Performance Study of Large Scale Network Slice Deployment in a 5G Core Testbed

Network slicing in 5G forms a pivotal enabler to satisfy unprecedented requirements in terms of data rate, latency, and capacity in the form of heterogeneous network services namely, enhanced Mobile Broadband (eMBB), ultra Reliable Low Latency Communications (uRLLC), and massive Machine Type Communication (mMTC) by building end to end logical network. In this paper, we study the performance of various network slice deployment modes in a Network Function Virtualization (NFV) based 5G Core (5GC) for High Availability (HA) and Reliable functioning keeping in mind Capital Expenditure (CAPEX) and Operating Expenditure (OPEX). Our experimental results on the in-house 3GPP compliant testbed prototype of 5GC show that the completely shared slice mode saves the resources and its cost (about 66.6% compared to the isolated mode). However, it risks HA and reliability of slice service by consuming additional time (about 40 seconds) during activation, for serving a maximum of 90K subscribers. Whereas, the completely isolated mode caters to reliable slice service using dedicated resources but, with higher CAPEX and OPEX to Telecommunications Service Provider (TSP)s. A reasonable middle-ground partially shared mode of slice functioning balances both reliability and cost concerns. We believe that this work will guide TSPs in building autonomous, smart management, and orchestration of network slice deployments.