Traffic Engineering

Kent W. England kwe at
Thu Sep 18 18:51:16 UTC 1997

At 01:23 PM 9/17/97 -0700, Michael Dillon wrote:
>The only area in which engineering can be used to solve your hot button
>issue is that if providers in your local area build a good local exchange
>point then you may be able to persuade more local end users to connect to
>local providers in order to attain better local connectivity.

TTraffic exchange is engineering tactical, not business strategic, for
national backbones. For regional providers, I think it may become strategic
and could be a flaw in the business plans of the backbone providers. If you
take web traffic and local traffic out of the national backbone business
plans, they have a lot less traffic than they expected. 

You build the local exchanges as the traffic builds. It is not economic for
a group of national backbones to build local exchanges when it is cheaper
to backhaul a couple megs of traffic a few hundred or thousand miles and
save on the number of exchange points. In the metro areas, it is still
often cheaper to backhaul (using CAP facilities) a few dozen miles than
build a POP.

Some RBOCs proposed NAPs in each LATA some years ago when the NAPs were put
out to bid by NSF. This was obviously premature. It is still premature,
although I bet there are more than a dozen LATAs where exchange is huge
(most already have NAPs or MAEs and private exchanges). The local exchanges
will be built -- it's just that we don't know whether they will be public
or private. A public exchange makes sense if there is a sensible sponsor
and lots of local providers with scale. If only a couple of national
backbone providers have the required scale, then the local exchange will be
a private interconnect.

Remember, the Internet started off as a pervasive but low density
technology. It was able to be pervasive only because it used the pervasive
telephony infrastructure. It was and is low density because it started from
zero. (Duh, you say. Right.)

The telephony infrastructure started off entirely local. When there was
high density local coverage, there was still no long distance. That came
only with the Bell System.

For anything that is two-way, pervasive (~100% coverage), and universal
(~100% density) -- what will the traffic patterns be? The Internet is only
the second two-way, universal, and pervasive technology ever. The first is
the telephone. If you extrapolate from the first to the second, then the
answer is that local traffic will dominate long-distance traffic. I think
that this will be true in any event because the tactical issue of paying
for bandwidth versus data storage will drive the Internet to local over
long-distance traffic patterns, because bandwidth is more expensive than
storage (for the foreseeable future).


More information about the NANOG mailing list