A Practical Navigator for the Internet Economy


Via LMDS and IP Over Glass Integrates Voice and IP Communication for Small and Mis Sized Business


We interview John Curran, Vice President for Internet Technology of NEXTLINK Communications, a facilities based CLEC. Curran points out that NEXTLINK does not resell anyone's phone service. It installs its own switches. In installs its own local fiber. It directly interconnects with inter exchange carriers and with the incumbent LECs. Last July for more than $700 million it bought not only 25% of the fiber in Level 3's current network but 25% of its future capacity as well.

NEXTLINK is position to offer a broadband telecommunications solution to small and medium sized businesses which have previously been under served. It is building its own local fiber networks in each of the 40 markets it now serves. NEXTLINK, which owns more LMDS spectrum than any one else, will use LMDS wireless coverage to get point-to-point links of up to OC3 into buildings which have too few customers to justify direct fiber connection to NEXTLINKs network.

The roof of a tall building with a NEXTLINK fiber POP inside may serve as a line-of-sight wireless base station to cover an area of more than two mile radius with point to point high bandwidth signals or point to point multiple T- 1s. Using multiple tall buildings will allow it to cover an entire city with wireless broadband connectivity for buildings too small to justify a direct fiber connection. It expects to have at least two of its markets lit with LMDS before the end of this year.

At the present time NEXTLINK is reselling PSI's Internet service. Since Curran will use NEXTLINK's ownership of Level3 fiber to build out its own nationwide ICP/IP backbone in the markets it serves, it is very unlikely to resell PSI capacity indefinitely. given the broadband nature of NEXTLINK services, John feels that when he opens his own backbone, he should have enough traffic to be able to nail down optimal peering arrangements.

Since NEXTLINK does not resell ILEC services, what it does need is effective interconnection to their networks. It has done this in all the markets that it has opened up and is co-located in more than 150 central offices of local exchange companies. Doug Carter the NEXTLINK CTO, brought John in earlier this year to basically look at what it takes to make all this happen - particularly with respect to data. John is building up the team which will execute the strategy that he outlines in his COOK Report interview.

When asked for his perception of the marketplace over the next year John said: "Increasingly the telecom manager and the IP manager in these companies are the same person. Particularly if you are a company in the small and medium size market you will very likely have only a single telecom vendor. If you are an ISP you have the challenge to think about how you will become a stand alone, integrated communications service provider. In the future there may not be a very robust stand alone ISP category."


A short excerpt from a NANOG discussion of Internet security issues as compiled in a 100 page report by the President's National Security Telecommunications Advisory Committee. The discussion concludes with some remarks on protecting the network from the propagation of bogus BGP routes.


ROSS CALLON, IRONBRIDGE Chief Architect, Looks at QoS, Standards and Engineering Issues of Next Generation Routers

We interview Ross Callon, Chief Architect for IronBridge Network's multi terabit capacity router which is to be released next year. While RED and Weighted Fair Queueing are effective at smoothing out router traffic in real time, traffic provisioning refers to the long term need of ensuring adequate available infrastructure for network traffic needs. Traffic engineering falls in between t he two and is designed to give net work engineers the capability of making decisions that will affect their network's performance two or three minutes to two or three hours into the future.

The discussion notes the traffic engineering BOF scheduled for the OSLO IETF and describes the concerns faced by developers when dealing with the standards process. Does one lay out one's proprietary solutions or does one with hold them? Callon, who is building the traffic engineering capability for the IronBridge router, describes his own position as being middle of the road.

The purpose of traffic engineering in general or a traffic management system in particular is to enable a network to use its infrastructure with maximal efficiency. While routers have traffic engineering capabilities built into them, the difficulty of gathering the necessary data on network state and expected behavior becomes much more important with the increased numbers and speeds of the data links.

The complexities involved in a traffic engineering system at multi terabit levels are a bit awesome. The system will collect on the order of 100, 000, 000 data points as often as multiple times per hour and apply its algorithmic intelligence to the data. Over time the traffic engineering algorithm may compute a new set of paths, based on changes in the traffic matrix or changes in the network state. Put less technically: it will generate output that manages network traffic with maximal efficiency.

Callon says: "Traffic engineering is a non-trivial problem which can surprise people when studied in detail. At first glance it looks like you have to use connections, because you need to use explicit routing. Then you discover there are algorithms where you can do a provably optimal job with datagrams. Then you might ask, will I need to set up too many label- switched paths, and am I going to get blown away with state information? Or is the rate of dynamic change going to mess me up if I try to use pre- setup connection oriented paths? It looks like this shouldn't be a problem either."

"The point of view that either you need to do datagrams for traffic engineering or you need to do MPLS for traffic engineering doesn't seem to hold water either way. It seems you can do a really good job with either approach (provided that the number of Label Switched Paths can be kept under control, implying that the number of core routers is small enough)." Callon believes that once routers are pushing data at terabit speeds, in order for the network to function effectively and efficiently, a sophisticated traffic engineering system is necessary. Disclosure: the Editor is doing some consulting for IronBridge.

NETHEADS VS BELLHEADS: Best policy Paper Since Isenberg's Stupid Network

Run don't walk to http://www.tmdenton.com/netheads3.htm This is a 45 page policy paper: "Netheads versus Bellheads Research into Emerging Policy Issues in the Development and Deployment of Internet Protocols" authored by Timothy Denton, with Francois Menard and David Isenberg. It is the greatest threat yet to the 43 million tons of buried local loop copper.

This paper is filled with information that is known to most readers of the COOK Report. What makes it brilliant and groundbreaking is that it is written with consummate clarity and in a way that it defines all its terminology within the text in a such a way that the reader never feels burdened. It articulates two opposing design philosophies of building next generation networks: the Internet design philosophy and the PSTN design philosophy.


ISOC ICANN Connection Foretold in Landweber Plan for Control of DNS - NSI Will Not Play ICANN'S GAME

We offer some highlights from the latest moves in the NTIA -Internet Society led effort to deliver the Internet into the hands of an entity without legitimacy or accountability. ICANN Watch unites Post Farber and Froomkin in a Web site which warns that "the shadowy outlines of a new kind of constitutional structure for cyberspace, centered around ICANN, have begun to emerge. The consequences of these developments for the Internet's future could not be more profound." Essay by Froomkin points out that Magaziner premise was for Clinton administration to create an entity that would impose uniform international codes of conduct on Cyberspace - regulation by deception and stealth. We publish ISOC's September 1995 plan for the takeover, control and taxation of DNS. Landweber's document makes it clear that ICANN is a front for the same ISOC objectives. Plan's to move the A server as a part of a vague CREDA with NTIA are discussed. ICANN registrars Registrar.com and AOL have paid some of the same attorneys involved in IAHC to write a WIPO friendly contract that gives ICANN power to revoke any domain name essentially "at will". Finally Joe Sims admits that the interim ICANN Board may ignore with impunity any policy presented to it by any supporting organization.