Shared Network Function Placement for Cloud Native 5G Core Networks

Abstract

Providing adequate operating resources for Network Functions (NFs) with minimal costs has been one of the primary concerns for cloud providers. The adequate operating resources definition has to cover the quality of service (QoS) constraints, which result in 'underutilized resources'. Studies in the NF placement domain that make use of these resources through sharing, employ NP-Hard algorithms. In this study, we propose an NF placement scheme called Cloud Native Network Function Sharing (CNFSH) using a priority-based load balancer and a polynomial time algorithm for dynamic 5G settings. Using CNFSH, the number of satisfied slice requests increases by 36%, and the average resource consumption per network slice drops by 28% compared to the NoShare model. Besides being dynamic and sustainable, CNFSH outperforms a previous sharing scheme [1], with 5% success rate for both of the metrics.