Research Of Data Deduplication In Data Disaster Tolerance Systems

With the rapid growth of information storage requirements in modern enterprises,the collection and processing of information has become one of the key technologies related to the survival and development of enterprises.At the same time,the reliability and security of information have also received widespread attention.Among them,data disaster recovery system is a kind of technical means that can effectively protect data security.The explosive growth of data directly brings storage space insufficient,Moreover,which brings forth challenges to the system's processing power and data transmission.Data deduplication is urgently required to alleviate these problems.The primary research contents of the article are as follows:Similarity detection is very critical to the performance of data deduplication.This technique reduces memory consumption and improves the data deduplication ratio by sampling.The paper improves the Enhanced Position-Aware Sampling algorithm(EPAS),proposing minimum value sampling algorithm based on Content-Defined Chunking(CDC),which introduces the sampling algorithm into the data chunking process,continuously calculating the hash value of the sliding window and then selecting the minimum value as a sample.The experiment shows that data deduplication ratio of the proposed algorithm is higher than EPAS algorithm at the same sampling ratio.How to efficiently delete the repeating data while ensuring the data confidentiality poses a great challenge to the data deduplication technology.This paper proposes a secure data deduplication scheme to be implemented in the integrated network.In this scheme,many key servers are distributed into the entire network so that the client can be linked to several nearby key servers at any time.Then a key generated by a key server is chosen randomly.The key fingerprint is saved to the key fingerprint server to ensure that the data with the same attributes can be encrypted by the same key for the purpose of ciphertext deduplication.The theoretical analysis shows that,compared with DupLESS and threshold blind signature,this scheme improves the security of the deduplication system,avoids a single point of failure,simplifies the signature verification,and thus reduces the computational overhead at the client side.