SPECIAL REPORTS: Optical Data Networks & Equipment
As standards addressing quality-of-service and management issues start to emerge, here's the business case for optical Ethernet.
JACK RAY, Applied Innovation
The recent financial downturn in the telecommunications industry and subsequent failure of many startup service providers seeking to bring new technologies to market are forcing many companies to reexamine their growth strategies. In hindsight, it is clear that, as an industry, we became so excited about the potential of new technologies to transform communications that we failed to pay enough attention to the business case for these new technologies.
Building long-term sustainable revenue and sufficient short-term returns on investment to maintain growth requires a strategy that focuses first on the market: What services are customers willing to pay enough for to make them profitable to deliver? That will then dictate network topology and, only then, will providers be in a position to make sound decisions regarding technology selection:
High-speed Internet access. Any discussion of subscriber services must begin with high-speed Internet access, which is rapidly approaching commodity status. Although the recent meltdown is enabling some providers to begin making small price increases, future competition is destined to drive down prices for high-speed access. That makes a business case built on providing Internet access only difficult to make. Every service provider is expected to offer this service, but only the biggest carriers will serve a consumer base large enough to remain profitable with it as their only product.
Storage-area networking. This market segment is rapidly growing and one of the few that actually demands high-bandwidth connectivity of 100 Mbits/sec or more. Although many business customers are still reluctant to outsource their data storage because of security and reliability concerns, the explosive growth in data required to serve a global customer base will make it increasingly difficult to manage these resources internally. Storage service-provider services will offer a more economical and, ultimately, more reliable solution to maintaining large data stores. Remote data storage also meets critical disaster recovery needs, preventing disruptions and discontinuity in business operations.
Video conferencing. While voice-only IP telephony solutions have found few takers, interest in videoconferencing is on the rise. Videoconferencing can dramatically reduce travel costs and greatly increase the accessibility of key decision-makers. Recent concerns for the safety of business travelers makes the business case for videoconferencing very appealing.
Ethernet virtual private networks (VPNs). VPN solutions offer corporations the ability to increase productivity by enabling secure access to critical data and applications across regional offices, providing access for telecommuters working from home and mobile workers connecting from the field. Workers can take their virtual office with them and maintain full productivity almost anywhere, if the security of these connections can be ensured.
Web hosting. Many service providers now offer turnkey Web-hosting solutions, which include not only an offsite, redundant platform with high-speed Internet access, but design and deployment assistance as well. The cost of truly high-speed, last-mile access to the Internet makes hosting these types of e-commerce sites at the business too high. Dramatically improved performance and reliability can be achieved by moving the host PCs into the core of the network where high-speed access is less costly and where service pro viders are capable of ensuring more reliable access to the business Website. For businesses engaged in e-commerce, loss of connectivity equates to shutting down the office.
Application service providers (ASPs). ASPs have struggled to find a widespread market for their products. However, growth is beginning in several key applications. Gartner Group estimates that the ASP market will hit $25 billion by 2004, up from $3.5 billion in 2001. Many corporations are seeing the value in ASP solutions for human resources software, expense reporting, e-commerce, customer relationship management (CRM), and financial services.
Transparent LAN. Providing direct network connectivity between remote offices allows the corporate network, whether locally or globally distributed, to appear as one network to users. That greatly reduces the complexity of network administration, increases performance, and enhances network flexibility and scalability. The corporate IT organization can roll out new applications and service more rapidly and at lower cost.
Network security. Today, the most critical intellectual property many corporations own is their data. Solutions that protect this data deliver high value. Service providers are responding by offering turnkey services that include customized network firewalls, automated virus scanning, protection against denial of service attacks, and even network security consulting.
Video-on-demand. The convergence of data, video, and telephony into a single network is telecommunications' "Holy Grail." We have yet to see this convergence occur for a variety of reasons. In the video marketplace, there are solutions for delivering video today that seem sufficient for most of the market. Video-on-demand, while offering additional flexibility, still competes with VCR, DVD, and TiVO systems. Any network design should anticipate being able to supply video-on-demand services in the future; however, a business case built solely on video-on-demand would be difficult to make in the short term.
IP telephony. The second leg of the converged network Holy Grail has also floundered in its introduction to the marketplace. Again, the issue is what additional value does IP telephony offer over competing solutions? Most consumers are generally satisfied with their current voice telephone services. The promise of free long distance is attracting a small market of early adopters, but dramatically falling long-distance prices, as well as competition from the wireless market, has stunted the demand for IP telephony. For new technologies to gain market share, they must offer something sufficiently advantageous over existing solutions. IP telephony has so far failed to identify any sufficiently interesting differentiation.
It's clear that while there are a range of services with the potential to drive future growth, the cost of deploying next-generation network technology probably cannot be justified by the potential market for any one of these services. A successful business strategy can only be built by offering a menu of these products from which customers can choose.
Faced with this wide variety of services and a rapidly changing marketplace, the choice of network technology is critical to success. The most common core technologies deployed today are frame relay (FR), ATM, SONET, and native Ethernet. Last-mile broadband solutions used to establish the physical connectivity to the end user include dial-up modems over POTS, xDSL, ISDN, and optical Ethernet. Which of these choices will enable the deployment of the widest range of services while also ensuring the flexibility and scalability necessary to meet future demands?
FR is widely deployed today for connecting corporate LANs. Most corporate networks are based on FR permanent virtual circuit (PVC) technology, which creates a point-to-point connection. Mesh networks are built to interconnect various offices. This approach offers sufficient quality of service (QoS) and is capable of supporting a multiprotocol environment. It can be managed with a variety of class-of-service (CoS) options to allow the service provider to offer a variety of service-level-agreement options at varying prices.
The drawbacks of FR are its rigid network architecture, cost, and the relatively low bandwidth options for most currently deployed networks. Adding offices to the mesh network usually requires adding additional PVCs to every office, making this technology less scalable and flexible than other options.
ATM networks offer the most comprehensive suite of QoS and CoS options of any networking technology. These networks can support both constant bit-rate (CBR) services, such as voice and video applications, as well as variable bit-rate (VBR) services such as data. ATM was de signed to be the next-generation in fra structure technology; however, as with all things designed by committee, the resultant overhead and complexity has made the technology costly and expensive to deploy.
ATM, like FR, is primarily a PVC-oriented network topology, resulting in large mesh networks or very large routers. An ATM network, once deployed, is a robust and reliable solution and found in the heart of most DSL network deployments around the world. But ATM has not seen the universal adoption that its designers intended, largely due to its complexity.
SONET is a widely deployed core technology. It is the entrenched technology in the core of the network and is often combined with ATM technology to offer a high-speed transport solution to the end user in the form of OC-3 (155-Mbit/sec) services. In most cases, SONET is invisible to the end user, performing only at the core with no terminating end points at the customer's network. SONET provides a time-division-multiplexed environment that can support both CBR and VBR services at carrier-level-quality standards. SONET can be deployed in a bidirectional ring topology for greatest reliability and boasts 50-msec or less restoration time in case of failure, which is sufficient in most cases to keep the end user from any noticeable service interruption.
As a core technology, SONET is king, but its high cost and inflexibility have seen the growth of SONET in the network slowing. SONET OC-192 (10-Gbit/sec) circuits are commonplace in the backbone of the Internet, and efforts are underway toward OC-768 (40 Gbits/sec), but carrying SONET overhead to these speeds makes for a costly equation.
Ethernet may be the most widely deployed networking standard, with more than 300 million Ethernet ports worldwide. Ethernet, as an end-to-end networking solution, is available for as little as one-fifth the cost of SONET and as little as one-tenth the cost of an ATM network. Its ability to directly interface to the widest range of networking devices greatly reduces the amount of required equipment. In addition, the cost of maintenance is reduced because carriers can leverage the economies of scale available in the enterprise networking environment.
IT professionals familiar with SONET are few and costly. Every network protocol transition point re quires additional expensive equipment. Converting from IP/Ethernet to IP/ATM or even IP/ATM/SONET in the core and back to IP/Ethernet at the destination network not only adds equipment cost, but it also greatly increases the cost of administration and maintenance. Ethernet is inherently scalable and works with a wide variety of physical media such as copper, cable, and fiber. It is highly scalable for large and small networks.
One challenge to promoting Ethernet as a core and metro edge technology for service delivery is the issue of QoS and security. Deploy ments today fall into two categories: IP VPN and virtual LAN (VLAN). VLAN enables the service provider to deliver a secure service to the end user at a relatively low cost. Emerging standards such as 802.1p, 802.1q, 802.1r, and 802.1s enable QoS and traffic management over the VLAN network. VLANs are relatively easy to manage, and the network topology can change without affecting the end user or altering the physical infrastructure.
Currently, VLANs face issues in larger networks based on the 4096 address limitation. Network-to-network transition points are complicated by the need to map VLAN IDs from one network to the other. That makes large-scale deployments unmanageable. More recent work on the Insti tute of Electrical and Electronics Engi neering (IEEE) 802.1q-1998 standard aims to create extended VLAN ad dressing, which will provide a solution to this limitation.
IP VPN is a more costly solution for service delivery, demanding sufficiently powerful hardware to support full line-rate encryption of the network connection to the end user. Its advantage is that it relies upon the end-user equipment to terminate the connection, so it is flexible and scales better than existing VLAN solutions. There are no inherent addressing limitations and it is compatible with any IP-based network services and applications. High-speed encryption silicon is rapidly reducing the cost of this type of solution. And there is a wide range of encryption standards to choose from.
Data encryption standard (DES) or 3DES is a direct algorithmic-based standard. IP security (IPSec) includes both data encryption and user authentication through public key encryption.
The Figure shows a metropolitan optical ring utilizing high-performance Layer 2/3 routers to create a 10-Gbit optical Ethernet backbone. The ring architecture is supported by either the IEEE 802.1 rapid spanning tree or the proposed resilient packet ring standard.
Redundant Layer 2/3 routers provide increased reliability in the collocation facility. Note that the collocation facility supports dual redundant Layer 4-7 switches for server and firewall load balancing and redundant firewalls for maximum performance and security.
The ring also has the capacity to support a server farm for Web hosting and content distribution, storage and Web caching network appliances, and an IP services delivery platform for VPN with full-line encryption between the collocation facility and end users.
Last-mile connectivity to the end user is critical to the long-term success of broadband technology. The vast majority of users still have services delivered over copper utilizing 56k V.90 modem technology. The xDSL technologies seek to extend the capabilities of the copper infrastructure by enabling network speeds of up to T1 (1.5 Mbits/sec) with the potential for rates as high as 45 Mbits/sec.
While this is a huge improvement over 56k modems, it is too limiting to enable the multiservice capabilities that will drive long-term profitability. High-speed, high-definition television-quality signals consume in excess of 20 Mbits/sec per channel. Storage networks typically are built on a minimum of Fast Ethernet 100 Mbits/sec and are migrating toward 1-Gbit/sec Gigabit Ethernet (GbE). Improvements in the ability to deliver high data rates over copper are inevitably not a final solution. These technologies may ultimately be relegated to rural environments where it is simply too costly to upgrade the copper and to intrabuilding last 100-yard deployment where the distance is limited and data rates can be maintained reliably.
Hybrid fiber/coaxial (HFC) cable networks have captured the lion's share of the residential broadband market. New technologies are emerging that may enable GbE data rates over this technology. However, most deployments have delivered a shared media to their customers, making this a less attractive service to most business users. Direct HFC connections could be deployed to the end user, but that raises the cost of the service and requires upgrading the entire cable network infrastructure. If this type of investment were to be made, it would be far more effective to install the physical last-mile solution with the most capacity: fiber optics.
Unlike any of the other last-mile solutions, the ultimate capacity of fiber-optic cable is almost unlimited. Bell Labs calculated the bandwidth-carrying capacity of currently available fiber to be in excess of 100 Tbits/sec. That is roughly 20 million times the capacity of a 56k modem connection and 600,000 times faster that T1 services. Fiber optics, much like copper deployments early in the 20th century, is the next 100-year technology.
Upgrades to copper will continue to be a band-aid solution, sufficient for near-term needs but facing inevitable obsolescence. Trends in fiber deployment technologies greatly reduce the cost of upgrading the last-mile copper plant. Tunneling devices no longer require digging up roads and sidewalks. Robotic devices string fiber-op tic cable through existing sewer lines. Metropolitan governments are subsidizing the deployment of fiber through right-of-way grants. New highway construction commonly includes fiber infrastructure.
Optical Ethernet stands alone as the most flexible, scalable, and lowest-cost technology available to truly support the rich mixture of services necessary to make both existing and emerging carriers profitable in the broadband marketplace. The almost unlimited capacity of fiber optics, combined with the ease of use and universal adoption of Ethernet technology around the world, makes this combination difficult to compete with. Emerging QoS and management standards will address many of the reliability and security issues that ATM and SONET currently offer high-end users. In a marketplace where technology is rapidly changing, the simplest solution with the greatest interoperability with end-user systems will enable service providers to be competitive over the long haul.
Jack Ray is the director of business development at Applied Innovation Inc. (Columbus, OH).
Whatever the purpose, all articles, product announcements, and advertisements in Lightwave can be reprinted to suit your informational needs.
For details, contact Kathleen McIntosh at: (603) 891-9203, fax (603) 891-0587, or e-mail: firstname.lastname@example.org.