Quantifying the Gain of Dynamic Network Slicing under Critical Constraints

By providing high flexibility and abstraction of network resources and functions, network slicing allows to partition a single physical infrastructure in multiple logical independent networks, called slices. To guarantee the required network performances, slices need to be logically isolated from each other. Therefore, advanced mechanisms are needed to efficiently manage the resource sharing among slices. On one side, static slicing guarantees full isolation but requires overprovisioning of the resources. On the other side, dynamic slicing exploits multiplexing gains by sharing resources among slices. In this work, we compare the two resource allocation schemes and propose a metric to quantify the slicing gain. We show that, also under critical constraints, e.g. in terms of stringent latency and throughput requirements, dynamic slicing outperforms the static approach.