Transitioning through uncertainty with multiservice provisioning
SPECIAL REPORTS: SONET/SDH, ATM & IP
Despite recent instability in the telecom world, shrewd network operators are saving costs by adopting multiservice provisioning platforms.
JOHN HEMINGWAY, Metro-Optix
The year 2001 was marked by uncertainty in life, uncertainty with the economy, and never-before-seen uncertainty in the telecommunications industry. While we can't solve all the world's problems, we can examine how service providers might address the changes occurring in network infrastructure as the move is made to converged networks in local and metro areas.
Transport systems are driven by the underlying traffic. In long-distance networks, the situation has been relatively simple. Service providers have chosen their preferred core traffic type-TDM voice, ATM, or IP-and focused on the infrastructure that gives the lowest cost per bit. Typically, high-channel-count DWDM systems feed core routers and high-capacity switches.
Quality of service (QoS) is provided by fixed bandwidth for voice circuits; multiple service classes with associated provisioning issues for ATM; and over-engineered capacity for IP traffic, migrating to MPLS with traffic engineering. Significant investment-and possibly over-investment-has taken place in long-distance routes, and the next logical bottleneck to address is in the metro space.
Demands and drivers
In local and metro areas, solutions for traffic growth and new services are much more uncertain. Demand is unpredictable, traffic mix is undetermined, customer churn is increasing, and end users are demanding a faster response for short-term requirements. There is no doubt that today's legacy infrastructure will transition to a "converged" voice and data network, but how fast and to what end result is the subject of much debate. From the customer premises, along the last mile, to a local point of presence, one factor seems certain: There will continue to be a patchwork of access methods. That translates into a requirement to handle a mixture of traffic types in the metro area, which shows no sign of converging.
Despite the conflicting demands, service pro viders share several key drivers in the metro area:
- Continued need to balance both voice and data traffic in networks. Rome wasn't built in a day, and network architectures are not redesigned overnight. Networks and their underlying traffic types migrate over time. Significant revenues continue to be voice-based and corporations are reluctant to switch mission-critical voice to data networks.
- Increased capacity requirements in the metro area. Traffic demands continue to increase and end-user broadband demand is unabated-eased only by cost-effective solutions.
- Competitive metropolitan environments. Deregulation has given rise to new opportunities, particularly in more profitable metro areas, to operators with solid business cases.
- Increased data traffic in carrier networks. Data traffic is continuing to explode, with new applications continually emerging. However, data revenues are not increasing in-step.
- Reduced infrastructure costs. Growing competitive pressures continue to place downward pressure on profit margins. Reduced infrastructure and, above all, reduced operating costs are necessary for long-term success.
These drivers are encouraging service providers to seek new solutions for today's uncertain environment. In the metro transport market, parallels can be drawn to recent uncertainty in the local Class 5 central switching market. Most observers agree that traditional Class 5 switches are outmoded, as can be seen by the precipitous drop in TDM-based equipment orders by such heavyweights as Lucent Technologies and Nortel Networks. However, next-generation packet-based systems have been slow to take off and are still considered immature and unproven. Most of the new service providers, basing their business plans on packet-only service, are now in serious financial trouble or defunct.
Undoubtedly, the prudent way is to take an incremental approach to service migration, based on a flexible infrastructure that can produce solid profits today by migrating to a cell- or packet-based structure as economics warrant. Unfortunately, taking this approach with traditional equipment leads to highly inefficient operations-overlays of incompatible devices, multiple different sets of equipment with unique management systems, and slow and time-consuming end-to-end service times. New techniques are needed to reduce cost, footprint, and power consumption while improving "service velocity" to customers.
Given the inefficiencies of older equipment, service providers must examine the latest metro transport alternatives. An apparently overwhelming number of options, both from new and established vendors, is in the potential shopping cart, but some careful screening can reduce the list significantly. Additional risk and uncertainty are definitely not the goal. Selection of the most flexible option will be critical to future success-and several metro technology alternatives exist.
DWDM has clearly been successful in the long-distance transport sector, based on the lowest cost per bit of already-consolidated traffic. Move to the metro area, however, and the solution is more murky-metro traffic patterns are more unpredictable; traffic is typically of lower capacity; more services require local switching and grooming; multiple traffic types need more dedicated wavelengths; and fiber quality is highly variable. These differences result in relatively high-cost solutions. The newer coarse WDM (CWDM) may be more applicable to metro areas, since there are fewer supported wavelengths and a wider spacing that is not so "fiber-demanding." However, standards and interoperability issues also come into play.
Data-centric solutions are another future possibility. Ethernet has appeal due to its low-cost user interfaces, user familiarity, and ease of use as well as new definitions such as IEEE 802.3z 1000B-LX defining Gigabit Ethernet up to 5 km on singlemode fiber and nonstandard 1000-LH up to 70 km.
Several service providers (Ethernet local-exchange carriers) are taking this route. But the applications have limited QoS guarantees when moving off-net. Resilience is a key issue that has yet to be addressed in Ethernet solutions, despite moves to rapid spanning tree restoration; rapid (50-msec) switch times and proactive performance monitoring are not available for business-class voice circuits on data-only solutions.
Recognizing the attractiveness of packet-based systems, the Institute of Electrical and Electronics Engineers is working on a new set of standards: 802.17 resilient packet ring, which addresses issues such as reliability, data-centric packet transport, and optimizing ring and mesh architectures. The standard is scheduled for completion in early 2003 but may differ significantly from today's plans due to major differences in opinion among the contributing committee-member companies.
Traditional SONET/SDH has been unsurpassed as a transport mechanism in the TDM world, largely based on reliability. Its ability to support 50-msec switching to backup paths, combined with extensive performance monitoring features, has resulted in the familiar 99.999% reliability in the carrier transport arena. Traditional SONET/SDH add/drop multiplexers had no data interfaces, but more recent equipment now offers "data-aware" interfaces such as the ATM user-network interface and Ethernet.
The use of SONET for data transport raises some issues: inefficient bandwidth usage of variable data rates when carried in a fixed-size pipe; more unpredictable demand than voice; and long service provisioning times. These issues are gradually being addressed. However, multiples of additional external equipment are still required, such as physical interfaces and patch panels to connect between rings and digital-crossconnect systems for grooming-all of which can be done electronically with next-generation equipment.
Next-generation SONET equipment currently on the market provides a combination of support for legacy traffic while efficiently migrating to a cell and packet network. Described as multiservice provisioning platforms (MSPPs), or bandwidth managers, these new products consolidate the functions of multiple network elements into a single element. These devices support legacy traffic via SONET/SDH to offer significant improvements in density and scalability, typically up to four high-capacity rings connected to multiple tributary rings in a single shelf.
To avoid manual fiber and electrical patch panels, high-capacity digital crossconnects are built-in as integrated features, typically at the STS-1 (52-Mbit/sec) and virtual tributary 1.5 levels. MSPPs are "data-wise," able to terminate multiprotocol data streams and perform either transport-only or full switching and consolidation locally.
MSPPs continue to support SONET/SDH structures for reliability and performance monitoring, but address data issues through other techniques. ATM can be transported in a SONET/SDH payload, or the MSPP may terminate the ATM signals and perform a service access multiplexing function or switch to allow statistical multiplexing or oversubscription for ATM virtual circuits.
Packet traffic can be handled at Layer 3 with integrated IP router functionality or at Layer 2 with virtual concatenation of Ethernet signals onto variable-sized SONET/SDH channels (from 1.5 to 1,088 Mbits/sec). The latter case has the advantage of being compatible with existing SONET/ SDH networks. With virtual concatenation, only the endpoints need to be upgraded, enabling the use of legacy transport.
Ideally, a protocol-agile MSPP can terminate and process traffic in its native format, with no protocol conversion. Transforming all traffic types to a common protocol such as ATM for subsequent switching and transport is an alternative method but has drawbacks, including additional cell/packet headers, assembly/disassembly latency, and delays-most noticeable with voice applications.
Despite the uncertainty in the telecommunications world, one thing is for certain: The demand for telecom services is not going away, and long-term success will come to operators who provide the most cost-effective and innovative solutions. Shrewd network operators are turning to MSPPs to lower the cost per bit of both legacy and data services to create an economical "converged" voice- and-data-network infrastructure.
MSPPs allow a flexible approach, which can be easily adapted to uncertainties in the marketplace, while improving service provisioning times and increasing bandwidth efficiency.
John Hemingway is a director of product management at Metro-Optix (Allen, TX). He can be reached at firstname.lastname@example.org.