The upcoming arrival of 400Gbit/s ZR/ZR+ QSFP-DD small form factor optics is creating high expectations. With 400Gbit/s DWDM coherent interfaces and the ability to be plugged into routers and switches, these new coherent optics promise faster connectivity between routers and switches at lower cost. In fact, the cost of these interfaces is expected to be so low that some network operators and service providers have started to rethink their network architecture and are now wondering about taking a fairly disruptive approach: Would it be possible to eliminate the DWDM optical layer? Would a network solution with routers equipped with these small-form-factor 400Gbit/s interfaces be a more cost-efficient alternative?
If you’re familiar with the cost of ROADMs, your immediate answer might be yes. But can this approach really deliver overall cost reduction? And even if it could, does it compensate the loss of the technical benefits provided by an agile DWDM optical layer? Let’s analyze it with a real-life and quite popular use case.
Figure 1 illustrates a classical network with hub-and-spoke traffic that is widely used in hierarchical networks. This network is deployed using two different technical solutions, with and without an agile DWDM optical layer. Table 1 compares the amount of IP and DWDM equipment required in each solution and shows the normalized capex cost for each network component.
In the scenario called “Agile DWDM”, the IP traffic is transmitted via DWDM wavelengths which bypass all intermediate nodes. Therefore, every hub-spoke traffic relation consists of two high-speed interfaces. With one 400Gbit/s traffic relation between the hub and each node, this solution demands a total of eight 400Gbit/s interfaces. Furthermore, this approach requires a ROADM-based optical line system (OLS) with Flexgrid function.
In the “IP-centric” scenario there is no transparent bypass. The traffic is always completely terminated and routed on the IP layer. This approach is based on a simple optical line system without ROADMs but it needs more 400Gbit/s interfaces and corresponding IP router blade space, as shown in table 1.
It’s assumed that the short node-to-node distances of the IP-centric scenario can be realized with coherent 400Gbit/s ZR interfaces, whereas the longer paths in the agile optics scenario need more expensive long reach interfaces. For a simple comparison we assume the use of coherent 400Gbit/s ZR+ in this case.
The results of the capex comparison of both scenarios are depicted in figures 2 and 3. The capex calculations show that both solutions would require a similar investment at day 1. However, the cost of the IP-centric solution would be much higher when the capacity-per-node increases. The cost savings derived from the elimination of ROADMs would be eaten by the higher number of DWDM interfaces and large routers that are required. These large blades fully equipped with multiple ports would also lead to a much higher power consumption, and therefore much higher operational costs (which is not covered by the simple model discussed here). Furthermore, the IP-centric solution would face other significant impairments such as a higher latency due to the need to terminate each packet in each router.
The results of this analysis show that IP-centric networks without an agile DWDM optical layer wouldn’t lead to significant cost savings, especially at later stages when the network capacity increases. What’s more, they wouldn’t provide the flexibility, performance and advanced features that network operators need to cope quickly and effectively with ever-changing data traffic demands.
Network operators and service providers need to invest in network infrastructure that can evolve along with customer demands. A remote reconfigurable DWDM optical layer with software-defined high-resolution flexgrid multi-degree ROADMs is key for future-proof network infrastructure that can scale and adapt to new data traffic demands and coherent technology evolution.