Research And Implementation Of OpenStack Storage Optimization And Load Balancing

OpenStack is one of the most active open source projects for cloud computing,it provides a complete infrastructure-as-a-service(IaaS)solution that can be used to provide services to the enterprise,and set up a public cloud platform to provide cloud services to the outside.Multiple components of OpenStack require back-end storage system support,Glance is responsible for store image data,Cinder is responsible for store volume data,Nova is responsible for store virtual machine data.The original OpenStack components Glance,Cinder,Nova's back-end storage technology is different,OpenStack needs to maintain three different back-end storage to meet the object/block/file and other access methods,which lead to cloud platform system maintenance costs are high,the interaction performance between components is poor,data storage reliability is not high and so on.How to provide the appropriate storage system has become one of the key issues to improve cloud platform performance.With the rapid development of cloud computing,more and more users choose to deploy the application services on virtual machines of the cloud platform.OpenStack virtual machine resources are pre-allocated,when the user requests increased,a single virtual machine will not be able to bear.At present,there are few researches on load balancing of virtual machines in cloud platform,most of them use traditional dynamic load balancing algorithm,the overall efficiency is not high.Therefore,how to combine the characteristics of OpenStack cloud platform to achieve load balancing,is also one of the urgent problems to be solved.In view of the above problems,this paper studies and implements Ceph unified storage technology and load balancing technology based on OpenStack cloud platform,and optimizes the performance of the system.First,build OpenStack unified block storage system based on Ceph RBD.By solving the problem of independent storage of Glance image,achieve OpenStack image layer,optimized for software deployment,and support for load balancing.On this basis,the load balancing strategy is designed according to the characteristics of the cloud platform,design and implement load forecasting algorithm based on BP neural network and support the implementation of online learning program.Finally,through the comparison experiment to verify the technical feasibility,the experimental results show that the proposed method can improve the performance of the improved cloud platform.