Communication service provider (CSP) networks face numerous challenges today. With mobile data traffic having grown 4000-fold in the past decade, consumers’ insatiable appetite for wireless content and mobile data is pushing CSP networks to crisis point. At the same time, the gap between wireless bandwidth demand and revenues is widening – and competition from over-the-top (OTT) services is growing.
As a result, CSPs are having to rethink their networks – starting with the design and dimensioning of their network infrastructure.
For many CSPs, network functions virtualization (NFV) and software-defined networking (SDN) are keys to a new network architecture.
Enabling delivery of higher levels of automation and faster and more agile services, these technologies result in new revenue streams for service providers as well as reducing CapEx and OpEx. Indeed, according to a recent IHS Markit report, 100 percent of SPs say that they will deploy NFV at some point – and many have already committed significant resources to virtualize their networks.
SDN and NFV release agility and automation, but what does network virtualisation look like? Primarily this involves shifting data flows from a traditional three-tiered approach to a high-density leaf-spine or mesh network with an east-west flow, deployed on the network edge. To support multiple migrations, careful planning at the infrastructure level is required from the outset.
CSPs must consider how the pod infrastructure will scale in order to realize the savings available and their migration plan must also be aligned with the evolving Ethernet roadmap to ensure application compatibility. The engineering team need to determine which specific type of optical fibers should be used, decide upon the optimal fiber count for the MPO trunk cables to ensure the best fiber use for anticipated applications, and plan to maximize fiber and equipment port density while keeping the network manageable.
By addressing these key considerations, CSPs will achieve a more flexible, adaptive and cost-efficient data center environment.
SDN/NFV has the potential to radically transform network capability, customer service and organizational profitability. These technologies have caught the attention of CSPs, whose adoption has incentivized standards bodies and vendors to develop a more robust SDN/NFV ecosystem that are able to support larger and more ambitious developments.
Success here, however, will rest on CSPs’ abilities to retool their facilities to accommodate these new technologies; there remains a challenge in developing the physical layer architecture that enables carriers to realize the benefits of SDN/NFV.
To support the low-latency, high-speed requirements of a highly virtualized cloud/compute environment, CSPs are beginning to shift their architecture to a two-tier spine/leaf network which is deployed in dedicated “pods” close to the network edge. The configuration is often referred to as a Central Office Re-designed as a Datacenter (CORD). However, whether deployed as part of a mobile telephone switching office (MTSO), centralized radio access network (CRAN) hub, or regional data center, these pods are actually part of a central office. Figure 1 shows a conceptual example of an SDN/NFV pod built with a two-tier spine/leaf architecture.
The CORD configuration has several advantages. Smaller pods use lower cost multimode fiber. Whereas the traditional central office infrastructure features singlemode fiber (SMF), the smaller SDN/NFV pods have shorter link distances. This makes them better suited for lower-cost MMF optic transceivers that use short wavelength transmission technology. MMF transceivers are designed around vertical cavity surface-emitting laser (VCSEL) technology that is far less costly than the lasers used for singlemode transceivers, presenting a huge CapEx advantage.
SDN/NFV provides carriers with the flexibility, speed and management capabilities to keep up to date. However, in order to reap the benefits, DSP engineering teams must rethink their traditional central office design and architecture to handle increasing customer demand for data, content and availability. As CSPs start to move their virtualization and cloud/compute resources to the edge and into compact pods, they are able to take advantage of the increased throughput and reduced latency of spine/leaf networks. The physical layer infrastructure within the pod is crucial for service providers to see success in this newly redesigned environment.
You can learn more about considerations for modern network infrastructure here.