Tài liệu Reducing Financial Risk docx

white paper Reducing Financial Risk in DS1 Service Delivery Executive Summary When taking into account all costs associated with DS1 service delivery, copper-based systems such as those using HDSL2 are far less expensive to implement and own than fiber-based systems for up to eight circuits to a single location. Data from North American service providers show that maintenance costs on copper circuits are higher than for fiber circuits. However, the data also show that capital outlays for fiber DS1 system hardware and cabling are higher than that of copper systems such as HDSL2. Taken together, the data show a lower cost of ownership for copper DS1 circuits. For network planners, the only sure bet is uncertainty because what the customer orders today may not at all reflect what that location needs tomorrow. The financial risk of choosing a DS1 delivery technology based upon an initial order can actually be enormous because, from one to eight DS1 circuits to a single location, the cost difference between copper and fiber DS1 technologies varies greatly with the number of circuits. Given this variable demand in the segment of customers using between one and eight DS1 circuits, copper solutions are far more economical and present far less financial risk than all other fiber- based solutions. 2 Introduction DS1 remains a steadily growing business for service providers. In July 2002, Gartner Group projected a compound annual growth rate for leased T1/DS1 services at 4.1% through 2006. Yet even as DS1 services are deployed, service providers still face important decisions in how those services are delivered in the access network: • Does the service provider use copper or fiber infrastructure? • What is the most cost-effective solution over time? Answering these questions extends beyond the simple cost of the equipment. Installation costs, maintenance activity, and pre-existing copper or fiber infrastructure are all factors that drive decisions on choice of technology for DS1 service delivery. ADC has studied these issues with the help of leading service providers in North America. This white paper provides an economic analysis of DS1 service delivery and offers important guidance in network planning. Assumptions This paper compares costs for one to eight DS1s to a single remote location at 5kft from the serving point, using the two predominate architectures and three technologies for DS1 service delivery. Those two typical networks - extensions from a SONET ring and point-to-point architecture - are shown in the following two diagrams. The diagrams also illustrate the technologies that include: • HDSL2, using the embedded copper network. • Point-to-point optical solutions, termed Fiber Loop Extenders (FLE) in this analysis. • SONET, including next-generation SONET, over the fiber network. It should be noted that although minor cost differences are seen at distances other than 5Kft, the data lead to the same conclusions within CSA (carrier serving area) radii. Six to eight DS1s are generally considered to be the point at which it becomes economically advantageous for end-users to use a DS3 rather than DS1 services, so this analysis has been limited to eight circuits. Network diagrams are shown below. SONET Ring Extension Architecture for DS1 Service Delivery Using Fiber Loop Extender (FLE), SONET, and HDSL2 Technologies 3 Data collected from major service providers in North America provide the basis for this economic analysis. The supporting data are grouped into three parts: • Maintenance costs over a five-year period; • Capital costs for hardware; and • Capital costs for cabling. The next sections of this paper explore those three component costs. Following that analysis, how those costs comprise total costs for projects are explored in two scenarios - when infrastructure is available and when it must be built. Taken individually, each set of data provides only limited insight for network planning decisions. For example, maintenance costs alone hardly provide adequate basis for choosing between copper and fiber DS1 technologies. Yet taken together, this validated set of data offers a more complete view of the costs for copper and fiber options for DS1 service delivery. Maintenance Costs The first set of variables for the model describes average maintenance activity on both copper and fiber DS1 circuits. Maintenance activity is defined as anytime a DS1 circuit is touched or tested due to alarms, customer complaints, circuit performance, and other instances. Public data from FCC filings and private data from service providers confirm that copper circuits encounter maintenance activity at a rate that is typically four times higher than that of fiber circuits. This higher maintenance rate is not surprising due to age of the copper plant and the inherent sensitivity of copper cables to environmental factors such as temperature and moisture. A detailed analysis of the technical factors comprising the copper failure rates is dealt with in a separate white paper available from ADC. The model also assumes a different type of dispatch on maintenance activities: copper circuits require more outside dispatch (75%) than fiber circuits (20%). Outside dispatch typically costs more because it usually requires both an inside and an outside technician working together. The combination of more maintenance activity and more outside dispatch on copper circuits results in higher costs for maintaining copper DS1 circuits. Compared to fiber circuits, copper is more expensive by a factor of seven. Point-to-Point Architecture for DS1 Service Delivery Using Fiber Loop Extender (FLE), SONET, and HDSL2 Technologies 4 Figure 1 shows that copper DS1 circuits are more expensive to maintain, especially as the number of DS1 circuits grows. For example, this chart shows that over a five-year period maintenance costs on four DS1 circuits are $1,440 for copper and $202 for fiber circuits. Refer to the Appendix of this document for cost details. Capital Costs - Hardware The data show that capital costs for fiber DS1 systems are much higher than for HDSL2. Using a distance of 5kft, HDSL2 hardware is the technology for copper. For fiber, capital costs for both SONET and fiber loop extender (FLE) systems are included in both point-to-point and SONET extension architectures. Equipment hardware costs do not vary with distances up to about 12kft. (Note: HDSL4 can be used to reach 30% greater distances without significant changes in the cost analysis.) It can be seen in Figure 2 that equipment costs for fiber DS1 solutions are higher than for HDSL2 for any number of circuits to a single location at 5kft in a point-to- point architecture. The data show that the gap in cost between FLE and HDSL2 hardware narrows a bit at four DS1 circuits. Even so, FLE hardware is more than double the cost of HDSL2 hardware ($2,752 vs. $1,280) at the nearest cost point, four circuits. At five circuits, FLE hardware is nearly three times more expensive than HDSL2 hardware ($5,504 vs. $1,899). Figure 2 also shows that SONET hardware is more expensive than HDSL2 hardware. At four circuits, SONET solutions are over five times as costly ($7,000 vs. $1,280). At best, SONET hardware costs nearly three times more than HDSL2 hardware ($7,000 vs. $2,469) with seven DS1s. The same conclusions hold true when comparing HDSL2 to fiber DS1 systems as a SONET extension, shown in Figure 3 - HDSL2 equipment costs less than optical solutions. The only difference is that the hardware required for delivering DS1s over fiber on a SONET extension is now even more costly than in a point-to-point architecture, making HDSL2 hardware even less expensive than fiber DS1 solution hardware. Cost details can be found in the Appendix of this document. 5 Capital Costs - Cabling Choices between copper and fiber cabling can greatly impact the overall cost of a DS1 delivery system. As shown in Figure 4, material and installation costs for copper cabling is significantly less expensive than that of fiber cabling. For example, at four DS1 circuits, the FLE fiber solution results in cabling costs that are 4.5 times as costly as that of HDSL2 ($4,500 vs. $1,000). At five circuits, the difference in cost becomes more dramatic: $9,000 for FLE vs. $1,250 for HDSL2. The details of this cost analysis can be found in the Appendix of this document. Choosing a System for DS1 Delivery When faced with choosing a technology for DS1 service delivery to a customer, network planners encounter basically one of two scenarios: • Infrastructure of copper and/or fiber is available, or • No infrastructure is available - copper or fiber may be exhausted, or the customer location is in a new development with no pre-existing infrastructure. These two situations are analyzed below. Infrastructure Available - Copper and Fiber When copper pairs and fiber cables are available, HDSL2 costs less for one to eight DS1 circuits when compared to fiber solutions in a point-to-point architecture. As shown in Figure 5, the cost differential is smallest at four DS1s where FLE hardware and maintenance combined is $234 more than HDSL2. However, the cost differential widens at five DS1 circuits where FLE costs 1.5 times more than HDSL2 ($5,756 vs. $3,699). SONET DS1 solutions in a point-to-point architecture are also more expensive than HDSL2, as shown in Figure 5. At seven DS1s, where the cost gap is closest, SONET hardware and maintenance are 147% that of HDSL2 ($7,353 vs. $4,989). If the customer later decreases service to five DS1s, SONET is now twice as expensive as HDSL2 ($7,252 vs. $3,699). In fact, SONET equipment and maintenance costs are an average of $4,901 (2.5 times) more costly than HDSL2 for one to eight DS1 circuits to a single location. On average, FLE solutions are far more cost-effective and present less risk than SONET, averaging less than 1.5 times (136%) that of HDSL2. We are faced with the conclusion that, given the changing nature of customer demand for DS1 services, HDSL2 technology delivers the least financial risk when deploying one to eight DS1s to a single location in a point-to-point architecture. 6 HDSL2 is also less expensive to deploy over existing infrastructure when compared to both SONET and FLE fiber solutions in SONET extension architectures, as shown in Figure 6. At four DS1s, an FLE solution costs 31% more than an HDSL2 solution ($3,554 vs. $2,720). The only time SONET solutions approach the costs of HDSL2 are in the 6-7 DS1 range while remaining dramatically more expensive at all other DS1 counts. As seen in Figure 6, the hardware and maintenance cost differential between HDSL2 and SONET DS1 solutions in a SONET extension architecture narrows with an increasing number of DS1s to a single location. At two DS1s, SONET is more than 3 times more expensive than HDSL2 ($4,601 vs. $1,430). This may be a cost difference that is hard for network planning to justify. There is a narrowing of the cost advantages of copper HDSL2 solutions at 6-7 DS1s. At a count of six DS1s, SONET is only 11% more than HDSL2 ($4,802 vs. $4,344), and at seven DS1s SONET actually costs slightly less ($4,989 vs. $4,853). Even so, the financial risk of a SONET solution for one to eight DS1s to a single location in SONET architecture is significant. On average, the SONET solution is $2,089 more expensive than HDSL2 for one to eight circuits. This $2,089 represents the average financial risk of choosing SONET over HDSL2 technology for a customer in the one to eight DS1 segment. When uncertainty is the only certain element in network planning for one to eight DS1s, an average $2,089 cost savings makes HDSL2 a financially less risky choice of technology. No Infrastructure Available The other scenario encountered by network planners in choosing DS1 technology occurs when there is neither copper nor fiber cables available. Now the equation involves not just hardware and maintenance costs, but also the capital costs of copper or fiber cables. In most cases, when no infrastructure is available the more likely choice for new cable plant is fiber. The ever-increasing bandwidth requirements of the access network have resulted in corporate philosophies that fiber, with its virtually unlimited bandwidth, always be installed to "future-proof" the network even when foreseeable needs only require copper. Yet in other cases installation of new copper is a valid alternative. When costs of material and installation of copper and fiber cables are added to the cost of hardware and maintenance, HDSL2 remains the least expensive DS1 delivery solution for one to eight circuits when compared to fiber solutions in a point-to-point architecture, shown in Figure 7. For example, costs for FLE and HDLS2 systems are closest at four DS1s, as shown in Figure 7. Yet the different is still large. At the closest, FLE is double the cost of HDSL2 for four DS1s in a point-to-point architecture ($7,454 vs. $3,720). However, if customer needs increase to five DS1s, FLE now costs 3 times more to implement than HDSL2 ($14,756 vs. $4,949). 7 The story remains the same with SONET solutions in the point-to-point architectures as seen in Figure 7. At four DS1s, SONET costs triple that HDSL2 ($11,702 vs. $3,720). The least cost differential is at seven DS1s where SONET costs 1.7 times more than HDSL2 ($11,853 vs. $6,739). No matter the number of DS1s in a point-to-point architecture where installation of fiber or copper is required, the economics say copper is the right choice. HDSL2 is also far less expensive than fiber solutions in SONET extension architecture when no fiber or copper is available, as shown in Figure 8. For example, at four DS1s, FLE is more than double the cost of HDSL2 ($8,054 vs. $3,720). At five DS1s, FLE costs are more than triple those of HDSL2 ($15,506 vs. $4,949). With the added cost of cabling infrastructure, HDSL2 is less costly that SONET solutions in SONET extensions, too. At seven DS1s, where costs are closest, SONET costs 1.4 times HDSL2 ($9,353 vs. $6,739). The cost gap widens quickly and significantly with more or less DS1s as seen in Figure 8. For example, at eight DS1s, SONET is 2.4 times more expensive than HDSL2 ($18,403 vs. $7,634). On average, for one to eight DS1 circuits FLE costs are $7,454 more than (2.7 times) HDSL2 over a five-year period when no cabling infrastructure is available. SONET costs average $6,026 more than (2.4 times) HDSL2. In either case, the financial risk of choosing a fiber-based DS1 solution over HDSL2 for customers in the one to eight DS1 segment is significant. Conclusion When compared to fiber-based alternatives, HDSL2 is the clear economic choice for delivering between one and eight DS1 circuits to a single location. This holds true even though the maintenance activity rate and the cost of maintaining the copper plant are both much higher than that of the fiber plant. The reason HDSL2 costs less is that in fiber solutions, capital costs for equipment and cable are significantly higher. Demand for DS1s in the segment of the market that requires one to eight circuits is variable. A business that needs five DS1s today may require three or seven in one year. The cost difference between HDSL2 and fiber DS1 systems is variable, too, at each number of circuits. On average, the cost for HDSL2 solutions is thousands of dollars less than costs for fiber loop extender (FLE) and SONET solutions. At four DS1s, paying an extra $234 to deploy a fiber solution might be reasonable. However, once that fiber solution is installed and the customer wants just one more DS1, paying an additional $2,057 for the fiber system may not be tolerable. ADC acknowledges that when faced with the need to install copper or fiber, carriers will most often choose fiber. The reasons for doing so are completely valid and justified, and are beyond the scope of this paper. However, the more likely choice to deliver DS1 services is using existing copper versus using either new or existing fiber. In those cases, the economics become a dominant concern, particularly in today's ultra-competitive and capital-constrained environment. Data from service providers show that on average HDSL2 is far less expensive to deploy. Given the variable nature of customer demand, the only way to minimize expenses and remove financial risk from DS1 network planning decisions for customers in the one-to-eight DS1 market segment is to choose HDSL2 technology. 8 Appendix - Assumptions and Calculations In comparing HDSLx and fiber technologies for DS1 service delivery, the following assumptions were used: • DS1 delivery to a single location of up to 8 circuits. • Maintenance rates and costs calculated over five years. • Both SONET extension and point-to-point alternatives examined. • Three solutions are presented at 5kft: – HDSLx, delivering one DS1 circuit over two copper pairs – Fiber loop extension, delivering four DS1s over two fibers (FLE) – SONET extension, delivering seven DS1s over two fibers Maintenance Costs - Copper Assumptions • Maintenance activity rate - 2.0% per month • Outside dispatch required on 75% of trouble calls – 4 hours 1 man outside, 2 hours 1 man inside • Inside dispatch (only) required on 25% of trouble calls – 2 hours 1 man inside * Loaded labor rate - $60/hour Maintenance Costs - Fiber Assumptions • Maintenance activity rate - 0.5% per month • Outside dispatch required on 20% of trouble calls – 4 hours 1 man outside, 2 hours 1 man inside • Inside dispatch (only) required on 80% of trouble calls – 2 hours 1 man inside • Loaded labor rate - $60/hour 9 Maintenance Costs - Fiber (continued) Costs for Hardware - HDSLx Systems Costs for Hardware - Fiber DS1 Systems 10 Costs for Hardware - Fiber DS1 Systems (continued) Cabling Costs - Copper Copper cable allocation • Cost for new copper cable includes material and installation – 100 pair distribution cable - $2.50/ft – Allocation for 2 copper pairs - $0.05/ft Cabling Costs - Fiber Fiber cable allocation • Cost includes installation, splicing, and material • 12-fiber buried cable - $5.40/ft; therefore a single fiber allocation = $0.45/ft . white paper Reducing Financial Risk in DS1 Service Delivery Executive Summary When taking into account. today may not at all reflect what that location needs tomorrow. The financial risk of choosing a DS1 delivery technology based upon an initial order can actually