Ante la recurrente necesidad de discutir este tema, creí conveniente rescatar este artículo que escribí en el remoto año de 2002 para la ahora extinta revista Latincom sobre la necesidad de contar con NAPs / IXPs en Latinoamérica.
Muchas de las ideas que se exponen parecen ser desconocidas para infinidad de “expertos” sobre el tema.
Are Internet NAP’s relevant for Latin America?
When Bill Clinton decided it was time for the Internet to be opened up for commercial use, the US government was still funding the NSFNet, “The Backbone” that interconnected all participating networks in just one single Internet.
A new market was going to be born, where any operator could invest on Internet infrastructure. Still, a new technical mechanism was needed to replace the NSFNet’s integrating role so all networks could remain interconnected with each other, without relying on the infrastructure of a single monopolistic provider.
The US government favoured a model of telecommunications hubs. Operators made bids to build and operate one of the four NAP’s (Network Access Points) that were established on Washington DC, New York, California and Chicago making possible for all networks to remain interconnected.
By 1997, it was obvious that just four points were not enough to cope with the increasing complexity of interconnecting more and bigger networks. Complains about NAP congestion were a common place and new business perspectives questioned the “interconnect to everyone” proposition; in particular, some large companies (MCI the most notorious example) considered that treating significantly smaller networks as “peers” was equivalent to actually subsidising competitors.
In a market completely free of government intervention, some companies were created with the sole purpose of profiting by solving the Internet’s interconnection issues. Some of them were dedicated to provide simultaneous connection to multiple networks ( “bandwidth aggregation services”), others created “private” NAP’s (by 1997 were around 80 NAPs in the world) and some other created overlaid networks for fast distribution of content.
But more relevant, most carriers decided to interconnect their networks in mutually agreed private interconnection points that used their own facilities, avoiding reliance on third parties. These private agreements have severely diminished the real importance of the four original public NAP’s and of the aforementioned companies that had to survive from other revenue sources like access, collocation or managed services. .
Most articles dedicated to the history of the Internet, however, ignore these facts and stop at the deployment of the four original NAP’s. Maybe that’s the cause of the strange relevance that some media in Latin America give to every NAP-related announcement.
Back in 2001 the Secretary of Communications and Transport of Mexico published its Sector Plan for the 2001-2006 period with an emphasis on extending the reach and quality of government services by providing Internet access in rural areas. The e-México project, also considered “the construction of a NAP that will allow the exchange of data traffic among operators, making access to the contents of the e-Mexico System more efficient”.
This brief announcement was the basis for at least five publications to celebrate the decision as “the end of the absolute dependence on US facilities for the exchange of national traffic”. Some analyst preached the benefits in terms of “improved performance and reduced costs to end users” and condemned the cartel of the three major carriers in the country that benefited from the use of the infrastructure of their US partners by refusing to interconnect in Mexico.
None of those publications meant to question the members of the “cartel” on the subject. If they would have done so, their conclusions would had been quite different: no benefit outside “making access to the contents of the e-Mexico System more efficient” it would be attributable to the e-México NAP since the operators had agreed long time ago to exchange traffic within Mexico using private peering points.
A lot of press attention has also been devoted to the so called “NAP of the Americas” owned and operated by Terremark Worldwide Inc and promoted as “the fifth NAP in the world”. Beyond the names, a NAP is as important as the traffic that flows through it and the “NAP of the Americas” with an impressive list of customers, lacks any carrier from the second largest market in the region (Mexico). Terremark is also partner in the “NAP de Madrid” in which Telefónica de España does not participate in; and the “NAP do Brasil”, that was created for ISP’s to avoid “relying on Embratel” (the largest fixed operator in the country).
If there is a company that could symbolically claim that operates “The Fifth NAP” that could be MFS (now part of MCI), the winner of one of the original four NAP’s that after building the MAE West (Metropolitan Area Exchange) in Washington, by its own initiative decided to create a similar “twin” site on the East Cost (the MAE East). Certainly Terremark it is a company with a valid business model that is using a trademark that legally owns. Still, no single company is in a position to claim that its facilities are the ones on which Internet access for all Latin-Americans relies upon. Claiming that the “NAP of the Americas” is the “Internet gateway for Latin America” is, in the best case, just questionable.
As a final thought, it should be expected that the debate on Internet interconnection will increase on the following years on Latin America. Smaller operator still think that they should be given the right for “no-cost”, “bill and keep” peering with large competitors. And, besides to what the OECD thinks is fair, few operators are happy about paying 100% of the dedicated lines necessary to interconnect to backbones in the US, given the fact that US operators use the same lines to terminate outbound traffic for their own paying customers.
The increasing importance of the Internet in general, will also push local governments to reconsider their position from a hands-off approach to regulate for the benefit of the general public. However, in all cases, do not expect to see a lot of NAP’s as part of the proposed solutions.