Monday, October 21, 2013

When "Carrier Class" is a Bad Idea

By definition, an Internet service provider cannot offer service to users who cannot afford to pay very much unless that ISP's costs also are low, even if that ISP might have other revenue streams than direct subscription or usage fees. 

Over the past decade, that has lead many smaller ISPs, serving hard to reach populations, to build and operate networks that are not "carrier class," on purpose. The reason is simple: carrier class networks would cost enough that end user costs could not be kept at reasonable levels.

That will be an even-tougher challenge in many parts of the world where monthly cash income is low enough that people can not afford to pay much more than cents a day. 

That is one problem Google's Project Loon is trying to solve, using some principles similar to low earth orbit satellite fleets. A LEO satellite fleet moves constantly through the sky, requiring tracking satellite dishes. 

As you might guess, that is too expensive for low-cost ISP service in many  to reach, low income areas around the world. So Project Loon is testing fleets of wind-blown balloons.

This is an early earth station, or antenna. Some of you won't be surprised if inside is found a Ubiquiti radio, which has become a mainstay of the wireless ISP business. 

Sometimes, people talk about "carrier class" as a good thing, and it is, in many communications contexts. Just as certainly, "carrier class" is not a good thing when a key objective is to provide service at historically low costs. 

When that is the objective, "carrier class" networks are in fact precisely what is not wanted. 

Despite the service issues that represents, "carrier class" also imposes cost. So one design foundation, for ISPs truly wanting to serve "underserved populations," is not to build carrier class networks, with all the reliability issues that will come with that approach.  

No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...