[asterisk-users] What are the various models of DID providers

Alex Balashov abalashov at evaristesys.com
Tue Jan 13 05:07:11 CST 2009


randulo wrote:

> Alex, thanks for the excellent explanations! Exactly what I was hoping
> for. 

Good!  Glad to help.

> Yes, in fact I should have said "origination and termination".

Ah.  Termination is a somewhat different game than origination.  From a 
technical point of view, most of what I said about origination also 
applies, and there's the same kind of arbitrage of underlying carrier 
pricing going on in the exact same way.

But all other things being equal, origination is an easier and better 
business model for an ITSP to be in.  That's why you'll notice quite a 
bit of origination providers who don't want to touch termination with a 
ten foot pole.

It has to do with costs, exposure and regulation.  In the US, at least, 
carriers *like* inbound traffic to be sent to them by other carriers. 
If you get a SIP trunk from Level3 and get some numbers from them, they 
take considerable delight in customers of other carriers calling you. 
Aside from using resources, it doesn't really cost them anything.  In 
fact, carriers actually pay each other to terminate calls amongst 
themselves;  this is known as "reciprocal compensation" or "intercarrier 
settlement."   When carrier A sends a call to carrier B, carrier B gets 
to bill carrier A a fairly infinitesimal but still nonzero rate.  The 
idea behind that is that the resources consumed do have a nontrivial 
cost associated with them, and unlike carrier A, carrier B has to accept 
the call and has no control over the volume.

Reciprocal compensation is a very complicated subject, and indeed, 
probably the single most broken aspect of telecom law in the US right 
now.  There are lots of things that are exempt from it with varying 
degrees of practicality and enforceability, such as VoIP traffic 
(defined as traffic that originates from VoIP customer premise 
equipment) and ISP-bound modem traffic[1].  Not everyone has to bill the 
same rate;  metropolitan incumbents mostly do, but rural carriers 
fixed-line carriers get to charge special - often much higher - rates in 
consideration of their low-ROI "rural" build-out characteristics[2]. 
It's an effective subsidy.

The point is that if you are doing origination as an ITSP, it's pretty 
standard to get unlimited inbound usage per channel (call) for a flat 
rate, which is something you can then in turn offer to your customers. 
It cuts out the exposure and makes for an attractive product, since 
people don't like paying usage (you don't pay it with a conventional 
landline).

If you're doing termination, though, you have to face this problem from 
the other side.  The carrier *will* bill you for usage, so at the very 
least, you have to make sure you aren't out money due to failures of 
billing reconciliation, rounding, etc. which keeps you awake at night 
worrying about the accuracy of your CDRs (Call Detail Records) relative 
to what the carrier sees.  Long-distance competition is a lot older than 
local loop competition (it was part of the Modification of Final 
Judgment that broke up AT&T in 1984;  AT&T's divested Regional Bell 
Operating Companies (RBOCs) would be allowed to retain their local loop 
monopoly in exchange for allowing competition in long-distance hauling), 
which helped domestic LD prices collapse over time along with the advent 
of CLECs and VoIP.  As a result, the margins are very, very thin; 
slight billing and rating errors mean you can be out a lot of money if 
you collect just a little less from your users than your carrier bills 
you.  Plus there's fraud to worry about.

It's actually a lot worse than that for the reasons I mentioned above; 
depending on which carrier the call is ultimately going to, *your* 
proximate termination carrier faces different access costs.  As a 
result, the country is split up into various "tiers" of traffic. 
Terminating traffic into RBOCs costs one thing, into mobile providers 
another, rural independents a third, and so on and so forth.  That means 
you have to be aware of the derivatives of these cost differences and 
bill your customers accordingly, and if you want to hear stories about 
how hard that can be, talk to some termination guys.

It is possible to get a "blended" rate based on an average derived from 
certain assumptions stipulated in the contract about the composition of 
your traffic destinations, but that has a number of problems.  You can 
be thrown off-guard by violations of the blend agreement if your carrier 
decides to pursue you for unexpected higher costs, and it doesn't get 
you the most competitive rate to offer to your customers for traffic 
destinations that *are* cheap to call.  Plus, with higher volumes, you 
get screwed because you can bill your customers one rate and make more 
money off the arbitrage to cheaper destinations while eating a loss on 
your gross margins on the more expensive ones.  So, higher traffic 
volumes tend to send an ITSP toward a "decked" or "tiered" arrangement 
with their carrier (by rate center, by LATA, by whatever the carrier 
defines as a tier, whatever).  And, of course, if you do wholesale 
you'll have some customers cherry-picking your expensive routes in a 
less than optimal proportion when that happens.

All of this can be dealt with, but requires some pretty serious billing 
innovation and demands a high level of accuracy.  If it's wrong or 
doesn't perfectly align with your carrier's billing process, you can be 
exposed.

So, termination is a much nastier business to be in, even if you do have 
blended termination rates.

-- Alex

[1]  There was a very fun game CLECs used to play in the late 1990s
      where they'd sell lots and lots of circuits cheaply to ISPs
      running modem pools, and collect a huge reciprocal compensation
      bill from the incumbents who were sending them an enormous amount
      of traffic (modem calls average considerably longer than voice
      calls) because most telephone service subscribers were customers
      of the incumbents.  The incumbents lobbied to get this stopped,
      resulting in a ruling circa 2000 that ISP modem traffic isn't
      subject to the same access charges as other types of locally
      exchanged traffic.

      Oddly enough, the justification in the opinion had something to
      do with the idea that the logical "endpoint" of an ISP modem
      call was "not local" since Internet destinations are not "local"
      in nature.  From a techie's point of view, this type of reasoning
      defies common sense.  But the incumbents tried hard.

      When access charges came about, the incumbents were big fans of
      the idea because they had most of the subscribers and expected to
      be net terminators (most of the time anyone's making a call, it's
      going to an incumbent number).  But the modem boom stood that idea
      on its head, at least as a large category of phone traffic, when
      ISPs bought cheap inbound circuits from CLECs.


[2]  Carrier access charges for rural operators created new and
      exciting arbitrage opportunities too.  If a rural carrier can
      bill massive access charges that a metro carrier can't, why not
      set up conference servers in those areas, enter into some sort
      of profit-sharing arrangement with the carrier for the access
      charges, and pump massive amounts of traffic in?  All hail free
      conference call services.

      There was a very infamous scandal about traffic pumping with a
      rural independent in Iowa in 2006, where tens of millions of
      dollars in (high) access charges were run up for AT&T and Qwest
      by a free conference service along these lines.

      These kinds of loopholes are on their way out now, and anyone
      trying to make a business model out of it is well advised to stay
      away.  CABS arbitrage is a well-known thing and various reforms
      pertaining to it are ongoing.

-- 
Alex Balashov
Evariste Systems
Web    : http://www.evaristesys.com/
Tel    : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
Mobile : (+1) (678) 237-1775



More information about the asterisk-users mailing list