Abstract
Cloud computing is a new emerging computing paradigm that advocates supplying users everything as a service. Compared with grid computing, the focus of resource management problem is transformed to resource virtualization and allocation rather than job decomposition and scheduling. It is more urgent to find better solutions for cloud resource allocation than ever before. Although there have been some research efforts in grid computing, most of them aim at maximizing utility of system and lack of analysis for competition between different users. Some researches consider competition analysis, but they assume that common knowledge is certain and known for every user, which is difficult to be applied in a global distributed cloud environment. In this paper, we hereby propose a new resource pricing and allocation policy where users can predict the future resource price as well as satisfy budget and deadline constraints. Experimental results prove that resource price can gradually converge to an equilibrium state by dynamic games and that cloud users can receive Nash equilibrium allocation proportion without other competitors' bidding information.