Inside the carrier crowd

In response to rapidly changing market conditions, carriers are purging their old infrastructures and building next-generation networks that will be device-agnostic, content-aware and capable of supporting ubiquitous access.

During the past decade, carrier networks have been impacted by mergers and acquisitions, by the emergence of packetized data technologies such as frame relay and, of course, by the Internet, which created an ever-increasing demand for IP-based, high-bandwidth access.

As a result, carriers today have several networks they use to deliver services. These networks are typically connected over a high-speed backbone consisting of Optical Carrier links, for example, OC-48, which runs at 2.488Gbps.

New demands, however, are forcing carriers to make major changes to their networks:

    Application service providers (ASPs) are demanding that carriers provide better service-level agreements (SLA) and quality of service (QoS) for VPNs;Residential customers, once satisfied with dial-up, are clamoring for broadband Internet access;Corporate customers want high bandwidth, and they want carriers to install and manage WAN interfaces at the corporate sites. Plus, corporations want to add new interfaces to their access requirements, such as Web phones and PDAs.Cable companies are challenging carriers and competing for residential and corporate broadband Internet business.Finally, data traffic exceeded voice traffic on carrier nets about a year ago and continues to grow at a much faster rate, which means carriers must change their networks to accommodate this continuous increase in IP-based traffic, especially media-rich content such as video.

In response, carriers are scrambling to build next-generation networks that will be content-aware and device-agnostic. That is, when a person plugs in a device, the network should be able to adjust the bandwidth requirements accordingly. To do this, the network should throttle between constant-bit-rate (CBR) applications such as voice and variable-bit-rate applications (VBR) such as data. It’s expected that most regional carriers will have this capability by the end of 2001.

Also, the network should permit access to the same content over various transports, such as broadband (DSL or cable), wireless and traditional wired lines. If a customer decides to use a Web phone in place of a PC to access Web pages or send messages, the customer should not have to use a separate account, carrier or application.

Content-based carrier networks

These content-based networks will have several layers, starting with a transport layer consisting of OC-48 through OC-768 optical links, as well as fixed wireless. A layer of routers will sit on top of the transport layer. Switches and caching servers will be tuned to recognize and handle changes in content format and access devices.

While some carriers are providing some content-based routing, the majority of features will not be seen until late next year.

QoS: The No. 1 job

Behind the scenes, caching servers, routers and switches will be delivering QoS by logically segmenting bandwidth to meet the requirements of the content format. The No. 1 priority for most carriers is to add QoS to the network backbone. The two protocols most recognized for this service are Multi-protocol Label Switching (MPLS) and Differentiated Services (Diff-Serv).

MPLS involves creating a specific path for a given sequence of packets, identified by a label put in each packet, thereby reducing the time needed for a router to look up the address to the next node.

Diff-Serv is a protocol for specifying and controlling network traffic by application or class so that certain types of traffic get precedence. For example, when using this protocol, voice traffic, which requires constant bandwidth or CBR, will get precedence over data traffic.

In the backbone, most carriers are looking to use MPLS to create efficiency. However, depending on what the customer’s interface requirements are, carriers may connect to customers by either MPLS or Diff-Serv, depending how the traffic profile is established (time of day, application, end user and so on).

Diff-Serv can be beneficial if a customer is using multiple devices, so if a service profile is based on the user, the network will adjust as the user changes devices or access points.

On golden PONs

Another technology being looked at to increase overall network speed, but especially to boost speed close to the corporate enterprise and residential areas, is Passive Optical Network (PON).

This is a system that brings optical fibre cabling and signals all or most of the way to the end user. Depending on where the PON terminates, the system can be described as fibre-to-the-curb or fibre-to-the-home.

The system consists of an Optical Line Termination (OLT) at the central office and a number of Optical Network Units near the customer. The unique capability of PONs is that the optical transmission has no power requirements or active electronic parts once the signal is going through the network.

As a result, it is durable and can be used as a backbone transport with other media, such as cable and T-carrier system. This helps to resolve the age-old problem of getting higher speeds closer to customers while keeping delivery and support costs under control. Last year, Bell South began installing a PON for about 400 residential customers.

IP over glass

Another innovation being studied as a pure backbone technology to provide a high-speed, end-to-end delivery of IP services is “IP on Glass.” This is a solution developed by CIENA, Enron and Cisco to eliminate the overhead in translating between Layer 1 through Layer 3.

In effect, this involves the use of a gigabit router equipped with software that is configured to determine which path the IP bit takes over the fibre network. The fibre network is using dense wave division multiplexing systems that create logical high-speed optical circuits.

This architecture is being targeted at replacing ATM, SONET and frame relay for carrier backbones.

SODA quenches thirst for content

Content networks, which most customers will see as the interface at their premises, represents the last layer of the next-generation network. A content network comprises servers, routers and switches, which are configured to provide the desired access to an application. To date most companies are doing this by combining off-the-shelf caching and load-balancing products.

Going forward, companies such as Cisco will be introducing full suites of products that work together to provide application availability and performance. Cisco incorporates Self-Organizing Distributed Architecture (SODA) intelligence into its strategy. This is a patented technology at the Massachusetts Institute of Technology that allows a large number of servers to configure themselves automatically into a sophisticated network for efficiently distributing media among themselves and routing requests to the nearest server for fastest delivery.

New service offerings

As the next-generation carrier network takes shape, carriers will be offering new services to enterprise customers:

    By 2002, many carriers will begin offering incentive plans to get companies to convert voice and video applications to IP backbones. Carriers would like to start dumping legacy systems, and to do this they will need to have two options: one that allows customers to connect their equipment to a carrier-provided MPLS gateway; and one that allows customer legacy equipment to connect to a carrier-supplied IP gateway.Data services will be increased to include security, directory and QoS. To support the growing number of VPNs, carriers will focus on providing a form of managed certificate-based authentication services such as public-key infrastructure (PKI). This will be used in conjunction with user ID and passwords for strong authentication. In large-scale VPNs, this would be viewed as a replacement authentication tool.A metadirectory will be implemented by carriers to support security services and improve the handling of configuration data between corporations’ enterprise systems and carriers’ systems. The metadirectory will also be charged with storing authentication certificates, user IDs and passwords for end users. This should be considered one of the vital components of carriers’ data offerings.QoS will be delivered to customers’ doorsteps by 2002 for data. In an effort to preserve customers’ investment in network capabilities, carriers will offer frame relay and native Ethernet access to the MPLS backbone through a carrier-provided MPLS gateway, which, depending on a customer’s access choice, may be on the customer’s premises. The other offering will be a native MPLS connection from customer- or carrier-provided equipment.Carriers will also start providing content-hosting services for customers. The content-hosting services will involve the carriers providing the storage for voice, videoconferencing and data as well as the necessary bandwidth to support the access requirements. Under content-hosting services, carriers will also offer content-rendering capability, which will convert the format of the information. To complement content hosting, carriers will partner with ASPs and other content providers to service different industries.

New relationship will emerge

All these changes in the carriers’ infrastructure will result in a major change in the relationship between companies and carriers in terms of costs, support, hardware and SLAs.

    Carrier costs are expected to decline as they migrate from the legacy infrastructure to pure IP networks. These savings will be passed on to corporate customers by converting from usage-based billing to subscription rate. In essence, enterprises will be charged a “per-seat” type of cost based on the overall volume of information sent over the carrier network. The per-seat charge will be based on the portfolio of services being secured from the carrier (security, wireless, wired-line and so on).Customers will have to decide whether they want the infrastructure that delivers the carrier’s new service offering to be collocated at their premises or at the carrier’s. The potential equipment that carriers will be looking to install at customers’ premises or on their sites, includes caching servers, content management servers, policy servers, routers and switches.The new equipment is going to require that firms reconsider what they want to own and support. Indeed, companies that once felt it was important to own everything (hardware and configurations) should rethink their strategy. Given the increased drain on companies and personnel, the preferred strategy should be to focus on owning the configuration of the new and current equipment and outsourcing the hardware break/fix to a carrier or third party.Service and support agreements need to be constructed carefully. Companies must demand that a carrier provide an administrative Web interface where corporate IT personnel can go to submit service change requests, view bills, review service-level performance and request new services. The SLAs must have more teeth and cover more areas. However, SLAs need to cover more than just network congestion. They need to focus on the performance and integrity for applications, data and networks.Carriers have finished much of their merger-related activity, and they are now retooling for a new set of requirements and customers. It will be a complex transition with big dividends at the end. However, for this to work, companies need to adjust their thinking toward carrier networks. View them not as an external network but an extension to the network and a business-enabling tool. Then decide the best position to take to deliver the service your business needs.

Gasparro is the chief technologist for Booz, Allen & Hamilton, McLean, Inc. He can be reached at