Research And Application On Data Possession Audit In Cloud Storage System

Cloud storage derives and develops from cloud computing technology and theory. So, it becomes one of the most attractive distributed storage method because of easy to use, pay-as-you-go and independence of time and space limitation. However, provided that CSP lacks of trust constraint, cloud storage client can not effectively gain control of management of their own data. The resulting security problems brought by it constrain wide application and development of cloud storage technology. The techonology of data integrity guarantee is key technology of guaranteeing data security in cloud storage and existing integrity detection research mainly focuses on improvement of verification algorithm, support of data dynamic update and restoration of bad data etc., most of them are still in theory research stage so they can’t suit for cloud storage environment mainly built in the way of cluster.This paper focus on building integrity detection model in cloud that is applicable for cluster environment as application target, studies and analyse current research status of existing data possession audit in detail and propose a cloud data integrity detection model-Parallel Audit for Data Possession(PADP) based on Provable Data Possession(PDP) and Data Possession Audit(DPA) that combines with specific characters of clustered storage environment. PADP comprehensively takes advantage of PDP and DPA and completely separates users from storage service provider, centering on the trust of a third party audit and adding audit log mechanism so that reduces users’storage and computation overhead with the security of audit process being ganranteed. This paper improves homomorphism verification algorithm based on RSA in PDP model, proposes and design a MapReduce-based Parallel Generation Challenge Proof Algorithm(MR_PGCPA) to problems of unsuitability for clustered storage environment in classic data possession audit scheme. And through theoretical analysis, this algorithm proves a good algorithm had excellent security performance. Finally, this paper designs and implements a PADP prototype system based on above model and algorithm and gives detailed use case, module and details of algorithm’s process design and implement. Test results show that, PADP scheme can reduce the user’s communication traffic and memory overhead significantly; When the size of storage file is large and the number of compute nodes is great, with the introduction of the MR PGCPA algorithm, the corresponding computational overhead can be reduced dramaticlly and the efficiency of detection can be improved.