The Design And Implementation Of A Distributed Cloud Storage Gateway

The widespread of Internet and exploding of information make data storage costcompany more than ever, and companies are facing unprecedented storage challenges.The newly-emerged cloud storage has been accepted gradually due to its highcost-effectiveness, expansibility, availability and reliability. Despite the rapiddevelopment of cloud storage and its huge market potential, defects like various accessinterface, insufficient security and slow access speed are found as well, which havealready become the biggest obstacle. To address these problems, this dissertationproposes a new scheme of cloud storage gateway, and implements a prototype systembased on the theory above. The new system could not only simplify the access pattern ofcloud storage gateway, but also boost its access speed exponentially, along withensuring the security of users’ data.The new system uses ordinary PC as the local server, employs a distributed cachesystem composed of ordinary PC as a fast primary storage appliance, and employs thecloud storage as the slow and massive secondary storage appliance. In the process ofswapping the file in the distributed cache to the cloud storage, the new systemimplements automatic compression and encryption, and similar automaticdecompression and decryption are also implemented when swapping files in the cloudstorage to cache. Meanwhile, this system could also provide local users with globalunified directory space and API in POSIX style. The specific work of this dissertation isorganized as follows:1) Proposes a scheme of cloud storage gateway, and implements a cloud storagegateway system based on the scheme.2) According to the application of the system, elaborates related criticaltechnologies like namespace management, address mapping, cache replacement, andreplica management in detail, and puts forward a new metadata organization and storagescheme, and an improved LNC cache replacement algorithm.3) Designs the overall scheme and detailed module arrangement of the system, andelaborates the primary flow of the system, along with how modules interact with each other.4) Elaborates the key namespace management module and cache space module,and illustrates the specific design and implementation scheme of key technologies andalgorithm, such as namespace management, address mapping, cache replacement andreplica management.5) Tests the prototype system in detail, test results show that the new system couldensure the security of users’ data, meanwhile improve the efficiency of cloud storageaccess effectively.