A Distributed Market Framework for Large-Scale Resource Sharing

Current distributed computing infrastructures, such as peerto-peer networks, grids, and more recently clouds, make sharing and trading resources ubiquitous. In these large distributed systems, rational users are both providers and consumers of resources. Currently, there is growing interest in exploiting economic models for the allocation of shared computing resources that incentivize rational users. However, when the number of resource types and users increases, computational complexity of the allocation algorithms grows rapidly and efficiency deteriorates. In this paper, we propose a scalable distributed market framework for the allocation of shared resources in large distributed systems. We use mechanism design to create a pricing scheme that allocates a request for multiple resource types, by trading economic efficiency for computational efficiency, strategy-proof and budget-balance. To address scalability, our proposed framework leverages on a peer-to-peer overlay for resource discovery and management. We prototype our framework using FreePastry, a popular overlay network based on the Pastry protocol. We show that our scheme is efficient and scalable using both simulation experiments and results from the deployment on PlantLab.