Data Distribution And Hybrid Redundancy Method In Multi-cloud Storage

With the rapid development of cloud storage,more and more enterprises and personal users move their data to cloud storage system.Many drawbacks such as vendor lock-in problem arise when relying on only one cloud storage provider.The drawbacks of single cloud have promoted the concept of multi-cloud storage.User data is distributed over multiple cloud storage providers in multi-cloud storage method.Data distribution method has become a research hotspot since proper distribution method can not only avoid vendor lock-in problem but also improve the storage performance and availability.But there are some defects with the existing multi-cloud storage methods,which include:(1)Using only one data redundancy technology.Existing distribution methods always use only replication method or erasure code to store user data.(2)Lacking data deduplication.Too much redundant data will occupy storage and cause a waste of money for users.(3)Being not user-friendly.This paper studies data distribution method in multi-cloud environment.The main contents of this paper are as follows:(1)The drawbacks of single cloud storage and the significance of multi-cloud storage are presented in the paper.Then a comparison to existing distribution methods is made and the architecture of multi-cloud storage is introduced.(2)This paper proposes a Data Distribution Selection Framework(DDSF)in Multi-cloud.DDSF gets user requirements through GUI,which is user-friendly.A specialist system based on rules calculates the weight of QoS parameters and then ILP solver calculates a feasible distribution solution.(3)A hybrid redundancy method(HRMD)based on deduplication is also proposed in this paper.HRMD adopts MD5 Hash algorithm to detect redundant data and uses both replication method and erasure code to distribute data.Redundancy technology used is based on the access frequency of the data.