What is the best way to migrate my existing network infrastructure toward a next-generation solution without throwing away my existing assets?
By Vinay Rathore
The optical infrastructure is the lifeblood of network operators and enterprise users around the globe. While enterprise customers don't necessarily ask for SONET/SDH as a network requirement, they do demand the reliability associated with SONET/SDH standards including the 50-msec automatic protection switching and the resiliency associated with ring-based networks. Enterprise users, network operators, and their customers are comforted to know that such levels of reliability and resiliency are based upon well-defined standards from the International Telecommunications Union (ITU) and American National Standards Institute (ANSI).
It has come to the point where end users expect SONET/SDH levels of service reliability and flexibility regardless of the service they may need. Emerging Ethernet services such as Gigabit Ethernet and 10-Gigabit Ethernet, which have limited OAM functionality on their own, rely on the SONET/SDH layer to overcome their shortcomings. While SONET/SDH has accommodated these emerging services in the past, the networks have struggled to keep up. Extensions to standards currently support Ethernet over SONET/SDH using virtual concatenation (VCAT), Link Capacity Adjustment Scheme (LCAS), and even a modified 10-Gbit/sec Ethernet called a WAN PHY interface, which neatly fits into the SONET/SDH OC-192/STM-64 (9.9 Gbits/sec) payload. While this interface tends to be more expensive than the slightly faster 10-Gbit/sec Ethernet LAN PHY (10.3 Gbits/sec), the slight sacrifice in speed and cost is offset by the improved reliability. That is where the service innovation has ended and the challenge for a new approach begins.
Widely deployed, well understood, and liked by many, SONET/SDH is not going away anytime soon. Further, any new technology introduced into the network must be able to ensure support for SONET/SDH, accommodate a wide variety of emerging services, and ensure that high-speed services can be delivered anywhere.
State-of-the-art SONET/SDH benefits from a large installed base and customer satisfaction. Implementation of VCAT and LCAS has enabled support for Ethernet point-to-point circuits that are compliant with the Metro Ethernet Forum (MEF) E-Line standards. With its ability to overcome many of Ethernet's shortcomings, such as performance management, trouble isolation, and widely understood operational procedures, Ethernet over SONET/SDH has been one of the fastest-growing transport options for network operators.
Operators continue to leverage this installed base and continue to improve SONET/SDH reliability by migrating away from traditional ring architectures toward more efficient and more reliable mesh architectures. Mesh-based networks offer greater levels of service flexibility and network reliability because each connection can have more than one restoration path.
However, the issue with mesh-based networks is the additional complexity associated with managing all of the available capacity of a network at any given time. Because mesh-based networks rely on any-to-any connectivity, managing the available resources can place a burden on network planning and implementation personnel whose sole function is to manage the network resources. One solution to overcome this complexity is to embed intelligence into the network equipment, otherwise known as an intelligent control plane.
For the layman, an intelligent control plane is the brains behind a smart optical switch, helping make decisions from simple provisioning tasks, circuit routing decisions, and -- most importantly -- complex restoration decisions in the event of single or multiple failures. The intelligent control plane controls the entire network down to the individual circuit, connection, and physical port. It accomplishes this by instilling control and routing knowledge directly onto each and every network element.
The ITU defines a standard to enable this technology as Automatically Switched Optical Networks (G.ASON). This technology has been proven in the field to not only reduce capital expenditures by 65 percent but more importantly reduce operational expenses by up to 85 percent, a critical step to scaling the network.
While initially created for SONET/SDH-based networks, mesh-based networking technology has been extended into the Ethernet domain and offers even more promise when it comes to routing high-speed optical services because the technology is protocol independent. While mostly deployed on optical switches, intelligent control plane software can reside on any optical or Ethernet switch as well as DWDM and/or ROADM device to enable a fully automated network infrastructure that requires little or no user intervention to manage even during catastrophic failures (Figure 1). This technology will be essential as we leverage what we've learned with SONET/SDH networks and apply it to the next-generation solutions.
If not SONET/SDH, what else?
Network operators and enterprise users should be thinking about two things when they contemplate next-generation networks:
1) What services will I need to support today and tomorrow?
2) How can I ensure my long-term networking requirements (e.g., performance) will be met?
To ensure support for legacy and emerging services, the ITU worked with network operators and vendors alike to develop a new standard that would grow with the network and customer requirements. As shown in Figure 2, the path forward for optical networks has been redefined into a new standard known as the ITU G.709 Optical Transport Network (OTN), essentially defining a service-agnostic, fully transparent optical layer that can be used for any service, without sacrificing performance while extending network reach.
Defined by the ITU and accepted by ANSI in North America, OTN is now a global standard for optical networking that will enable the migration of SONET/SDH while preparing for a future of new services, higher speeds, and more Ethernet. Even more important, OTN was designed for high-speed services that include all-optical reconfigurable optical add/drop multiplexers (ROADMs), which enable the routing of optical wavelengths at 10 and 40 Gbits/sec.
The question of flexibility
The ability to build flexibility into optical networks continues to be a challenge. While optical transport networks have always been designed to carry large volumes of traffic from Point A to Point B at the lowest cost-per-bit, adding the ability to groom, re-route, and add/drop services anywhere has always been a challenge. However, with the continued price declines in the components marketplace and some innovative designs, combining flexibility of software into optical networks can be accomplished at prices equivalent to traditional rigid transport technology.
To properly add flexibility to optical networks, there are three areas that can and should be implemented: service-level flexibility, sub-wavelength-level flexibility, and wavelength-level flexibility. The combination of these three elements offers the ultimate in flexible optical networks.
Service-level flexibility can easily be accomplished by enabling software-defined line cards. Somewhat common in data equipment, support of multiple data protocols and speeds on a single interface has been a method for reducing cost and complexity. However, in the optical space this is a new paradigm.
As shown in Figure 3, this approach enables the use of a single programmable optical line card that can support multiple services types and speeds through on-demand firmware downloads that can be reprogrammed as often as needed. The financial benefits of such programmability include faster revenue delivery, fewer spares, and less time planning the network.
Sub-wavelength-level flexibility is not a new paradigm. However, legacy networks require a separate add/drop multiplexer (ADM) to efficiently groom the wavelengths, which adds complexity and cost and reduces value. By integrating an ADM as part of the transport network element, greater wavelength efficiencies can be achieved. In fact, sample network configurations show 78 percent improvement in wavelength utilization using an integral ADM as part of the network. This is ideal for low-speed services up to 10 Gbits/sec.
Wavelength-level flexibility requires optical routing, which can involve specialty components, designed to route wavelengths and groom entire WDM networks for services that require 10- or 40-Gbit/sec interface types. This level of flexibility can be achieved by using ROADMs, and depending upon the flexibility required should be able to route within the same ring or even offer the ability to route to another ring. Multi-ring interconnectivity can be accomplished using a wavelength-selective switch (WSS) for a ROADM, which offers a photonic switch for all-optical routing of services that can scale from 2.5 to 40 Gbits/sec without any hardware changes.
Once combined, wavelength and sub-wavelength level grooming capabilities create a hybrid optical/electrical ROADM as shown in Figure 4. The platform offers highly efficient routing of both high-speed and low-speed services and a level of flexibility that is ideal for the delivery of services from 10 Mbits/sec to 40 Gbits/sec. More importantly, operators can start off with a sub-wavelength level of grooming, which is low cost, and migrate to high-speed wavelength routing when the network reaches a critical mass of 10 and 40-Gbit/sec services.
Is it worthy?
Designing the next-generation network is always a challenge. The old adage, "If it ain't broke, don't fix it," does not apply here, assuming you want to grow your service revenue and lower your cost structure. SONET/SDH networks will always remain near and dear to our hearts, but flexible OTN and ROADM will replace many of them to enable new Ethernet services. The very heart of these networks will likely be controlled by a G.ASON-based intelligent control plane to fully automate the network operation, management, and most importantly, restoration.
By combining granular cost savings, dramatic overall cost reductions can be achieved. Reduction in spares, higher utilization of wavelengths, on-demand service delivery, and fully automated, self-healing network operations with service-level agreements can increase revenue and lower the total cost of network ownership by up to 60 percent. If that's not enough, consider the improvement in scalability and flexibility in your network that is offered in this new architecture.
When SONET/SDH was conceived, it was primarily designed to support voice services. With the rapid growth of new optical services driven primarily by Ethernet, it is clear that SONET/SDH has reached its limit. Leveraging the low cost-per-bit of transport networks, with the flexibility of a transparent optical protocol that offers optimal grooming, network automation and greater levels of resiliency are no longer just lofty goals for network operators, but are available today. As network operators work to sustain legacy revenue and migrate their SONET/SDH infrastructure toward a service-enabled, OTN-based ROADM network, they must consider the flexibility they will require and evaluate the proper blend of hardware-based flexibility and software-based automation that will serve their needs.
Vinay Rathoreis senior director of product marketing at Ciena (www.ciena.com). He has 16 years of experience in service provider networking from Sprint, MCI, Global One, and Alcatel. His areas of expertise include network engineering, operations, sales, and marketing in both wireline and wireless systems and with networking technology that includes IP, MPLS, Ethernet, ATM, and optics.