Research On Optimization Policies Of Resource Provision Based On Cloud Computing Environment

Cloud computing is a new business computing model with resource pool. It is the development of distributed computing, parallel computing, utility computing and grid computing. Cloud computing allows users be able to use the cloud service through network, users can acquire or release resources in accordance with their actual needs. Cloud providers provide users the required resources through the cloud data center with virtualization technology. Due to the different purchase time, different tenure of use and different brands of cloud data center servers, existing cloud data centers tend to be heterogeneous. With the development of cloud computing, the scale of cloud data centers are also increasing. Resource provisioning and high energy consumption problem in cloud computing environment are becoming critical factors that affect the performance of cloud computing.This paper studies cloud resource provisioning technology systematically and optimizes the resource provisioning polices from the aspects of payment and energy-saving in the cloud computing environment. First, a multi-dimensional resource combinatorial pricing mechanism is proposed to address the issues in the single-resource pricing and to reduce the high purchasing costs in a heterogeneous cloud environment. The pricing mechanism comes from CPU, memory, storage and bandwidth on the system. A cost-optimized resource provisioning policy (CORPP) based on the game theory is applied on this mechanism to maximize the system utility. Not only a Nash equilibrium is considered between cloud users, and furthermore, CORPP also considers Stackelberg equilibrium between clients and the cloud provider. Second, aiming to the problem of high energy consumption of datacenters in the cloud computing environment, this paper models the energy consumption with the respect of migration energy consumption. We further propose an energy-aware virtual machine migration algorithm EABFD. Through the efficient migration of virtual machines, the cloud servers whose load are below the low threshold can be turned off or switched to the sleep state and thereby reducing energy consumption.This paper details the simulation platform CloudSim and implements the proposed CORPP and EABFD algorithms by expanding the CloudSim. The simulation results show that the multi-dimensional resource combinatorial pricing is more suitable for expressing pricing in cloud environment when running various cloudlets. Additionally, CORPP reduces users’total costs effectively compared to the commonly used Round-Robin policy when using both random cloudlets and the PlanetLab cloudlets. The EABFD algorithm can effectively reduce the energy consumption of datacenter in the cloud computing environment compared to the NEA algorithm, the DVFS algorithm, the ST algorithm and the DT algorithm.