Packet Design continues to receive requests from service providers who want to do multi-layer optimization and orchestration via SDN. They have long wanted to transport services over the most efficient layer automatically and at scale, but SDN has only recently made this a possibility. Here’s why the ability to compute best paths across both the optical and IP/MPLS transport network layers is such an advantage and what’s required to make it happen.
At present, operators manage these layers separately. Because the optical layer is cheaper and has lower latency, operators use it to offload certain types of traffic, such as bulk data transfers, from the packet network. However, this engineered traffic must stay there and not cross any intermediate WAN locations.
However, most traffic policy and management is currently done at the IP/MPLS layer. Layer 1 and Layer 3 engineering and planning teams rarely collaborate on network planning. Without a means to route traffic dynamically across the most advantageous network, service providers are less agile and not optimizing their excess capacity.
By merging the management and orchestration of the optical and IP/MPLS layers, SDN promises to provide a flexible pool of resources that can automatically and intelligently respond to changing network conditions and business demands. Centralized service orchestration can direct traffic across the optimum layer based on cost, policy based constraints, and quality of service requirements. Modifications can originate from the controller, the operator, or a third-party traffic engineering application such as Packet Design’s Explorer Suite.
Service providers want to use multi-layer optimization and orchestration for services and problem remediation that might now take them hours, days or weeks to plan and provision. A few of these include diverse path provisioning, services activation, disaster recovery, network capacity/throughput maximization, and data sovereignty protection. For instance, if a fiber optic cable is severed, a multi-layer auto-recovery service would calculate and provision the most appropriate alternate route in seconds – preventing loss of customer connectivity.
However, SDN controllers themselves do not have the management intelligence needed for viable multi-layer optimization. For example, a controller will not know if a cable has been cut, or how to compute the best alternate path, or ensure that the new path will not adversely affect other applications and services.
Any network automation, including multi-layer optimization, via SDN is only possible with real-time telemetry and analytics at scale. SDN telemetry consists of real-time visibility into control plane and data plane metrics such as the network topology, IGP and BGP routes, traffic utilization, current state of the network, latency, jitter, and device performance. SDN analytics are the actionable conclusions drawn from this telemetry, such as which path to take.
These must come from an SDN analytics tool such as the Packet Design Explorer SDN Platform. The intelligence from this software replicates the expertise of network planning groups, assessing the network’s readiness and capacity for making significant changes, acquiring a new enterprise customer, or turning on a new service. Once the SDN analytics software determines a workable action or solution based on telemetry data, the SDN controller or orchestrator can provision it in the network.
With visibility into the underlying optical layer combined with multi-layer analytics, operators can ensure optimal traffic engineering that takes advantage of the performance, protection, and cost characteristics of each layer. What used to take hours, days or weeks can now take minutes, increasing agility and delivering a greater return on network assets.