Research On Key Technologies Of Cloud Storage

Nowadays,cloud storage technology which focuses on the storage and management of data is getting more and more attention due to the massive data stored in the network.In the cloud storage network,data is stored in the cloud data center.Data may be intruded by illegal users,resulting in data leakage or even tampering.In addition,the failures of equipment and the impact of various natural or man-made disasters,results in the loss of data.It has brought huge losses to users and service providers.Therefore,it has become a hot research area on how to design an effective storage strategy to improve the reliability of data in the cloud.Firstly,the research work on redundant data storage schemes commonly used in cloud storage services has been done.There are two types of the scheme: data storage schemes based on complete replication redundancy and data storage schemes based on erasure code redundancy.In order to solve the problems of large access time and low data reliability in current data redundant storage strategies,an adaptive data redundancy scheme for cloud storage(ADRS)is proposed.ADRS combines fragment replication with LT code to solve the problem that the existing redundant strategies can not adapt to the dynamic network environment.Simulation results show that ADRS fully takes the advantages of both fragment replication and LT code.It can reduce access latency and improve the reliability and stability of the system at the cost of increasing a small amount of storage space.In cloud storage systems with frequent failures,data replicas are widely used to improve data availability.However,when users need a large number of remote replicas,most replication schemes do not adequately consider the cost and latency.Therefore,an improved dynamic replication management scheme(IDRMS)is proposed.Firstly,the minimum number of replicas needed in the current network under availability requirements is calculated.Then the candidate nodes for storing replicas are selected by using B*tree.The location of replicas is adjusted by introducing prediction model.Also,the nodes' popularity is defined to transfer requests from hot nodes to cold nodes,so as to balance the load between nodes.Finally,simulation results show that IDRMS outperforms other classical replication management schemes in terms of communication cost and load balancing in cloud storage systems.