Virtualization Technology Blending for resource-efficient edge clouds

Abstract

Edge computing brings virtualized services closer to users to improve their operation (e.g., in terms of communication latency, reliability, or data privacy) and is considered as a main technological enabler for 5G and beyond ecosystems. Edge cloud orchestration approaches are typically adapting server and network resources to rapid changes in the workload, mitigate resource exhaustion, or address service performance requirements. In this context, we introduce and investigate a new cloud orchestration strategy, called Virtualization Technology Blending (VTB). VTB considers edge clouds supporting alternative virtualization options, including multiple unikernel flavors and container builds of the same or a similar service, which can be effectively blended to handle cases with challenging resource requirements, since they exhibit diverse resource requirements and performance capabilities. Here, we provide an elaboration of VTB, backed by a relevant optimization framework, in order to validate its feasibility and assess its benefits. According to our results, the proposed strategy augments the number of users that can be served up to 74% and improves the server and network resource utilization up to 50.9% and 34.4%, respectively.