Fair-efficient allocation mechanism with meta-types resources in cloud computing

Abstract

Fair and efficient resource allocation is a fundamental goal of cloud computing systems. However, diverse user requirements and heterogeneous resource types make it difficult to balance utilization efficiency and user-perceived fairness. To address this challenge, we propose a meta-type-based resource allocation mechanism, GAF-MT, which is based on the principle of asset fairness. GAF-MT introduces meta-types to model structured resource groupings and supports user-specific requirements while reducing fragmentation. We design a scheduling algorithm to find feasible solutions and implement GAF-MT based on GUROBI. Extensive experiments in small-scale and large-scale user environments show that GAF-MT not only ensures fairness, but also significantly improves resource utilization and maintains high performance even under high user loads.