Performance and viability analysis of deploying cloud-native 5G autoscaling platforms

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

5G mobile network technology is undergoing rapid deployment. Autoscaling in 5G refers to the dynamic allocation and removal of network functions based on real-time service demand. It provides additional capacity to serve new users, while avoiding the risk of excessive costs. In this paper, we compare two stateless 5G autoscaling platforms: CoreKube and free5GC-helm, both deployed on the Hetzner Cloud platform. We utilize PacketRusher to generate high load for autoscaling evaluation, and collect metrics for analysis. Additionally, we analyze the bottleneck and autoscaling problem of free5GC-helm, providing guidance for real-world deployment. Our investigation revealed that the free5GC-helm scaling mechanism quickly encounters bottlenecks, primarily due to decisions made within the network repository function.
Original languageEnglish
Title of host publicationin Proc. The 1st free5GC World Forum 2025, in conjunction with ACM CCS 2025
Publication statusAccepted/In press - 11 Aug 2025

Keywords

  • 5G
  • network function virtualization
  • Autoscaling
  • Kubernetes
  • Stateless Network

Cite this