Component technology can have a significant effect on capital and operating expenses.
MICK WILCOX, Cielo Communications
Over the past few years, increasing demand for bandwidth and the continued growth of the Internet fueled a build-up of communications networks, particularly in the core. Last year, the bubble burst, as buyers, carriers, and investors realized that their core networks had been overbuilt and additional equipment deployment to the core would not be required for some time.
Unfortunately, the current economic slowdown has caused many analysts, investors, and consumers to conclude that the entire network had been overbuilt. In reality, eliminating bottlenecks in the core of the network was only one piece to the bandwidth delivery puzzle. The fact that consumers are still experiencing slow download times during peak usage periods proves that many bandwidth issues remain. Therefore, the growth of bandwidth in the core did not eliminate the bottleneck; it just shifted the problems to the sectors of the network closer to the end user.
This "new" location of the bottleneck is in the metro network. The solutions and economics required in this space differ substantially from those in the long-haul (LH). The ultimate objective for the metro space is to lower the cost of bandwidth delivery to the customer, which involves increasing bandwidth and lowering the network cost of ownership. Thus, despite the decline of the LH and ultra-long haul markets, the metro market is expected to be a growth area, with some analyst estimates showing increases in metro data traffic as high as 100% per year.
While it is easy to mistake metro networks for "miniature" versions of LH networks, solutions developed for long-haul are inefficient, costly, and may even inhibit growth when applied to the metro due to its unique architecture and needs. Metro requirements are driven by the variety of protocols used, the unique topologies of metro networks, and the economics of delivering high bandwidth. As opposed to the direct transport links that characterize the core, the metro's architecture consists of fiber-optic rings or mesh architectures that run throughout the metro and connect signals from the LH network to edge access equipment, campuses, and customer premises. These rings or meshes use numerous singlemode fibers to carry multiple data signals-composed of a variety of speeds and protocols-at 1310 nm. Additionally, distances are determined by the application space, most of which are less than 25 km in link length.
Cost constraints are an equally important consideration, because they limit and delay upgrades to the network. By understanding the different elements to the cost structure, system providers can reduce and even eliminate unnecessary expenditures by selecting the best solutions.
With the increased competitiveness for the telecommunications providers, the over-arching goal moving forward is to lower the cost of network ownership. A majority of networks' total cost of ownership (TCO) comprise two major categories: capital expenditures (capex) and operating expenditures (opex). Capex, which accounts for 50-60% of TCO, consists of cable costs, including connectors and the fiber cable, and system costs for the networking equipment. Capex at the network-element level is largely driven by optical components and silicon. These components in turn drive many secondary costs associated with power supplies, chassis, cooling systems, and printed circuit boards (PCBs) to name a few.
Opex accounts for about 30% of TCO. These expenses include network administration, provisioning, maintenance, floor space, power, cooling, and battery backup. The number and efficiency of network elements largely drives these expenses, and the silicon and optics, again, are the determining components.
The demands and cost considerations of the metro's unique architecture lead to several keys for success for the network. These keys are parameters that the metro must conform to if it is to quickly and efficiently meet its goals of delivering more bandwidth while lowering associated costs. These requirements, summarized in the Table, will be critical to maintain the viability of advanced metro networks. The keys to success outline the constraints of the next-generation metro network and also provide a tool to evaluate proposed solutions.
An important factor in gauging the keys to success for the metro is whether proposed solutions are actually the best fit for the network. When taking into account the metro's different cost and technology constraints, some solutions such as DWDM do not appear to be the same panacea for the metro as they were for the LH network. Although larger metro rings (>40 km) may use 1550-nm DWDM technology, the prospect of deploying the same solutions in all metro environments is not cost-effective due to the inherently higher cost of DWDM components and equipment, which is contrary to the cost demands of the metro.
DWDM is optimally suited for applications where one of two conditions exist: a shortage of fiber, which DWDM capitalizes on by sending multiple wavelengths over a single fiber, or a link length greater than what 1310 nm can meet-anywhere from 40 to 400 km. Ultimately, there is a crossover point in the economics of deploying DWDM technology: It should be used when laying new fiber is more expensive than upgrading the components of the existing network. However, the combination of an abundance of unused (or dark) fiber in the metro with power, cost, and real estate limitations hinders DWDM technology from being the best solution for most applications.
Defining fiber-optic modules
The keys to success emphasize performance goals and cost targets for new products. As such, the complete set of metro requirements is driving evolutions in fiber-optic technology to rapidly enable the network to meet its cost and bandwidth delivery targets.
As a critical subcomponent necessary for optical-electrical conversion, optical transmitters and receivers are the prime determinants in the overall cost of a system. In fact, the type of optics used can affect nearly all areas of capex and opex. These areas include the basic cost of equipment, the facilities space required to deploy the equipment, power consumption, and ongoing operations and maintenance. Fact is that advanced metro systems cannot be economically deployed en masse with optical components that cost tens of thousands of dollars, occupy expensive board space, and are inefficient in their use of electrical power.
Along the same lines, currently available optical interconnects are large, expensive, and power-hungry modules based on traditional Fabry-Perot (FP) or distributed-feedback (DFB) lasers. There are opportunities for next-generation technology providers to alleviate this problem by providing a competitive interconnect that is smaller in size, lower in power dissipation, and lower in cost. Such fiber-optic modules would enable network systems vendors to offer solutions with higher total bandwidths, increased densities, and lower overall network cost of ownership.
Rather than being an isolated capital expense, next-generation optical interconnects actually would create a ripple effect of expenditure reductions that would affect almost every area of capex and opex, including the total cost, amount of equipment, real estate used, and power consumption of a system. Thus, the best hope for meeting growing bandwidth demand, while lowering costs, is to select fiber-optic modules capable of enabling systematic cost reductions across the board.
Therefore, just as the metro network must adhere to a general set of keys to success, so too must the optical modules. As metro networks span the gap between local enterprise and LH networks, systems designed for metro applications must be more flexible to handle a diverse range of optical input speeds and protocols. This diversity in the metro space necessitates that a next-generation metro-network element must handle traditional OC-3, OC-12, OC-48, and OC-192. It must also accommodate LAN and SAN protocols, including Fast Ethernet, Gigabit Ethernet (GbE), 10-GbE, Fibre Channel, and Escon. In contrast to the core, which puts an emphasis on maximizing bandwidth and transmission distances, the metro environment requires rate-flexible designs to efficiently address the diverse and protocol-agnostic needs of the application space.
Another must for interconnects is economical and efficient 1310-nm optics. Moderate link-to-link distances define the metro space in that 90% of the interconnects' service distances are less than 25 km. That is quite different from LH topologies where fiber is scarce and greater link distance is needed, making expensive DWDM solutions more viable than installing new fiber. The 1310-nm sweet spot is further supported because a majority of the singlemode fiber in the metro is optimized for 1310 nm, meaning that 1550-nm DWDM systems cannot be deployed without significant added expense.
1310-nm multichannel modules
As mentioned, many factors affect the cost of network ownership, including space requirements, power, cable, connectors, and system costs. Integrating 1310-nm vertical-cavity surface-emitting-laser (VCSEL)-based optical components can dramatically lower the costs of bandwidth, thereby enabling an evolution of network architectures and networking equipment. These benefits come from the inherent properties of the VCSEL that facilitate modules of smaller size, lower power dissipation, lower heat generation, and lower cost.
VCSEL technology is the key to changing the cost structure of optical components, as proved by the past success of 850-nm VCSEL technology in data communications networks. That is the result of more efficient wafer-level testing of electrical and optical characteristics, less expensive packaging, and higher yields. VCSELs also offer a substantial reduction in power consumption that positively affects many aspects of system design. Finally, VCSELs offer the only credible path to monolithically integrated laser arrays.
The impact of metro requirements on optical module design translates to smaller, denser, and more power-efficient 1310-nm optics. Today's discrete optical solutions, in the form of small-form-factor (SFF) modules, are as small as they can get due to the size of the optical connector. To continue the progress toward higher interconnect density, new packaging and laser technology is required.
One such interconnect technology that can meet those requirements is multichannel array modules. Multichannel optics use revolutionary packaging designs to incorporate monolithic arrays of lasers and detectors to create extremely dense modules, each containing multiple (eight or 12) independent optical transmitters or receivers.
The density of multichannel array modules is compelling: A line card fully populated with 18 SFF transceivers can now be replaced by a line card with six pairs of 12-channel array modules, increasing the overall port count to 72 ports (see Figure 1). That results in a four-fold increase in the aggregate port count within the same footprint. If each port supports 2.5 Gbits/sec, that's an overall increase in bandwidth per card from 45 Gbits/sec to 180 Gbits/sec. The increased port density, along with better energy efficiency, translates into lower costs associated with cabling, connectors, chassis, power supplies, and fans.
Systems and benefits
The benefits that accompany the combination of these two disruptive technologies-1310-nm VCSELs and multichannel optics-will enable unprecedented efficiencies in delivering high bandwidth to the end user. The effects of array technology on performance are dramatic; the technology offers systems several cost benefits that can be categorized according to capex and opex.
Capex. Traditional DFB-based optical systems have been unable to keep pace with the steep price/performance curve required for rapid market expansion. By shifting away from FP and DFB-based interconnects to multichannel 1310-nm VCSEL-array-based interconnects, systematic cost savings can be realized:
- Reduced power consumption. The VC SEL-based 1310-nm transmitter arrays operate at 50-70% less power than FP and DFB SFF modules, translating to savings in power supply costs and reductions in cooling costs (fans).
- Smaller size. Multichannel VCSEL arrays occupy up to 25% of the space of equally performing discrete SFF transceivers. This space savings has multiple benefits and a major impact on TCO, including 75% reduction in the necessary PCB space, elimination of multiple chassis, and facilitation of massive integration with octal serializers/deserializers
- Simplified cabling. With multichannel arrays, 24 independent fiber cables and 48 connectors are replaced with two parallel ribbon fiber cables and four connectors. Crossconnects can be effectively handled in a patch panel, simplifying the overall cable plant (see Figure 2). Thus, cabling and connector dollars are reduced, and ease of use is dramatically improved.
Opex. Although often left out of the cost equation, ongoing operational costs can be significantly reduced by the incorporation of 1310-nm VCSEL-array-based technology into next-generation metro-networking systems:
- Reduced space requirements. With increased pressure on maximizing return on collocation expenses, efficient use of floor space is critical to a service provider's ability to compete. Multichannel modules enable denser systems that increase bandwidth while actually reducing floor space footprints. Smaller footprints translate into reduced leasing costs to collocate networking equipment.
- Reduced energy costs. One of the primary ongoing expenses to a service provider is energy cost, including the cost of providing power to the systems, battery backup, generators, and air conditioning systems to remove unwanted heat generated by the equipment. VCSELs at 1310 nm consume less than half the power of current alternatives, thereby reducing power and cooling requirements for the service provider.
- Lower cost of maintenance. Higher reliability and lower power dissipation will also have a direct impact on the cost of maintenance. Because of their unique structure, VCSELs have inherently higher reliability than other lasers. These benefits have been well documented with 850-nm VCSELs and are currently being proved for 1310-nm devices. This higher reliability can reduce the number optical-interconnect field failures, thereby reducing maintenance and repair costs. Lower power dissipation translates into a reduction in the overall operating temperature of the networking equipment. Lowering the system temperature improves the overall lifetime of all components, further reducing maintenance costs.
VCSEL technology at 1310 nm is revolutionizing the design and performance of optical components. The optics affect many of the cost components of the entire metro network and have the potential to redefine the cost structure of network ownership. Figure 3 illustrates the dramatic reductions in power supplies, fans, equipment, and real estate needs that can be gained by implementing 1310-nm multichannel array modules in a system.
These multichannel modules clearly break from traditional optical components, enabling higher bandwidth while reducing capex and opex. Systems that evolve according to the keys to success, using multichannel modules, demonstrate that profitability for service providers and affordable, high-speed access for consumers need not be mutually exclusive. Indeed, 1310-nm VCSEL-based modules are more than just better components; the technology enables more efficient metro networks and brings the promise of affordable high-speed access closer to the end user.
Mick Wilcox is a product marketing manager at Cielo Communications Inc. (Broomfield, CO).