Research On Data Classification And Resources Scheduling Strategy For Cloud Storage System

With the advent of cloud computing era,more and more people will store data in the cloud storage,cloud storage brings convenience,but also increases rapidly with the increase of data volume.High energy consumption and low efficiency of data center have become one of the factors restricting the development of cloud storage.In view of the shortcomings of single consumption reduction technology and incomplete consideration of user quality of service(QoS)factors in cloud storage systems,This paper aims to combine data classification strategies to meet consumers multi-perceived QoS constraints,according to the current load situation of the system and the various QoS requirements submitted by consumers.Classification strategy stores data in different areas,and flexibly stores data according to different heat to achieve consumption reduction.This paper discusses the background and development status of the subject,and elaborates the significance of studying energy-saving strategies.The key technologies involved in this paper,such as data classification strategy,VM placement and migration strategy and CloudSimDisk simulator.Based on the analysis of the current research status of energy-saving technologies at home and abroad,this paper proposes a hot-spot data classification strategy(HDCS)for energy-saving of cloud storage systems,and a strategy for placement and migration of virtual machines(VMs)in cloud storage centers.The specific research work is as follows:(1)In order to make better use of data and reduce power consumption,we design and propose a hotspot data classification strategy(HDCS)for energy saving in cloud storage system.HDCS firstly uses data classification strategy to divide data into three types:hot data,cold data and duplicate file area,and places them in different storage areas.In view of the data classification strategy,the corresponding performance and energy consumption model is established,and the energy-saving properties of the data classification strategy are analyzed theoretically.The theoretical deduction proves that the proposed HDCS can effectively reduce energy consumption.In addition,in order to verify the energy-saving effect of HDCS,two different methods(HDFS default,HDCS)are compared.The simulation results based on CloudSimDisk show that HDCS can reduce energy consumption,with an average reduction of about 12%.(2)With the further development of virtual technology,the use of virtual machines has become a trend.Usually,several or dozens of virtual machines will be placed on a physical machine.In cloud storage centers,how to place and migrate virtual machines with minimal power consumption is a key research issue.Virtual Machine(VM)migration is the process of migrating VM from one physical server to another.It provides a variety of benefits for data centers in various scenarios,including improved performance,fault tolerance,manageability,load balancing and power management.However,due to the constraints of Service Level Agreements(SLAs),sometimes the migration of VM will lead to performance degradation,especially in the case of meeting key business objectives,which can't be ignored.In this paper,we propose a new VM placement and migration algorithm,which takes into account the quality of service requirements of different users,and can also reduce energy consumption and SLA violations caused by insufficient CPU utilization in data centers.The research work mainly focuses on resource allocation based on heuristic algorithm,which allocates users tasks in the form of Cloudlets to virtual machines that consume the least energy.The model is implemented using a service-oriented architecture and simulated using the CloudSim toolkit.Finally,we compare our algorithm with non-power sensing(NPA),dynamic voltage and frequency scaling(DVFS),single threshold(ST)strategy and minimization migration strategy(MMP).The experimental results show that the proposed algorithm can use lower energy consumption under the premise of guaranteeing customers QoS.In summary,in view of the existing problems of cloud storage system consumption reduction technology,the proposed MMP algorithm based on QoS constraints has made a beneficial attempt to reduce consumption on the premise of meeting customers QoS requirements,and achieved certain results.