Efficient Auction Mechanism with Group Price for Resource Allocation in Clouds

With the rapid grows of cloud-based internet application, a need for efficient resource allocation, load balance and cost management increases. In this paper, we propose a group-auction based mechanism for the cloud instance market to efficiently allocate resources. In the market system, resource providers offer resources in the form of virtual machine. Users submit their bids. The proposed system assists to decide how providers allocate their resources to users efficiently. Group auction is also used to reduce cost and improve sales volume for the participants. To find how to allocate resources, the utility optimization problem is formulated as a Distributed Constraint Optimization Problem (DCOP), and we apply a distributed message passing algorithm, max-sum algorithm using the group auction formation game to find the nearly optimal solution with quadratic time complexity. In comparison study, we deploy the system on two scenarios. In terms of resource efficiency and acceptance rate, we compare our algorithm with greedy algorithm and random algorithm. Experiment results show that our system performs very well in probable setting.