Table of Contents
Fetching ...

ClusterSlice: A Zero-touch Deployment Platform for the Edge Cloud Continuum

Lefteris Mamatas, Sotiris Skaperas, Ilias Sakellariou

TL;DR

ClusterSlice addresses the challenge of deploying Kubernetes-based edge cloud slices over heterogeneous bare-metal and virtual resources. It adopts a declarative, CRD- and Operator-driven architecture with node-level Resource Managers and cluster-level Slice Operators to automate, standardize, and reproduce edge deployments across multiple domains and Kubernetes flavors. The work contributes an open-source platform that unifies infrastructure-, platform-, and application-as-a-service for edge slices, demonstrated via a multi-cluster Liqo-based demo spanning local testbeds and CloudLab. The results suggest zero-touch deployment, high reproducibility, and scalable federation across domains, enabling rapid experimentation and evaluation of edge-networking configurations. The platform's modular abstractions and broad compatibility position it as a practical tool for edge-continuum orchestration and research.

Abstract

We demonstrate ClusterSlice, an open-source solution for automated Kubernetes-center deployments for the edge continuum. ClusterSlice is an infrastructure-as-a-service, platform-as-a-service, and application-as-a-service solution, supporting: (i) declarative deployment slice definitions; (ii) infrastructure-on-demand capabilities over multiple heterogeneous domains; (iii) composable Kubernetes deployments, supporting multi-clustering as well as various Kubernetes flavors and intra-cluster/inter-cluster network plugins; (iv) configurable application deployment; and (v) experimentation automation.

ClusterSlice: A Zero-touch Deployment Platform for the Edge Cloud Continuum

TL;DR

ClusterSlice addresses the challenge of deploying Kubernetes-based edge cloud slices over heterogeneous bare-metal and virtual resources. It adopts a declarative, CRD- and Operator-driven architecture with node-level Resource Managers and cluster-level Slice Operators to automate, standardize, and reproduce edge deployments across multiple domains and Kubernetes flavors. The work contributes an open-source platform that unifies infrastructure-, platform-, and application-as-a-service for edge slices, demonstrated via a multi-cluster Liqo-based demo spanning local testbeds and CloudLab. The results suggest zero-touch deployment, high reproducibility, and scalable federation across domains, enabling rapid experimentation and evaluation of edge-networking configurations. The platform's modular abstractions and broad compatibility position it as a practical tool for edge-continuum orchestration and research.

Abstract

We demonstrate ClusterSlice, an open-source solution for automated Kubernetes-center deployments for the edge continuum. ClusterSlice is an infrastructure-as-a-service, platform-as-a-service, and application-as-a-service solution, supporting: (i) declarative deployment slice definitions; (ii) infrastructure-on-demand capabilities over multiple heterogeneous domains; (iii) composable Kubernetes deployments, supporting multi-clustering as well as various Kubernetes flavors and intra-cluster/inter-cluster network plugins; (iv) configurable application deployment; and (v) experimentation automation.
Paper Structure (4 sections, 2 figures)

This paper contains 4 sections, 2 figures.

Figures (2)

  • Figure 1: A Multi-Cluster & Multi-Domain Slice Descriptor
  • Figure 2: Compute Resources after the Slice Deployment