Design And Implementation Of Resource Scheduling System Based On IVCE

The efficiency of resource scheduling has a crucial impact on the performance of cloud computing environment.Now there are many researches on resource scheduling in the area of computing environment both at home and abroad,but all of these researches tend to study how to reduce energy consumption reasonably and how minimize cost as much as possible.Thus,there are few use of resource management and dynamic allocation on resource scheduling,besides,there are also fewer research on dynamic prediction and allocation in resource scheduling.However,how to select and set the appropriate resource scheduling strategy is based on the accurate prediction of resource workload in the environment.Furthermore,knowing the information of the resource dynamical availability in time plays an important role in resource scheduling.Therefore,in this paper,resource monitoring,resource detection,resource forecasting are applied to the resource scheduling system in order to realize the dynamic management of resources,The main contents of this paper including:(1)A resource availability detection algorithm based on hybrid kernel function and SOM algorithm is proposed.The traditional SOM algorithm has poor classification effect when the dataSet is non-linear.So,considering the feature of resource in cloud computing environment(Non-Linear),the paper applies hybrid kernel function in the traditional SOM algorithm which contains of global kernel function and local kernel function.The simulation experiment shows that the resource availability detection algorithm proposed in the paper has a better classification effect which makes the F_Score is increased by 29%averagely compared with the traditional SOM algorithm,the simulation experiment shows the effectiveness of the proposed algorithm.(2)A Short-Term Resource Load Forecasting Algorithm based on IVCE Platform is proposed.Usually,a single prediction model can not get good prediction result,may have poor accuracy.Therefore,a short-term load forecasting algorithm which is based on sliding window technology,improved K-means algorithm and Elman Neural Network algorithm is proposed in this paper,the combination of multiple algorithms and multiple algorithms help to avoid limitations which is caused by single model.Furthermore,experiments made in the paper shows that the improved K-means algorithm can overcome the shortcomings of the traditional K-means algorithm which needs to specify the number K of clusters in advance.In the experiment,after clustering,it indicates that the data in the same class is relatively dense and the distance between the clustering centroids in different class is far.In addition,compared with several single prediction models,the proposed algorithm error value(MAE)is reduced by an average of 0.03,the prediction algorithm proposed in this paper has an obvious advantages.(3)The Resource Scheduling System based on IVCE Platform is designed and implemented.The system contains of four modules:resource monitoring,resource detection module,resource prediction module and resource allocation module.The resource monitoring module provides data support for resource scheduling system.The resource detection module provides the resource availability detection function in order to provide reliable virtual resources for the system.The resource prediction module realizes the resource short-term load forecasting.The resource dynamically adjusts the amount of virtual resources according to the resource requirement output by the resource prediction module.In this paper,resource prediction,resource detection and resource monitoring are applied to the resource scheduling system,and a Resource Scheduling System based on IVCE State-Sensing Platform is designed and implemented.The main function of the resource scheduling system is to allocate the available resources by reasonably configuring the resources from resource pool.In this paper,the resource scheduling system monitors the virtual resource data and implements the flexible resource allocation scheme in order to reduce the unnecessary resource consumption and improve the resource utilization rate.