Consumer Equilibrium And Pareto Optimality Based Cloud Computing Resource Allocation Strategies

As a new computing pattern, Cloud computing connects to computers through Internet, abstracts dynamic resources from the hardware and software by the use of open technology and standards, and provides to users appearing in the state of service, so users can get a superior user experience. However, as a new thing, most parts of Cloud computing are still not solved well, such as resources scheduling, and how to use reasonable tactics to guarantee users'satisfactory and prove servers'utilization is the most important aim of resources scheduling. Traditional scheduling methods usually judge it by servers' utilization or responding time, but the ultimate aim of Cloud computing is satisfy users'needs and whether the users can get the most benefits through the less cost. Obviously, traditional scheduling methods cannot do that. Under the circumstance of resources scheduling, it should change from a system-centric to user-centric, as to improve customer satisfaction in the premise of ensuring system performance.This paper does many researches about Cloud computing and resources scheduling methods summarizes the problems, then applies Consumer Equilibrium theory and Pareto Optimal theory to the resources scheduling of Cloud computing and brings up the model and algorithm. This algorithm is different from the traditional resources allocation strategies and fit to the commercial characteristics of Cloud computing, which mainly reflects in two points:first, it is order to enhance consumer satisfaction and maximize the users'utilities from the consumer point of view; second, it can improve the servers'utilization and make full use of the limited resources. Combining these two theories aims at maximizing the tasks utilities and resources utilization. After a reasonable analysis, this paper finishes a test by the use of simulation software called CloudSim, which simulates two cloudlets to be allotted resources to get the maximum utility under the conditions of limited resources, known cloudlets costs, known cloudlets preferences, known resources cost.In addition, the paper extends the basic Pareto optimal theory model on two tasks and two kinds of resources into multitasking model of the two resources for the further work.