Research On Optimization Method Of Replicaallocation In Distributed Block Storage System

With the rapid development of cloud computing,big data,mobile,social networks and the acceleration of enterprise digital transformation,data will continue to grow exponentially and explosively.Faced with the growing rapidly total data and the increasingly high requirements of real-time access,current storage architectures and replica allocation methods face enormous challenges and bottlenecks.Most current algorithms only consider the balance of data distribution,and it is easy to cause unbalanced use of resources in clusters with different hardware configuration.In a complex and diverse physical environment,how to provide a reliable,efficient and reasonable allocation strategy is the core of a distributed storage system.Therefore,the storage replica resource allocation optimization problem of has become a hot issue in the field of distributed storage.On this background,research on optimization method of replica allocation in distributed block storage system has important and practical value.This paper is mainly as follows:Current distributed block storage system's replica allocation algorithm only considers the storage space usage rate and cannot make full use of other physical resources.According to this problem,this paper proposes a replica balance allocation model.This model considers the CPU,memory,storage space,storage performance and network bandwidth.In order to solve the model effectively,an improved whale optimization algorithm is proposed.This algorithm combines the beetle antennae search algorithm and differential evolution algorithm to improve the search ability of the algorithm and avoid falling into the local optimal solution.Contrasting experiments have been carried out on different cluster sizes with many swarm intelligence optimization algorithms and the replica balance allocation model is applied to the actual system.The experimental results show that the improved whale optimization algorithm is better than other swarm intelligence optimization algorithms in solving the replica balance allocation model.The use of replica balance allocation model in the actual system can improve the balance of the physical resources of the system.In some application scenarios,the distributed block storage system consider not only the balance of physical resources but also the localization of data resources.According to this problem,this paper proposes a multi-objective replica allocation model that considers physical resource balance and data localization.In order to solve this model effectively,an improved multi-objective whale optimization algorithm is proposed,which is based on the improvedsingle-objective whale optimization algorithm,combined with the external archive and crowding distance sorting technology.Contrasting experiments have been carried out on different cluster sizes with different multi-objective optimization algorithms and the multi-objective model is applied to the actual system.The experimental results show that the improved multi-objective whale optimization algorithm is better than other algorithms in convergence,distribution and coverage.The use of multi-objective replica allocation model in an actual system can greatly improve the system's data localization and access performance while slightly reducing the system's balance.