Wan virtual sensation

New technologies are breaking down the service walls that once divided carriers. That means the status quo, in which business users get Internet protocol services from an Internet service provider and fractional T-1 frame relay service from a local exchange carrier's central office, is rapidly changing.

As carriers roll out digital subscriber line (DSL) technology, virtual WAN multiplexing is emerging as the access network architecture that carriers can optimize.

Incumbent LECs own the dedicated DS-3 connections and copper loops used for the voice network and Internet access. But a growing number of competitive LECs are competing furiously with one another and with the incumbents for end users. More than 1100 new entrants in local markets had been approved nationally as of January, according to the National Association of Regulatory Utility Commissioners.

CLECs today claim 1.6 million access lines, or 2.6% of the $102.7 billion local telephone market, according to Merrill Lynch Global Telecom Research. This proportion is expected to rise to 5 million access lines, or 4.5% of the local market, by 2000. The United States Telephone Association cites 1700 interconnection agreements between incumbent telephone companies and competitors.

The typical interconnection agreement, which gives a CLEC access to a high-speed trunk to direct traffic to its off-site switching center, is the first step in an 18-month process. The CLEC must ink an agreement with the incumbent for a 10 x 10 ft. collocation cage before it can install equipment.

Moreover, under such an agreement, although a CLEC can terminate lines and multiplex traffic in its CO equipment, it cannot switch traffic between subscriber lines inside the cage. The CLEC must backhaul its traffic to a switching center where it can switch or groom the traffic to hand off to an ISP. Only after this preparation can the CLEC actively market its services to a CO's customers.

Carriers can use DSL to offer value-added networking services by optimizing transmission over existing copper pairs. DSL provides LAN-like availability and supports a range of speeds from 128 kb/s to 1.024 Mb/s. Products can reach 22,770 feet at 128 kb/s.

End user traffic is transmitted to a DSL access multiplexer (DSLAM) at the CO, then to a frame relay switch and out of the CO to the ISP (Figure 1). Because DSL breathes new life into the existing copper loop, it is fast becoming the strategic weapon of choice for both CLECs and LECs, enabling them to satisfy pent-up demand for high-speed data services, high-speed Internet access and corporate enterprise networks.Issues in network operationsBef ore they can provision DSL services to customers, CLECs and LECs must design a DSL access network architecture that can deliver LAN data from the end user to the ISP without involving the carrier in protocol addressing. They also must minimize WAN virtual circuits and minimize change for the end user and ISP.

Carriers can choose from several multiplexing schemes inside the DSLAM for the access network. The best way to assess these schemes is to compare the effort and expense required to perform various network operations. These operations include:

* Adding a new DSL subscriber. A proper network architecture lets each operator act independently without having to coordinate service technicians or delegate operational authority.

* Adding a new ISP. Unless carriers offer Internet services themselves, they must negotiate service agreements with ISPs. The effort required to add new ISPs should be minimal.

* Changing a DSL subscriber from one service provider to another. ISPs can experience churn rates of up to 10% a month. Even if the churn rate for DSL service is 1%, a carrier with 100,000 subscribers will change 1000 subscribers a month from one ISP to another. The process must be fast and simple.

* Providing end user security. If an end user's network is connected to the Internet, it must be secure. However, a service provider that forwards network basic input/output system traffic can compromise the end user's perception of network security. No carrier takes "network neighborhood" so literally that it allows users to see into their next-door neighbor's hard drive.

* Separating traffic. The carrier should be able to divide traffic into separate classes easily-premium vs. regular-and to provide fair access to the network to all users, regardless of traffic class.

* Managing DSL services. Some vendors offer management tools that let carriers add subscribers; configure, diagnose and monitor the DSLAM; and partition the network across multiple ISPs and corporate user groups. These tools are especially useful for data CLECs such as NorthPoint, which can use them to satisfy service-level agreements.

* Mixing and multiplexing. Carriers can choose from three DSL multiplexing schemes. The DSLAM can multiplex traffic at the circuit layer on asynchronous transfer mode or frame relay; it can multiplex traffic at the IP layer; or it can multiplex traffic into a virtual WAN at the Ethernet layer. The choice of multiplexing technology determines the switching equipment used by the CLEC or LEC, and it affects the form of the ISP traffic handoff.

Circuit multiplexing This is the technology of choice for carriers serving a small number of medium-sized to large corporate subscribers with sophisticated enterprise networks. Such subscribers employ permanent, high-speed fractional T-1, full T-1 and even a few T-3 connections, and they require large blocks of permanent IP address assignments. Although corporate customers are few, they generate the bulk of an ISP's traffic.

In a circuit-multiplexed network, the carrier provisions a separate permanent virtual circuit (PVC) between each subscriber and the service provider. To carry traffic through a DSLAM, each virtual connection is identified by a virtual connection number-virtual path identifier/virtual channel identifier for ATM and data link connection identifier for frame relay (Figure 2).

To add a new DSL subscriber, the carrier must provision the circuit, and the service provider must provision a circuit termination. To change a DSL subscriber from one service provider to another, the carrier must replace the original PVC and its termination with a new PVC to the new service provider.

Circuit multiplexing inherits the traffic security and traffic fairness of the underlying virtual circuit technology. Frame relay implementations use frame relay committed information rate. For ATM, circuit multiplexing uses whatever ATM traffic classes-generally unspecified bit-rate, variable bit-rate and constant bit-rate-are implemented in the DSLAM and in the underlying carrier switches.

Today, PVCs are the basis of circuit multiplexing. However, every carrier planning a circuit multiplexing DSL service is upgrading to a switched virtual circuit (SVC) ATM service to handle the flood of move-add-change orders they expect from millions of DSL subscribers. Many LECs have adopted a circuit multiplexing model because it simplifies their DSLAMs.

IP multiplexing In an IP multiplexed network, traffic to each service provider is multiplexed using IP addresses. Upstream traffic is sent to a service provider based on the IP source address of the packets, while downstream traffic is demultiplexed based on IP destination addresses. IP filters provide traffic security, and IP flow management provides traffic fairness.

A change order requires the LEC to work with the ISP before moving, adding or changing subscribers. Once the ISP brings circuit connectivity up to a DSLAM, it does not have to reconfigure its network when the carrier adds new customers (Figure 3). The carrier and the ISP need to coordinate IP address assignments, however, because the carrier must administer IP addresses in the range managed by the backbone ISP.

ISPs use a form of IP multiplexing for dial subscribers who typically connect to heavily overbooked dial-up modem ports using the Internet's point-to-point protocol (PPP). When the user dials in from a single PC, the ISP assigns an IP address to the subscriber's PC for the length of the session and provides all backbone network services such as domain name service, e-mail and routing.

The earliest DSL trials used IP routing and IP multiplexing because the earliest DSL modems supported only direct Ethernet connections. Address coordination between LECs and network service providers quickly proved to be too complex to manage. LECs found the IP multiplexing model broke down because many enterprise customers used private IP address space.

Virtual WAN multiplexing IP multiplexing, which is appropriate for networks in which the carrier and service provider are the same organization, has the advantage of a simple ISP handoff. Circuit multiplexing stays at Layer 2 in the DSLAM and provides high-speed, dedicated access. However, neither method is optimal for DSL. The virtual WAN, which combines the best of both worlds, enables transparent delivery of high-speed DSL traffic over the copper loop, and the carrier does not need to be aware of the protocol.

ISP traffic for a DSL network employs a hybrid of the temporary PPP handoff and permanent connection models. On one hand, DSL subscribers have broadband, always-on connections. On the other hand, many subscribers have one or a few machines, which may not always be on.

In a virtual WAN multiplexed network, traffic to each service provider is multiplexed at the Ethernet layer in the DSLAM (Figure 4). As in an IP multiplexed network, the carrier moves, adds and changes subscribers. The ISP can terminate a CO's population of DSL subscribers on a single, clear-channel T-1 or T-3 access circuit while retaining control of its own IP address assignments.

A virtual WAN multiplexer leverages standard IEEE virtual LAN technology, but it implements some special rules to ensure traffic isolation and security. For example, it distinguishes between upstream ports pointing to the ISP and downstream ports pointing to individual DSL subscribers. Broadcast and unknown traffic are only sent upstream from the source. Thus, network basic input/output will not leak from subscriber to subscriber, and the network learns new addresses quickly. ISPs also use tunneling and VPN technology to provide access security in a virtual WAN.

Virtual WAN multiplexing fits naturally with the Internet focus of data CLECs' customers. It also gives the CLECs a competitive advantage over the LECs because they can use it to simplify the traffic handoff to their ISP customers and improve operational efficiency.

For Internet-focused CLECs, virtual WAN is emerging as the DSL access network architecture of choice.