e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publicación
Artículo científico (JournalArticle)
Architecting Multi-Cluster Layer-2 Connectivity for Cloud-Native Network Slicing
Connecting distributed applications across multiple cloud-native domains is growing in complexity. Applications have become containerized and fragmented across heterogeneous infrastructures, such as public clouds, edge nodes, and private data centers, including emerging IoT-driven environments. Existing networking solutions like CNI plugins and service meshes have proven insufficient for providing isolated, low-latency and secure multi-cluster communication. By combining SDN control with Kubernetes abstractions, we present L2S-CES, a Kubernetes-native solution for multi-cluster layer-2 network slicing that offers flexible isolated connectivity for microservices while maintaining performance and automation. In this work, we detail the design and implementation of L2S-CES, outlining its architecture and operational workflow. We experimentally validate against state-of-the-art alternatives and show superior isolation, reduced setup time, native support for broadcast and multicast, and minimal performance overhead. By addressing the current lack of native link-layer networking capabilities across multiple Kubernetes domains, L2S-CES provides a unified and practical foundation for deploying scalable, multi-tenant, and latency-sensitive cloud-native applications., This article has partially been supported by the 6G-INSPIRE (Enabling Native-AI Secure Deterministic 6G Networks for Hyper-connected Environments) project, funded by the Spanish State Research Agency with reference PID2022-137329OB-C42; the NEMO (Next Generation Meta Operating System) European project, funded by the Horizon Europe programme with grant agreement 101070118; and by the CyberNEMO (End-to-end Cybersecurity to NEMO meta-OS) European project, funded by the Horizon Europe programme with grant agreement 101168182.
Proyecto:
EC / HE 101070118
,
EC / HE 101168182