New Ethernet/IP in the access network

Access networks play the most critical infrastructure and operational role as carriers focus on wringing out excessive and unsustainable operating expense and seek to grow revenue. Carriers realize that access networks—strategically close to customers, yet the most expensive portion of carrier infrastructure—are ripe not only for modernization, but that making the right access architecture investment is critical to their ongoing business success.

Fundamentally, the nature of telco service delivery is changing. While the cornerstone service will always be primary lifeline voice, a carrier must deliver broadband services that require up to 100 times more bandwidth per service than traditional voice.

Over the next 10 years, nearly every carrier’s revenue streams will depend on:

• Continued support of POTS and associated special services, but with much better operating margins

• Broadband penetration growing as much as 15% and mass-market penetration rates of greater than 30% to 50%, but with much better operating margins

• A portfolio of video services, especially core services like broadcast TV and video-on-demand, while keeping operational expense in check despite the tremendous spike in bandwidth that’s needed

• An expanding uptake of business services, especially Ethernet-based connections (over copper as well as fiber), but with the ability to support them at remote terminals rather than forcing expensive direct central office connections

• An increasing demand for new voice and unified services enabled by VoIP and next-generation softswitching

• The increasing need to transport all services on the same local loop

In addition, carriers are marching forward on the transition to converged packet infrastructure—at the core, at the edge, in services—and an effective access network must be aligned with this packet evolution in order for carriers to extract the kind of value expected from these packet infrastructure investments.

Carriers understand how critical it is to make investments in the right access network architecture because of its critical role and long installed lifespan. It must simplify operations, provide a margin-aligned sustainable cost structure and offer the scale and flexibility to cope with the next 10 years of unpredictable service mix and escalating yet lumpy uptake rates of big-bandwidth services. A properly designed access network puts the carrier in control of its cost structure and its service portfolio.

Accordingly, the hunger for bandwidth at low prices will drive Ethernet into the local loop and shape the service-to-the-subscriber (STTS) architecture in the 21st century. Specifically, a New Ethernet/IP access network will enable carriers to overcome the substantial barriers to healthy margins and deliver new bandwidth-intensive services that carriers need for a sustainable business model. Access networks based on legacy technologies saddle carriers with high fixed costs, high incremental costs, high operating expenses and an inability to support even modest uptake rates of emerging services. A New Ethernet/IP access network not only overcomes these barriers, but also positions carriers with the kind of service mix and scale-range to economically cope with uncertain localized uptake rates.

Carriers’ business goals—namely, continuing to reap revenue but improve margins on POTS services and increase DSL take rates from 5% to 8% up to mass-market levels of 30% to 50%— have important ramifications for access networks. Carriers can no longer afford access networks that are not optimized to handle this uncertain mix of voice, data, and video services across a wide range of small to large remote terminal sites with entrée costs that are prohibitive and operating costs that are unsustainable. The remainder of this article describes the economic advantages of New Ethernet/IP access networks versus New SONET/ATM and NGDLC architectures. We define these three access network architectures as follows:

• New Ethernet/IP: Broadband loop carrier using a distributed architecture, Ethernet Layer 2 transport and IP Layer 3 switching

• New SONET/ATM: New designs based on traditional SONET and ATM technology

• NGDLC: Traditional legacy DLC with upgraded broadband capabilities employing a mix of TDM, ATM and SONET technology

The economic advantages of the New Ethernet/IP access network architecture are demonstrated by examining a study done by Network Strategy Partners on 8,448 subscribers served by Remote Terminals (RT) located on four access networks with a wide subscriber density range. Any protected path topology could be used without significant impact on the overall cost analysis. The wide range in node and network sizes is intended to reveal the cost of the three architectures over all of the typical sizes found in ILEC systems. Node sizes are in even, 48-port increments so none of the architectures are put at a disadvantage by using partial capacity.

The scalability and flexibility of the New Ethernet/IP architecture is demonstrated by modeling a three-phase network buildout. All three phases assume 100% POTS penetration.

• Phase I Business Model: Maximize Operational Efficiency From POTS and Data.

  Phase I models maximizing operational efficiency from traditional POTS and data services. The service mix consists of POTS and 10% data uptake. Phase I is intended to represent typical current operations.

• Phase II Business Model: Initial Triple Play.

  In Phase II data penetration increases to 25% and video service is introduced with an uptake rate of 15% of subscribers. Phase II video service is broadcast TV and Pay Per View (PPV) Phase II is intended to represent operation that may be expected one year after launch of video services.

• Phase III Business Model: Mature Triple Play.

  In Phase III data uptake increases to 50% and video services increase to 40% uptake rate. Phase III video service increases the number of broadcast and PPV channels and adds video-on-demand. Phase III is intended to represent the long term demands on the access network as broadband data and digital video become mass-market services.

Total Cost of Ownership (TCO)--consisting of capital, network operations and environmental expenses--for each of the three architectures was calculated by Network Strategy Partners. (The full study, including graphs, is available at http://www.nspllc.com.)

The New Ethernet/IP architecture has superior total cost of ownership compared to the other two architectures, enjoying a 29% total cost of ownership advantage over the New SONET/ATM design and 69% advantage over the NGDLC design. This superiority is attributable to lower capital and network operations expenses.

The New Ethernet/IP architecture enjoys a consistent capex advantage over both NGDLC and New SONET/ATM in all three project phases. It provides lower capex in Phase I where only POTS and modest levels of data are delivered and increases its advantage with the increased data and video penetration and usage modeled in Phases II and III. NGDLC becomes a particularly expensive architecture for Phase III with its high video penetration and usage.

The New Ethernet/IP architecture has a very small common cost and its port (incremental) cost is significantly less than the other two architectures. This cost advantage is due to the simplicity of the all Ethernet/IP design. The NGDLC design has especially high common cost because a separate CWDM system is needed to accommodate the video traffic added in Phases II and III. These capital expense advantages mean that carriers no longer have to pay a high upfront penalty for expensive common system components, and once installed, continue to derive per-port cost savings. This becomes especially critical as carriers seek to add RTs to shorten DSL loops in order to support more bandwidth.

Shortening DSL loops results in smaller serving areas with associated greater uncertainty in service mixes and uptake rates. The New Ethernet/IP cost structure, consequently, reduces the investment risk of service upgrade initiatives. New Ethernet/IP access networks are more economical across all sizes of RTs from the smallest (48 subscribers) through the largest (1,152 subscribers). Its Central Office Terminal (COT) also is lowest-cost. This enables carriers to adopt a single architecture and use a single product family across the entire access network, rather than using different platforms for different size RTs.

The use of fewer internal ports is a source of economic advantage for the New Ethernet/IP architecture. The NGDLC uses the largest number of internal ports in Phase 1. This is due to its use of lower speed OC-12 SONET versus the other architectures’ use of OC-48 SONET or multiple Gigabit Ethernet for internal connectivity. In Phase 3 NGDLC requires additional CWDM ports to support advanced video services. These additional ports add directly to capital expense and indirectly to both capital and operating expense. NGDLC also has the additional complexity of adding IGMP Multicast equipment to support the video service introduced in Phase 2. The excessive number of internal ports adds to design and operational complexity with resulting capital and operational cost penalties.

Efficient bandwidth utilization on the feeder network is another source of economic advantage for the New Ethernet/IP architecture. It handles voice traffic more efficiently than both New SONET/ATM and NGDLC and is dramatically more efficient than NGDLC in handling video traffic. The dramatic swing in bandwidth requirements caused by the addition of video underscores the importance of the New Ethernet/IP architecture’s capability to reduce the risk of future service uncertainty.

The New Ethernet/IP architecture has a significant feeder bandwidth advantage over the other architectures due to the incorporation of VoIP processing within the RT, permitting use of dynamic bandwidth allocation. This yields important bandwidth efficiencies since feeder capacity is only used when a caller is speaking. Since callers usually take turns speaking, this provides at least a 50% bandwidth efficiency improvement—speaker pauses add additional dynamic bandwidth allocation efficiencies.

Data bandwidth is planned as 1 Mbps per subscriber with 50:1 oversubscription in Phases I & II, reduced to 25:1 oversubscription in Phase 3. The three technical approaches consume network bandwidth for data service with about the same efficiency. In practice, New Ethernet/IP consumes less overhead bandwidth than SONET/ATM, however this advantage is generally ignored in typical network planning and so is not considered here. (That is the 1 Mbps planned per customer absorbs the overhead regardless of the amount actually consumed by the technical approach.)

Video services are planned using 3.8 Mbps per video stream. Each active channel consumes that much bandwidth from the nearest multicast site. In the NGDLC network, multicast is from the Central Office where the additional multicast capable devices have been installed. The New SONET/ATM network multicasts at every site, but incurs additional costs for IP processing capability. The New Ethernet/IP network has multicast from every site, with no additional costs.

The New Ethernet/IP architecture enjoys a substantial network operations expense advantage over the other two architectures. Service provisioning expense is the largest contributing factor to the cost advantage while NOC and Field Support also add to New Ethernet/IP’s network operations expense advantage.

The New Ethernet/IP architecture’s significant service provisioning and configuration expense advantage is due to its elimination of per subscriber/per service/per ISP virtual circuit provisioning. The New Ethernet/IP design requires only one-time specification of service policies. This eliminates most of the service provisioning and configuration work. Field and NOC support are lower for the New Ethernet/IP architecture because it employs fewer network elements and system components than the other architectures.

The New Ethernet/IP architecture is built upon the philosophy of Occam’s Razor, a logical principle attributed to the mediaeval philosopher William of Occam that says “What can be done with less, is done in vain with more.” This economic analysis demonstrates the merit of the philosophy. The simple modular design built upon Ethernet and IP yields lower Total Cost of Ownership than the Next-Gen DLC and New SONET/ATM architectures because it handles voice, data and video in one unified manner which results in fewer network elements, system components, and internal ports. This simple unified solution, also, dramatically simplifies provisioning and configuration management from a per-service, per-subscriber and per-ISP process to one-time specification of service policies.

Michael Kennedy is the managing partner and co-founder Network Strategy Partners, LLC and can be reached at mkennedy@nspllc.com.

Kris Sowolla is the director of marketing at Occam Networks and can be reached at ksowolla@occamnetworks.com.

Visit Network Strategy Partners online.

Visit Occam Networks online.