The rise of technologies like AI and IoT has increased the need for high-performance, low-latency computing. Edge computing addresses this by extending cloud resources closer to client applications. Effective workload distribution across this edge-cloud continuum must however consider both network proximity and available computing resources to avoid potential overloads. To achieve this, we developed a computing-aware framework for dynamic traffic steering using Kubernetes API, MP-BGP, and SRv6. Specifically, we use Kubernetes APIs to determine computing capacity and BGP to advertise it, while SRv6 steers traffic based on proximity, resource availability, and QoS requirements. We implemented a proof-of-concept of the proposed framework and validated it using GNS3 and Unix-based routers running FRRouting, demonstrating its effectiveness and performance.

Keywords: Measurement;Quality of service;Load management;Low latency communication;Artificial intelligence;Optimization;Edge computing;Pragmatics;Edge computing;SRv6;CATS;load balancing;traffic steering

DOI: 10.1109/ICCCN65249.2025.11133863