Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SlicePilot: Demystifying Network Slice Placement in Heterogeneous Cloud Infrastructures

Ioannis Panitsas, Tolga O. Atalay, Dragoslav Stojadinovic, Angelos Stavrou, Leandros Tassiulas

TL;DR

This work tackles the NP-hard problem of placement for end-to-end network slices across heterogeneous multi-clouds in 5G. It introduces SlicePilot, a modular framework that profiles real-world traffic, estimates per-VNF resource demands, and uses a disaggregated MARL scheduler to optimize VNF placement with cost efficiency and SLA adherence. Through a real multi-cloud testbed and OpenStack/Kubernetes integration, SlicePilot demonstrates significantly faster decisions (up to 19x) and near-optimal deployment costs (within 7.8% of ILP) while trading some SLA violations under high load, illustrating its practicality for real-time deployments. The approach advances network slicing by combining traffic-driven demand estimation with scalable, per-slice RL agents, enabling responsive and scalable management in diverse cloud environments.

Abstract

Cellular networks are comprised of software-based entities, with main functions encapsulated as Virtual Network Functions (VNFs) deployed on Commercial-off-the-Shelf (COTS) hardware. As a key enabler of 5G, network slicing offers logically isolated Quality of Service (QoS) for diverse use cases. With the transition to cloud-native infrastructures, optimizing network slice placement across multi-cloud environments remains challenging due to heterogeneous resource capabilities and varying slice-specific demands. This paper presents SlicePilot, a modular framework that enables autonomous and near-optimal VNF placement using a disaggregated Multi-Agent Reinforcement Learning (MARL) approach. SlicePilot collects real-world traffic profiles to estimate resource needs for each slice type. These estimates guide a MARL-based scheduler that minimizes deployment costs while satisfying QoS constraints. We evaluate SlicePilot on a multi-cloud testbed and demonstrate a 19x speed-up over combinatorial optimization methods, while keeping deployment costs within 7.8% of the optimal. Although SlicePilot results in 2.42x more QoS violations under high-load conditions, this trade-off is offset by faster decision-making and reduced computational overhead. Overall, SlicePilot delivers a scalable, cost-efficient solution for network slice placement, making it suitable for real-time deployments where responsiveness and efficiency are critical.

SlicePilot: Demystifying Network Slice Placement in Heterogeneous Cloud Infrastructures

TL;DR

This work tackles the NP-hard problem of placement for end-to-end network slices across heterogeneous multi-clouds in 5G. It introduces SlicePilot, a modular framework that profiles real-world traffic, estimates per-VNF resource demands, and uses a disaggregated MARL scheduler to optimize VNF placement with cost efficiency and SLA adherence. Through a real multi-cloud testbed and OpenStack/Kubernetes integration, SlicePilot demonstrates significantly faster decisions (up to 19x) and near-optimal deployment costs (within 7.8% of ILP) while trading some SLA violations under high load, illustrating its practicality for real-time deployments. The approach advances network slicing by combining traffic-driven demand estimation with scalable, per-slice RL agents, enabling responsive and scalable management in diverse cloud environments.

Abstract

Cellular networks are comprised of software-based entities, with main functions encapsulated as Virtual Network Functions (VNFs) deployed on Commercial-off-the-Shelf (COTS) hardware. As a key enabler of 5G, network slicing offers logically isolated Quality of Service (QoS) for diverse use cases. With the transition to cloud-native infrastructures, optimizing network slice placement across multi-cloud environments remains challenging due to heterogeneous resource capabilities and varying slice-specific demands. This paper presents SlicePilot, a modular framework that enables autonomous and near-optimal VNF placement using a disaggregated Multi-Agent Reinforcement Learning (MARL) approach. SlicePilot collects real-world traffic profiles to estimate resource needs for each slice type. These estimates guide a MARL-based scheduler that minimizes deployment costs while satisfying QoS constraints. We evaluate SlicePilot on a multi-cloud testbed and demonstrate a 19x speed-up over combinatorial optimization methods, while keeping deployment costs within 7.8% of the optimal. Although SlicePilot results in 2.42x more QoS violations under high-load conditions, this trade-off is offset by faster decision-making and reduced computational overhead. Overall, SlicePilot delivers a scalable, cost-efficient solution for network slice placement, making it suitable for real-time deployments where responsiveness and efficiency are critical.

Paper Structure

This paper contains 13 sections, 16 equations, 8 figures.

Table of Contents

  1. Introduction
  2. BACKGROUND
  3. SLICEPILOT FRAMEWORK
  4. User Plane Traffic Capturing Block
  5. User Plane VNF Stressing Block
  6. Slice Placement Block
  7. EXPERIMENTATION ENVIRONMENT
  8. EVALUATION RESULTS
  9. Configurations and Settings
  10. Placement Algorithms
  11. Experimental Results
  12. RELATED WORK
  13. CONCLUSION

Figures (8)

  • Figure 1: 5G system architecture.
  • Figure 2: UPTCB: Collecting and profiling over-the-air slice-specific user plane traffic for realistic replay and analysis.
  • Figure 3: UPVSB: Replaying slice-specific traffic through the user plane for VNF resource profiling.
  • Figure 4: Traffic profiling of eMBB, URLLC, and mMTC slices, illustrating packet rate (left) and inter-arrival time (right).
  • Figure 5: User plane VNF resource usage for eMBB (red), URLLC (green), and mMTC (blue) slices under increasing user load.
  • ...and 3 more figures