Research On Key Technologies Of Load Balancing Based On OpenStack

With the rapid development of cloud computing technology in the new era,it has become an important point in the world of IT industry.Every country in the world has attached great importance to cloud computing technology,and has conducted in-depth research and improvement.How to improve the service quality of cloud computing in the cloud environment,how to save the resource cost and how to improve the resource usage rate in the cloud environment is of great significance for cloud computing.Although the virtualization technology in the cloud environment has valuable advantages,the shortcomings are not to be underestimated.Because the cloud hosts in the cloud platform are virtualized,the performance of each configuration is quite different.In addition,the types of tasks that arrive are different,and the demand for resources is different.Therefore,these factors will cause the cloud host to experience machine downtime,which causes some nodes in the cluster to receive a large number of tasks,and some nodes have no task to process too busy and appear in the system.The embarrassing phenomenon is ultimately serious until the cluster is not working properly.In order to solve the above mentioned malignant problems,this paper selects the OpenStack cloud platform to make in-depth research on the scheduling strategy of LVS clusters applied to the platform and the back-end metadata storage strategy of OpenStack.In turn,the problem of solving the imbalance of the cluster load is achieved.This paper mainly proposes two innovations.The first innovation was optimized for the internal job allocation strategy of the OpenStack cluster.The article conducts in-depth research on the internal workings of OpenStack.The underlying mechanism of Linux Virtual Machine Cluster(LVS)applied to OpenStack,existing scheduling algorithms,etc.are discussed.Finally,a new scheduling strategy is obtained by combining LVS minimum weighted link algorithm and ant colony algorithm.The first step is to redefine the weight of the node,and secondly,according to the characteristics of intelligent optimization of the ant colony algorithm,find out the optimal set of server nodes in the cluster,and finally use the original weighting for the optimal node group.The least connected algorithm(WLC)selects a machine node with the best load performance.Then the task sent from the outside world is handed over to the best node for processing.Then,the relevant experimental scheme wasdesigned to test the ant colony optimal node group algorithm.The final result proves that the ant colony optimal node group algorithm proposed in this paper is more suitable for large-scale clusters that need to schedule resources than ant colony and WLC strategies.Another innovation goes deep into the OpenStack backend storage Ceph cluster to improve the management of metadata in the cluster.First,the node weights are reconstructed,and the migration strategy is implemented based on the traffic and the migration amount.The experimental results show that the improved scheme of metadata management is more effective than the original scheme.