The intense competition for residential and business video, voice and high-speed Internet services is forcing MSOs to look for new revenue-generating engines. The result? MSOs are re-evaluating their next generation of services and the networks required to sustain them for the long run. Faced with the challenge of future growth and market leadership retention, MSOs require an optimal network transformation path, one that will enable them to increase and differentiate their existing residential offerings, while simultaneously addressing the business services market – all over a single, optimized, converged network.
Traditionally, cable operators have deployed solutions based on fixed, overlaying, but separate networks, comprising best-of-breed WDM, MSPP and Ethernet switch platforms. While addressing the need for increased capacity and fiber relief, the result is a costly, complicated and inflexible implementation.
Provisioning, maintaining and deploying such networks is a complex affair that requires multiple manual operations and frequent re-engineering of the network, resulting in high total cost of ownership (TCO).
For example, adding new channels or rerouting existing ones in such rigid networks is complicated, labor intensive, and usually affects the flow of traffic. Even minor changes can require re-engineering of the network.
A far more efficient approach is to deploy a converged optical Ethernet networking solution, integrating Carrier Ethernet with reconfigurable optical add-drop multiplexer (ROADM) technologies.
Figure 1: MSO target architecture.
The major benefits of this approach include simplified management, shorter service provisioning time, reduced network complexity and significant cost savings. ROADM and Carrier Ethernet switch routers (CESRs), as well as sophisticated, unified, end-to-end network management systems, hide the complexity involved in the integrated, simultaneous management of the optical and data layers.
ROADMs allow carriers to route any wavelength (or any combination of wavelengths) from/to any node, without the need to install additional devices, thereby reducing time-to-market for new services. In addition, ROADMs offer “colorless” ports, enabling the operator to remotely select, and then reselect, the specific wavelength to add/drop at the node. This inherent flexibility is of particular benefit when traffic is hard to predict, or when its load, end-points or routing is expected to change frequently. When coupled with optical transport network (OTN) technology, these next-generation systems offer longer reach, easier maintenance, improved resilience and better quality of service (QoS) through unified monitoring of all wavelength and sub-wavelength services.
A further improvement is to include built-in packet support to better cope with Ethernet services by mapping packet data into WDM lambdas or OTN payloads. With a WDM overlay in place, CESRs can be directly connected to their own wavelength channels and can share the same fiber. Moreover, CESRs and ROADMs can be integrated to form an agile, multi-service transport solution. This is done via unified management, enabled by multilayer, multi-technology, end-to-end service management. The convergence of data and optical layers provides Ethernet aggregation and scaling capabilities for growing metro and regional networks.
MULTILAYER TRAFFIC ENGINEERING
The shift in function of DWDM from “dumb pipes” to an intelligent transport layer has created a need for new management tools. Service providers require a single, multilayered network management system (NMS) to efficiently and seamlessly manage increasingly complex networks. The NMS provisions, monitors and controls all network layers, managing the optical and Ethernet layers, as well as the physical layer.
Wavelengths can be used, dropped and shifted, to be picked up and reused as needed. This creates an optical Ethernet infrastructure that allows rapid provisioning of new connectivity paths for the ever-increasing amount of data traffic. Only a single, integrated management system can provide a complete view of the entire network at a glance, and can include network optimization tools that enable rapid, efficient network design based on network demand and span information.
The powerful advantages of this approach are best illustrated with the following real-life scenario.
Figure 2: Multilayer traffic engineering.
A set of local hubs serving the end customers lies on one physical ring. The physical ring is subtended by a headend that provides access to all services (such as broadcast TV, VOD and Internet). This is a relatively simple, yet realistic, realization of the MSO architecture presented in Figure 1. This situation can be modeled as a ring topology (bottom of figure 2) with a hub traffic pattern (top of figure 2). The challenge is to configure the optical layer, which determines CESR interconnectivity, and the routing tables that are dictated by such switch connectivity. The most obvious possibilities are:
• Ring-based architecture: Each switch is connected to its neighbor. In this structure, all traffic is groomed in each one of the switches.
• Star architecture: There is a (protected) lightpath between each one of the hub switches and headends.
While the ring-based architecture is optimal when traffic per site is low, it becomes increasingly inefficient as traffic grows, since switch capacity is wasted for unnecessary grooming. For instance, when the traffic volume entering/leaving each site is close to the wavelength capacity, the optimal scheme is star optical architecture. In this case, entering data traffic is consolidated by the Carrier Ethernet switch and then encapsulated on a wavelength that is transparently transported up to the headend, saving switching ports, as well as optical client interfaces. However, when the flow per site is low, this scheme becomes inefficient and results in poor wavelength utilization.
This simple example (fixed topology and steady traffic pattern) serves to illustrate that no configuration is perfectly suited to any bandwidth scenario. One option is preferable to the other under certain conditions, and neither has the flexibility to justify its implementation as an optimal solution.
When advanced mathematical optimization techniques are employed, we find that as traffic grows, there is a succession of different configurations that provide optimal resource utilization corresponding to minimal network cost. Clearly, the ring is the optimal solution for very-low data flow, but when data approaches the wavelength capacity divided by the number of hubs, clusters provide a far more efficient solution. In Figure 3, a particular example comprising six hubs is presented. Note that when traffic reaches 1/6 of wave capacity, two clusters of three nodes each (where traffic is groomed in one of the hubs) provide the most efficient aggregation scheme. However, when traffic reaches 1/3 of wavelength capacity, a reconfiguration is necessary, since two hub clusters would ideally provide the optimal solution.
The diagram illustrates successive network reconfigurations as data per site grows from very low values to wavelength-level capacity. The optimal solution should therefore provide optical reconfiguration, combined with path re-optimization on the data plane. This points to a flexible optical layer, combined with unified management of the optical and data layers, which is the only way to minimize the capital expenses, even in this simple case!
INTEGRATED PACKET-OPTICAL TRANSPORT SYSTEMS
As the heart of any operation center, the management system plays a central role in opex savings. Advanced management of packet-optical networks takes advantage of the converged transport layer for provisioning, while providing tiered multi-technology views (packet and optics) for easy navigation. Advanced maintenance and monitoring capabilities, such as multilayer alarm correlation and root-cause analysis, are key features and include suggested proactive actions for reducing network troubleshooting time. In addition to standard features, such as the ability to display available, active or blocked channels in real time for any route in the network, the NMS also includes detailed performance monitoring at the wavelength and service level for complete QoS visibility.
Figure 3: Capex analysis: Minimal network cost (packet-optical integration) vs. traditional approach.
The integration of the transport Ethernet and flexible optical layers facilitates additional capital savings, such as:
• Elimination of regenerators, making the ROADM-based network more cost effective than a pure DWDM network (typically beyond the fifth wavelength).
• Elimination of back-to-back connections between separate systems, a change that typically results in up to 45 percent savings.
• Remote provisioning and reconfiguration, eliminating the need for massive upgrades whenever additional wavelength services are provisioned.
• The use of one converged solution with the same look and feel and alarm systems throughout, reducing network operation and maintenance by minimizing the need for team re-training.
Cable operators expanding their IP-based service offerings can benefit from a unified network architecture for IP, Ethernet and MPLS services in a common framework that bonds WDM/ROADM and Carrier Ethernet technologies. The management system with the ability to control all network elements is a key factor for delivering basic and advanced transport services with a simplified point-and-click interface.
The unique value of converging ROADM capabilities with Carrier Ethernet lies in its flexibility, providing virtually seamless implementation of any service, at any time, at any location within a unified transport platform. Multi-degree ROADM technology provides tightly integrated networks and provisioning of new high-capacity services, and Carrier Ethernet capabilities dynamically allocate and manage traffic transported on the network. The key enabler is a single, integrated management system providing a holistic network view that includes optimization tools.
These functionalities allow MSOs to build next-generation networks in a cost-effective manner by reducing the number of necessary components. Moreover, converged packet-optical solutions require less switching capacity and network interfaces on the switches and routers, thereby minimizing the size of the routing tables. All this simplifies network architecture, since fewer platforms are required for installation, management, maintenance and troubleshooting.
The converged network is a lean, mean telecom machine – a cutting-edge response to current demands that not only addresses the needs of operators, but also sets a new standard for network efficiency and service quality.