Saving the access network without destroying the infrastructure

The access network is the door to all services. It brings world commerce to the office and world knowledge to the classroom. It enables familiar voices to traverse cities or continents. And it speeds emergency services to their destination at three touches of a keypad. Over the decades, incumbent carriers have evolved this essential network to supply a wealth of voice and data services in addition to traditional analog telephone services. However, this evolution has not been straightforward.

The steady proliferation of workstations and personal computers brought with it an escalating need for data transmission services. Carriers rose to the challenge by progressively adding data-handling equipment to the time-division multiplexed (TDM) voice network. The resulting overlay networks were first deployed to supply high-capacity 1.544Mb/s (T-1) and fractional T-1 services. As technology advanced, 45 Mb/s (T-3) and other synchronous, high-bandwidth service terminations were provided, and carriers began delivering multimedia services via synchronous optical transport (Sonet) systems. Technology developments have led to high-speed Internet access services via various DSL standards and the emergence of native IP services delivered over Ethernet.

Because every service required specialized equipment and facilities to deliver it, the once well-structured, carefully planned voice network fairly creaked beneath the weight of multiple overlays. Yet despite the complexity imposed by overlay networks, carriers managed to operate with the efficiency needed to deliver the expected "five nines" reliability. Carriers achieved this feat through an incremental yet massive investment in operations support systems (OSS), organizational procedures and a workforce of thousands of employees trained to set up, provision, operate and maintain the network. The synergy of these resources constitutes the present mode of operation (PMO) that has kept access networks running reliably all these years.

Although the PMO carriers have developed is efficient, the overlay networks it supports have become increasingly inefficient. With wireline and wireless voice revenues leveling off, carriers realized the next avenue of profitability would be high-speed data services. This anticipation led to a heavy investment in the core transport network, which in turn created the present dilemma: The core is primed for high-speed services, but the access network isn't up to the task. The overlay networks carriers have come to rely on have grown too complex, costly and unmanageable to handle the escalating levels of data traffic. Unless carriers can solve the bottleneck caused by the "box bloat" in the access network, they may not realize their anticipated return on investment in the core network.

The complexity crisis

A typical access network includes a confusing array of digital loop carriers (DLCs), DSL access multiplexers (DSLAMs), span termination systems, optical transport aggregators and multiplexers (Sonet ADMs), and systems to cross-connect (DCSs), aggregate or disaggregate service interfaces. Most recently, carriers have also added circuit-to-packet voice gateways and Ethernet multiplexers. The number of network elements required to satisfy the projected demand for broadband would be reason enough to exchange overlay networks for simpler topologies, but vendors also seem to be retreating from acting as system integrators, giving carriers this additional unwanted burden. These factors cause a number of specific operational problems for carriers.

The Babel factor

As a rule, overlay networks are made up of specialized network elements from different manufacturers-a virtual recipe for incompatibility. Making all these boxes from diverse vendors work harmoniously forces carriers to become a systems integrator. This is not their core competency, and having to upgrade hardware and software in a piecemeal fashion is an unending burden carriers can scarcely afford.

Weakness in numbers

It is difficult to improve capital efficiency and decrease operating expenses when there are multiple types of network elements to buy and maintain. Each network element purchased brings its own common control (governing internal control, switching and synchronization) and its own cost and failure rate. Multiple network elements also require a corresponding number of interconnecting links, which must be planned for and provisioned, and are prone to connection errors and failures.

Location problems

Increasing the bandwidth to the end user requires bringing the first node within DSL reach-ideally within 12,000 feet. This necessitates remote terminals located outside the central office (CO). Once the area needed to house the network elements has been negotiated with municipalities, the controlled environment vaults (CEVs), huts and cabinets need to be powered and protected from environmental extremes. If there is anything that comes at a premium in the outside plant, it's space and power.

Obsolescence and inadequacy

Many of the NEs used today are either facing obsolescence or have already been discontinued. This is particularly true of DLCs, the mainstay of voice access. Most of the next-generation DLCs (NGDLCs) were designed a decade or more ago and have reached their advancement potential with the addition of new features-including limited broadband capability. Their inability to keep pace with the demands of today shows up in switching and processing limitations, power draw and cooling problems. All NGDLCs have limitations on the number of high-speed service interfaces they can support-some require separate transport facilities from the remote terminal back to the CO. Few NGDLCs support Ethernet, and none support it efficiently. Most NGDLCs are also incapable of aggregating data traffic at the central office interface to an asynchronous broadband, service (ATM) network, requiring an optical concentration device just to consolidate sparse traffic from multiple sources.

The promise and threat of converged systems

Consolidation is the obvious solution to the problems caused by overlay networks. By replacing multiple network elements with a single "macro-network element" that can perform diverse functions, carriers will be able to enjoy space and power savings hitherto impossible. Recognizing this opportunity, established and start-up vendors have begun developing consolidated systems and courting the incumbent carrier market. To date, however, incumbent carriers have been unimpressed with these consolidated systems because they are all disruptive to the telco PMO. While promising to simplify the access network, today's consolidated systems threaten the very infrastructure that keeps the access network functioning smoothly and allows it to scale profitably. This threat generally manifests itself by over-reliance on IP, or by causing unthinkable disruptions.

Heavy reliance on IP

Converged systems based on an all-packet network may require carriers to install voice-over-packet-to-circuit gateways for interfacing with Class 5 voice switches, while requiring their customers to adopt new customer premise equipment. Many of these systems were intended for the largely defunct competitive local exchange carrier (CLEC) market. Now, the creators of these IP-only systems are scrambling to re-target the incumbent local exchange carrier (ILEC) market, but are having difficulty getting their products to comply with ILEC standards, practices and interfaces. Vendors of IP-only systems tend to make unrealistic assumptions regarding the willingness of carriers to abandon Class 5 switches and install an entirely new IP-based switching and transmission network. 

Disruptive functional partitioning

Some vendors are trying to address both the metropolitan and the local access markets simultaneously by building switching and routing functions into their converged systems. This decision leads to additional cost and inherent complexity. More importantly, it disregards the current functional split between access, switching and transmission. This split in functionality is partly historical, but objectively, high-value service terminations are best concentrated in the network core to minimize expensive equipment and personnel since the network edge is where the multipliers are.

Bringing switching and routing functions to the access network goes against all past and current practices in functional partitioning. Field technicians, for example, are trained to deal with copper- and fiber-transmission problems and multiplexing issues. Complex functions (including switching) are best left to the central office or network operations center (NOC) where specially trained personnel are available. And if scenarios predicting the separation of incumbent carriers into wholesale and retail entities to facilitate loop unbundling come true, access multiplexing and service processing will become even more distinct and stratified.

Disruptive operations

Network equipment vendors tend to believe deeply in their product technologies, but this belief can blind innovators to a less attractive-but no less important-element of the network business equation: manageability.

How well a product can be managed ultimately determines how successfully it can be deployed. Unless it is stationed in an entirely new network, a product's best chance of success lies in its ability to fit the telco PMO. Some converged systems vendors believe that the value proposition of their product will justify retooling an incumbent's entire back office along with massive retraining of personnel, so these vendors design interfaces and operations methods to fit new technologies. But this strategy ignores the legacy software and interfaces for which legions of central office, NOC and field personnel have been laboriously and expensively trained. Other vendors assume that a token integration with legacy OSSs for fault management and loop modeling will be sufficient. Their converged systems end up lacking the kind of OSS compatibility ILECS need in their end-to-end procedures. Still other vendors believe that their proprietary element management systems are the solution to all network management woes.

The operations arena is where many an equipment provider-having demonstrated its technological prowess-loses the opportunity for product deployment. Underestimating the need for efficient, reliable and automated management in large carrier networks can spell disruption for ILECs and disaster for converged systems designers. To be successful, players in the converged systems race need to design solutions that aren't just forward-looking, but which are immediately expedient.

A winning strategy

The successful new consolidated platform will be one that truly simplifies the access network and reduces capital and operational expenses as much as possible. It should also work within the carrier's PMO to maximize its investment in Class 5 voice switches, back office operations support systems and trained personnel.

To achieve the utmost network simplicity, a new system should integrate the functions of many network elements, reduce interconnections and replace hard cabling with internalized "soft provisioning." It should support all existing services-including traditional business and residential telephone services, "high capacity" synchronous service interfaces, and the widely deployed asymmetrical DSL services. Such a system should also have "legs" so it supports emerging technologies such as DSL variants and native Ethernet services as they become commercially important, and it should also be able to deploy services built around "deep fiber" technology.

By following the current model of simple multiplexing and aggregation/concentration, a new system can avoid disrupting the functional partitioning carriers have carefully established. However, as noted earlier, successful access systems should have enough capacity and flexibility to flow with evolutionary changes in the access network over the next 15 years, respecting the depreciation cycle carriers use for their access network infrastructure.

Above all, the consolidated solution must be compatible with the OSSs used throughout the telecommunications industry. These systems may be decades old, but they still drive the work orders and tally the phone bills. Backward compatibility is therefore the key to a forward-looking system, and backward compatibility will be driven by innovative software rather than innovative hardware.

Designing for the telecom market

One can always optimize for the future, but it's folly to ignore today's realities. Engineers should keep the following goals in mind when designing consolidated systems for telco carriers:

Study customer operations, methods, and procedures.  Strive to create products that aren't just physically compatible with the telco environment, but culturally compatible as well.

Automate the most frequently used operations first. "God boxes" are dead. Instead of trying to deliver a system that is all things to all markets, focus on delivering the most useful network elements functions. Leave room for future enhancements as they become necessary to the user.

Make the system protocol-independent. The access network is a busy place. Central office, NOC and field or craft personnel all need to interface with the system, and access for all of them should be equally straightforward. The architecture should therefore support interfaces to the legacy OSS, interfaces to the more recent data management systems, and a serial "craft port" for field operations. A deployable, scaleable access system should be technology agnostic but fanatical about pragmatically delivering services to customers.

Design machine interfaces that allow new technology to resemble legacy systems. Although an access system may be "converged," it should function as though it were a sub-network of individuals NEs. This will enable CO, NOC, and craft personnel to interact with the system in a way that is familiar, efficient, and requires little or no retraining.

Plan several OS integration efforts. The system will need to undergo the OSMINE integration process if it is to be used with the prevalent operations support systems. Fund and execute the legacy OS integration plans first, and schedule enough time so that the integration effort does not impact the product release date.

Most importantly, do not be seduced into expecting an industry of such prominence to conform to every new product technology that comes along. Remember that telco carriers operate the world's largest, most reliable public network, and the carriers have long-established ways of operating. Their pragmatism is born of deep investments and the mandate to maintain five-nines reliability year after year. By creating solutions that work with-not against-telecom people, processes and systems, today's network equipment designers will help to preserve the infrastructure that makes access networks work.

Catherine Millet is Vice President of Strategic Marketing for Entrisphere Inc.

Visit Entrisphere Inc. online.