LEC for the privilege of delivering voice traffic the “last mile”. The traffic pumpers try to stimulate the largest volume of incoming traffic possible to increase the amount of the access charges they collect from the IEC. Clever!

On our side, we have a flat-rate-all-you-can-eat business model. Success in this model depends to a large degree on having a low cost structure, and having subscriber usage which conforms to the “norm.” Extremely high volume callers can distort this model, and these traffic pumper destinations give such subscribers a reason to call.