Towards Efficient Greedy Allocation Schemes for Double-Sided Cloud Markets

The increasingly growing supply and demand of infrastructure as a service (IaaS) makes cloud trading possible and desirable in open cloud exchange marketplaces. The automation of cloud services trading in such marketplaces is an essential next step in the cloud market evolution, and it requires a proper market mechanism to effectively and efficiently trade IaaS. Due to the high problem complexity, associated with complex cloud services configurations, the research conducted so far has not resulted in an efficient and effective market allocation schemes for trading large public clouds. In this paper, we make the first step towards effective and efficient market mechanisms for cloud services exchange by addressing the problem of cloud services allocation in double-sided cloud markets. We consider realistic cloud exchange marketplace scenarios, and propose a family of combinatorial greedy mechanisms together with the two types of sorting criteria functions for homogeneous and heterogeneous resource allocation. We perform extensive experiments in order to analyse the impact of the mechanisms' parameters on the allocative performance and identify the sets of parameters that lead to the maximization of the mechanisms' efficiency based on the desired objectives. Our study demonstrates that the proposed greedy mechanisms are effective and efficient for the large-scale double-sided allocation problems. The experimental results also show that the sorting criteria for homogeneous goods is more efficient when it heavily relies on the contract's surplus, and the allocation mechanism based on resource scarcity factor is more efficient when the mechanism favours better-balanced contracts.