Jim Southworth Looks At Telco & Isp Business Impact - G.Lite Standard Helps Bring Down Per Port Cost, CLECs Push For Massive Deployment As Alternative To Data Over Cable

pp. 1 - 12

We interview Jim Southworth, Chief Technologist of Concentric Networks on DSL technology, deployment and business impacts; offering in 14,000 words a mini -encyclopedia on the subject according to Jim.

DSL(Digital Subscriber Line) is a radio frequency carrier that is multiplexed onto the copper local loop by a Digital Subscriber Line Access Multiplexer or DSLAM, a device placed at the telco central office. This device generally increases the amount of data the copper can carry by a factor of between 10 and 100 depending on the variety of technology used. The current drawbacks are that DSL will extend over a distance range of only about 3.5 to 4 miles from the central office under the best of conditions. Under current conditions, when DSL is fully deployed it, will likely be able to reach 80% of American homes. It is hoped that range will hit 5.5 to 6 miles within another year. This will extend its reach to more than 90% of homes.

Deployment is being pushed most heavily by national CLECs Northpoint and Covad. ILECs have been dragging their feet on deployments, especially in areas where it could replace T-1 business local loops. The revenue stakes are huge for we are talking about using a specially tailored flavor of DSL called HDSL2 to replace leased lines that the ILECs get perhaps $300 a month for with a service that would bring them about $12 a month in income.

Many flavors of DSL exist. They fall into two general categories. Symmetric and asymmetric. Most flavors are asymmetric featuring higher incoming bandwidth than out going. RADSL or Rate Adaptive DSL adapts to the distance from the Central Office, and to the other characteristics of the line with which it is presented. IDSL has been adopted from IDSN. Northern Telecom offers a flavor called Etherloop and a Paradyne one called MVL that on an instantaneous basis varies the amount of bandwidth needed to go in each direction based on a reservation protocol that says: hey, I need to do this to get from here to there. What can you give me? Finally VDSL will attain very high speeds on very short local loops.

Pricing on DSLAMs is usually specified in terms of so many dollars per port where port is defined cost of the customer modem and the cost of the DSLAM prorated over the number of customers it would serve. For MVL and some other implementations involving as many as 100,000 ports covering multiple central offices but no more than one good sized city, a per port cost of as low as $100 has been attained. A more average cost would be about $400 per port.

In the greater New York City region there are about 18 different ILECS and CLEC serving customers from many different central offices. In the absence of generally agreed on standards, this creates a situation where different DSLAMs from manufacturers such as Nokia, Alcatel, Copper Mountain, and Cisco may be in use in the same CO. Alcatel has been the manufacturer of choice for DSLAMs and customer modems for the ILECs.

In early July of 1999 the G.lite standard gained final acceptance. G.lite is intended for installation in the CO and customers home without forcing a truck to drive from the CO to the location of the end user, testing the circuit ever step of the way.

While SBC has been most active in the deployment of DSL, ILECs have been loath to install it on a reasonable cost basis because it will cannibalize their T-1 business. This situation has led to CLECs pushing to get their own DSLAMs into ILEC offices. Even though their CLEC status gives them the right to place DSLAMs into central offices, they have had to fight every step away for the space to do so. The ILECS succeed in imposing many impediments in the way of DSL rollouts. For example, because the ILEC technology in Boston is limited to a range of no more than 18,000 feet, you will find that you cannot buy a local loop longer than 18,000 feet in that area. Given all this ILEC foot dragging the saying goes that the most important person in a CLEC is its chief legal officer. But with CLECs beginning to be able to offer 2 megabit per second symmetrical HDSL2 lines, even the ILECs will find them selves forced to start offering HDSL to compete with the CLECs in its high density areas. This same market will start doing voice over HDSL at the same time. Cheap HDSL business local loops will enable all businesses to develop new applications including virtual collaborations real time and video monitoring

Southworth points out that bandwidth intensive DSL will increase ISP's upstream communications costs and is likely to lead to important changes in the business model of the industry. He believes that the costs of broadband infrastructure will be so high that ISPs with a subscriber base of less than 100,000 will find it very difficult to acquire and pay for. However he adds that we are not going to suddenly see 4,000 small ISPs die off. DSL will create a synergistic opportunity that will enable the smaller ISPs to become resellers of the larger ISPs bandwitdh into the communities they serve. It is important that this can happen because smaller ISPs have the infrastructure, customers and trust necessary to provide good customer service to their communities in a way that large ISPs cannot accomplish. Opportunities for small ISPs to partner with larger ones in all manner of content and service distribution should increase. The interview concludes with a discussion between Southworth and a COOK Report subscriber of the security issues involved in having an ìalways onî network connection.


Focuses On International Services And Internet Connection For New European Competitive Players

pp. 13 - 20

We interview Steve Heap, Teleglobe's VP of Global Network Planning and Engineering and Bob Collet, VP of Data Services. Originally Canada's international carrier, Teleglobe runs a large voice network, and ATM network that is very good at carrying video and a very large TCP/IP network. Its customers are carriers, ISPs, content providers, and business customers needing global connectivity. BT, ATT and MCI-Worldcom are competing in local markets with the very same start up telecom provider to whom they would like to offer international connectivity. Whereas these global competitors of Teleglobe have grown by acquisitions and face the problems of integrating the resulting networks, Teleglobe's strategy is to expand by growing their unified global system to one with 160 access points around the world. Not having to integrate different architecture allows them to grow at less cost and deliver more consistent service than their competitors Since they are not offering local service they aren't seen as competitors by their smaller customers and by the PTTs of the countries in which they operate.

The new global backbone powered by Alcatel equipment (40 wave lengths of 10 gigabits each) both under sea and on land that they are deploying will give them end to end control of their infrastructure into and out of 160 cities around the globe. This will facilitate their delivery of broadband QoS. While Teleglobe will offer IP directly over fiber, it will also offer IP over SONET and over ATM. It is watching the development of optical switching very carefully. Currently Teleglobe is in 27 cities in Europe, Canada, the US, Latin America and Asia.

Teleglobe considers itself to be the world's largest provider of satellite based Internet backbone services having no less than 66 different countries connected in that manner. As Canada's Intelsat provider it has direct access to Intelsat at cost. Many of its satellites links are used to update web caches.

Because of the high cost of bandwidth, the Australian IXC Telstra has a 90 megabit per second simplex satellite link to Teleglobe. Web proxy caches can come from that simplex link. Although Teleglobe may have customers who are using it to run a VPN, it does not emphasize such services. Teleglobe runs IP into more than 100 countries and has an IP backbone in the US and Canada with nodes connected at OC-12 speed. In addition the backbone has access nodes all connected with multiple OC-3s.

In the 100 countries in which it operates, it connects just over 200 ISP's and if you add up the routes of their customers, that equals to about 15% of the entire Internet routing table. Teleglobe's customers, also include large PTT's today, such as Telecom Italia and Telephonica, Singtel and Rosstelecom and a multitude of others. Teleglobe is a customer of Akami which has developed an intelligent distributed web mirroring service. Both Akamai and Sandpiper provide an ability to cache or mirror those elements that would traditionally be uncacheable.

Teleglobe has put most of its structure under the ocean. It has leased fiber in Europe and has also built its Internet backbone on a pay-as-you-go basis. I think it turned out to be exactly the right thing to do, because Teleglobe has been able to be profitable every step of the way. About half the cost of its five-year Globe Service expansion will be financed from Telglobe's cash flow


pp. 21 - 22, 24

We present a debate from NANOG in mid June on the comparative efficiencies of web caching versus multi-casting.

Short ICANN Over View

pp. 12, 20

Tony Rutkowski describes the meta issues in the ICANN debate and Jim Dixon, identifying ICANN's plans to control the Internet by means of control of a single authoritative root issues a blistering declaration of independence.