Multi-Access Edge Computing: A Comparative Analysis of 5G System Deployments and Service Consumption Locality Variants

Abstract

Multi-access edge computing (MEC) is a key technology for 5G systems and offers cloud computing capabilities to application developers and content providers along with an information technology service environment at the edge of the network. In terms of system deployment, MEC infrastructure may, among other options, exist either at the edge of the radio access network (RAN), or at a central office (CO). Moreover, various MEC service consumption variants can be considered, depending on the degree of proximity between service consumers (i.e., MEC applications) and service producers (i.e., running on MEC platforms or exposed by service-producing MEC applications). Consequently, different system scenarios are expected to achieve different performance, e.g., in terms of the round-trip time (RTT) experienced at the user side. In this article, we present a first MEC performance case study to focus on different system deployments and service consumption localities for a number of workloads. It is observed that in all cases, the RTT is highly impacted by the MEC application workload; higher RTT is observed when a MEC application consumes remote MEC services; whereas low RTT gain is observed when moving from smart CO to RAN edge deployments, regardless of the considered workload.