by Thomas Strasser
The emergence of high-definition television (HDTV), IPTV, and video-on-demand services, along with the advent of peer-to-peer video applications, has placed a strain on the bandwidth requirements and operational efficiency of carrier networks. As the growth of these high-bandwidth services puts increasing demands on today’s communication infrastructure, existing architectures and their underlying subsystems are proving costly and unmanageable.
To date, fully featured reconfigurable optical add/drop multiplexers (ROADMs) have demonstrated their effectiveness in metro core optical networks, providing the ability to provision and monitor optical wavelengths via remote management software. As the metro core network becomes automated with core ROADMs and as fiber-to-the-home deployments bring fiber (and higher bandwidths) to the residence, carriers now face new questions about upgrades, especially in that segment of the network that lies between the core and customer: Can carriers harness the total cost-of-ownership (TCO) advantages of current ROADM platforms and utilize them closer to the edge of their networks? Can they gain these advantages at modest incremental cost? What if carriers can eliminate amplifiers and have “40-Gbit/sec ready” networks that are futureproof? And, finally, how do they address the problem of uncertain demand, which has created premature network bandwidth exhaust?
Telcos and cable service providers are asking these questions, but are seemingly resigned to the fact that current ROADM economics just don’t work in the edge of the network. They are frustrated that deploying current fixed optical add/drop multiplexer (FOADM) devices forces them to accurately anticipate traffic demands and predict capacity exhaust as their networks grow. This article tackles such issues directly and presents economic alternatives to the limits imposed by earlier technologies.
Planning optical networks requires a great deal of forethought. Designers must accommodate the immediate service requirements of their networks and forecast the impact of future end-user services on the associated infrastructure.
For complex optical mesh networks and networks deep in the metro core, current core ROADM equipment provides complete automated access at all times to any wavelength at any node, with impressive cost and planning benefits. These platforms reduce network TCO and give carriers much needed flexibility. Unfortunately, such functionality comes at a price that cannot be supported as the point of deployment moves toward the edge of the network. It is here that a carrier typically depends on FOADMs for equipment (capex) savings because of core ROADM platforms’ initial prohibitive expense. As a lone alternative, FOADMs are less than perfect-networks deploying them require a more generous optical power budget, scheduled disruptions in service, capacity clairvoyance, and a tolerance for stranded bandwidth.
However, emerging edge ROADMs enable cost-effective automation in the optical edge. These recently introduced ROADMs include small, tunable fiber-optic band-pass filters and wavelength routers enabling flexible add/drop of up to eight wavelengths in the C-band to cost-effectively address low-bandwidth applications. These devices contrast against classic core ROADMs, which provide a full complement of multidegree add/drop functions for more than 80 protected wavelengths across multiple bands.
To illustrate the difference between FOADMs and this new class of ROADMs, let’s assume a simple, protected six-node backhaul network (see Fig. 1). We can use Monte Carlo simulations to evaluate the impact of uncertain traffic demands on network exhaust. For comparison with FOADM equipment, we’ll consider typical 100-GHz-spaced systems using 4-skip-0 banding filters with a system capacity of 40 wavelengths. At the hub node location in the collector ring, we’ll place a 40-wavelength demultiplexer and multiplexer.
Figure 2 quantifies the network exhaust scenarios due to uncertain planning and unavailability of bandwidth. The graph plots the likelihood that a carrier will strand a wavelength given a range of planning scenarios. The purple shaded area (“red zone”) represents the system performance that carriers settle for today. The right edge of the red zone illustrates the best-case scenario where the actual evolution of traffic in a uniform fashion matches the uniform assumption made during filter placement. In this ideal case, 50% of the networks will have stranded wavelengths by the time two-thirds of their system wavelengths are deployed, implying that carriers are effectively buying a 27-wavelength system while actually paying for a 40-wavelength infrastructure.
The left edge of the red zone illustrates the worst-case scenario stemming from the best of intentions. In this case, extra filters are deployed at one node with the expectation that more traffic will be serviced at that node-a classic case of preplanning based on market forecast. Blocking occurs quickly, with more than 90% of the networks stranding wavelengths when only half of their system wavelengths are in service. Carriers that have deployed FOADM systems confirm these results reflect wavelength utilization experienced in real networks.
More disturbing to carriers is the uncertainty associated with network exhaust. Note the overall size and breadth of the red zone for the FOADM test cases in Fig. 1. By the time 50% of networks strand at least one wavelength, an individual network may have deployed between one- and two-thirds of its wavelengths. This is an enormous gap and creates considerable ambiguity in the timing of each new network overlay. As a result, carriers cannot reliably predict when their networks will reach exhaust. Without a graceful equipment growth model, planners are unable to accurately allocate budgets and implement network overlays.
Compare these attributes to the advantageous characteristics of modular edge ROADMs. Referring again to Fig. 2, the green zone represents the same scenarios used to quantify FOADM subsystem performance, but replaces them with three-port edge ROADM subsystems. Immediately apparent are both the diminished size and movement of the green zone to the right. In all network scenarios, the three-port edge ROADM outperforms its FOADM counterpart by increasing the number of average deployable wavelengths in each network and imparting a much higher degree of certainty in forecasting network exhaust. For the first time, carriers can both dramatically increase the lifespan of a network and anticipate the next network overlay to within a couple of wavelengths.
These advantages can also apply to more complex networks in the metro core, where multiple central offices are connected with a logical mesh traffic pattern, interspersed with several edge nodes that communicate exclusively with core nodes. Figure 3 shows a 10-node metro network utilizing three core and seven edge nodes, with half the traffic being confined between core nodes and the other half going between edge and core nodes. In this scenario, we’ve placed core ROADMs (with up to 100% add/drop of any wavelength at any node) at the core locations. The comparative wavelength utilization for FOADM-based networks versus edge ROADM-based networks is shown in Fig. 4.
FOADM wavelengths are stranded in 50% of the networks before half of their wavelengths are deployed. At this same probability, the performance of the modular edge ROADMs approaches that of a network with core ROADMs at all nodes. This result suggests a profound monetary impact independent of network perspective. All cost and management advantages that the modular ROADMs bring to the edge manifest themselves in more complex networks that mix and match core and edge ROADMs. If carriers demand ROADM functionality at edge node prices, they can place the modular ROADMs at 7 of 10 locations and slash the initial common equipment costs by 50%.
The business case for core ROADM technology is now well established in the market. The TCO benefits mainly derive from the savings in labor and operational expenditures (opex). To date, these monetary benefits have not mapped well to the network’s edge, but the modular edge ROADMs change the equation. For the first time, carriers can realize the cost savings associated with core ROADMs closer to the edge of their networks and eliminate the added financial penalties of FOADM-based networks. These benefits reveal themselves when we look at a carrier’s common equipment costs as the traffic demands increase.
Figure 5 compares the capex advantage of the FOADM and the modular edge ROADM in the collector ring network of Figure 1. We calculate the common equipment costs as a function of increasing network capacity. Note that transponder costs, which will dominate the costs at high wavelength counts, are not included in this exercise. Several features stand out from this analysis.
First, we notice the significantly reduced initial cost structure of the edge ROADMs, providing a 66% capex reduction over core ROADM alternatives. We also note that edge ROADMs at the first wavelength demand a very small price premium over static FOADM systems. As network demands grow, static systems with fixed filters encounter bandwidth exhaust, leading to an early and inefficient requirement for new overlay networks. Modular edge ROADM subsystems, on the other hand, are more adaptable to uncertain sources of traffic demand and provide extended life spans, thus reducing the need for overlay networks. As a result, these systems provide the lowest capex. When one adds labor savings, improved planning, and rapid provisioning of new services to the equation, the carrier’s advantage in going toward edge ROADM-based platforms becomes economically compelling.
We also can consider a more realistic case of traffic growth to highlight the robustness of edge ROADMs. The example covers a case where a new office complex spurs significant growth at one location, which we can model as one edge node generating 3× the expected traffic demand. For this example, the inability of a static system to accommodate unpredictable traffic patterns is highlighted by the dashed line in Fig. 5. On the other hand, there is virtually no difference in network equipment cost certainty between the ideal and realistic cases when deploying flexible edge ROADM devices.
In a similar manner, the economic advantages of edge ROADMs extend to more complex networks in the metro core. The combination of incremental capex coupled with the established ROADM opex advantages enables breakthrough technical and financial gains in most network applications.
Today’s metro core and edge networks favor the deployment of edge ROADMs. For the first time, carriers can benefit from the recognized opex savings associated with core ROADMs at the incremental capex costs enjoyed at the edge. Planning and budgeting becomes easier with newfound predictability in network modeling. More efficient wavelength deployment extends the life of the network and postpones the need for upgrades and overlays. These concrete savings, coupled with the opportunity to eliminate amplifiers and prepare the network for an inevitable 40-Gbit/sec upgrade, make the advantages even more substantial.
Edge ROADM devices capture the right balance between required features and acceptable price points. These lean ROADMs complement core ROADMs and make a strong case for the elimination of all FOADMs in all edge networks that need to be agile and scalable to support the uncertain and demanding bandwidth needs of emerging services.
Thomas Strasser is founder and chief technology officer of Nistica Inc. (www.nistica.com).