Future X Network Cost Economics: A network operator's journey through virtualization, automation and network slicing

Virtualization and programmability of network resources has emerged as a key prerequisite for telecom operators to efficiently deliver digital services at web scale. Driven by the high bandwidth and low latency needs of a wide variety of applications, these services require a highly-distributed network capable of flexibly supporting real-time customization of network elements. Creation of service slices, enabled by closed-loop, end-to-end automation, is a key requirement of such a network. Evaluating the economics of the digitized network resources operating in steady state is clearly important for determining its long-term viability. However, in order to evaluate and optimize their return on digitization investment, operators also need to understand the behavior of network cost and the impact of automation during the evolution itself as a traditional, physical, and static network is transformed into a virtualized, programmable and dynamic one. In this paper, we identify two intermediate stages between the two end states described above and quantify network cost behavior for a typical transformation path comprising these points. We show that the intermediate stages incur higher costs, resulting in a hump in the cost curve, due to lack of automation and incomplete virtualization. Without full automation, creating slices further exacerbates network cost performance, resulting in up to 30% cost penalty during transformation relative to the initial physical network. However, with full automation, the digitized end state can lead to 32% cost reduction. We conclude that complete automation is a critical enabler of the significant network cost savings possible through digitization.