Research On The Battle Clouds Oriented Information Resource Scheduling And Optimization Methods

With the deepening of the army information construction,the combat cloud has become an important development direction of the army information construction.As the U.S.military believes,cloud computing is a disruptive technology that can "change the rules of the game," and the future combat cloud will certainly become an important part of the military information grid.Although the demand,mission,main functions and usage modes of combat cloud have been demonstrated and studied,it still faces a serious challenge in the actual deployment and application-resource scheduling problem.Compared with the traditional cloud computing model,the combat clouds face the high fault-tolerant requirements brought by the confronting battlefield environment and the large data processing requirements brought by the explosive growth of massive battlefield data.Especially in the current battlefield cloud generally adopts cloud-to-end fusion architecture,the traditional cloud computing resource scheduling method not fit the environment,so it is very important to study the information resource scheduling and optimization problem under the combat cloud scenario.In this paper,we study the information resource scheduling and optimization problem under combat cloud scenarios.In order to effectively meet the basic requirements of the combat cloud environment,facilitate the further design and improvement of the information resource scheduling model,this paper decomposes the scheduling problem in the combat cloud scenario into three major tasks: computational resource scheduling,cooperative storage resource scheduling and network reserved bandwidth resource scheduling.On this basis,the main challenges and constraints that they have to face in the process of information resource scheduling are analyzed in detail,and the scheduling strategies and methods for different types of resources in the operational cloud are further designed,and the corresponding resource scheduling algorithms are given to meet the basic requirements of the operational cloud.At the same time,we try our best to exploit the advantages architecture to achieve the goal of effectively improving the utilization efficiency of operational cloud system resources.Finally,experiments are carried out to verify the proposed theory.(1)design of general scheduling framework for computing resources.This paper designs a new computational resource scheduling framework which is suitable for the actual needs of combat cloud and the requirements of cloud-to-end fusion architecture.In this paper,we first analyze the fusion architecture of combat cloud and the main characteristics of computing resource scheduling tasks.Then,based on the analysis results,a general scheduling framework for computing resources and its specific processes are designed,and a representative example is designed to describe the main work of the framework in detail.Under this framework,the operational cloud can flexibly combine different scheduling management objectives to handle different tasks.(2)research on cooperative scheduling method of combat cloud storage resources based on alternating direction multiplier methodIn order to effectively integrate and schedule the storage resources in the multi-device of the combat cloud,and significantly improve the storage performance of the combat cloud,the main storage modes in the combat cloud are systematically modeled and analyzed.Based on this model,a cooperative storage strategy which meets the requirements is designed concretely,and an executable algorithm is given combining with the alternating direction multiplier method(ADMM).Finally,the implementation of the distributed operation of the algorithm is discussed.This method considers and integrates heterogeneous information of all participating nodes in the cloud.After a limited number of iterations,a feasible storage resource scheduling scheme is obtained.This method can minimize the latency of task execution and the total cost of system operation,maximize the use of node local information and ensure the overall reliability of the system.At the same time,each node in the combat cloud can self-adaptively schedule storage resources to complete storage tasks under the premise of guaranteeing energy efficiency and controlling the risk of node withdrawal.(3)optimization of dynamic scheduling of reserved resources for combat cloud networkIn order to provide an optimal dynamic scheduling strategy to deal with the challenges faced by combat cloud in the scheduling of reserved bandwidth resources,this part firstly models the main factors involved in combat cloud network resources,and analyzes the key issues.On this basis,the main scheduling strategy is given.Considering the agility requirement of network resource reservation bandwidth scheduling,an on-line algorithm is developed based on Lyapunov optimization theory through the design of real queue and virtual queue.Finally,the theoretical performance of the proposed online algorithm is analyzed from the aspects of queue boundary and worst delay.This novel strategy can systematically and dynamically allocate bandwidth resources for multiple applications and services in a combat cloud in a locally optimal way.(4)experimental test and analysisIn order to verify the performance of the proposed strategy and algorithm,a large-scale simulation experiment is carried out on the platform of MATLAB,CloudSim and other platforms based on random composite demand and real demand data to schedule computing resources,storage resources and network reserved bandwidth resources.Through the comparison and analysis of the experimental results of various typical similar algorithms,the function and mechanism of the techniques and methods proposed in this paper,as well as the performance performance under the actual requirement data set are further discussed.The experimental process and results fully demonstrate the characteristics,advantages and disadvantages of the various methods designed in this paper,as well as the importance of practical problems,and lay a solid foundation for the further deployment of the methods in the actual production system.