Improving Primary Storage Performance Using Data Deduplication and Pre-fetching Methods in Cloud

The Cloud Storage Server (CSS) provide huge storage facilities to many data users. The clients outsource large amount of data to the CSS. One of the challenging issues to CSS is data redundancy. Many researchers have proposed advanced techniques to eliminate data redundancy, in which data deduplication is one of the techniques to eliminate duplicate data in the cloud. To improve the primary memory management, the authors have introduced the Performance-Oriented I/O Deduplication (POD) in the existing scheme. The existing scheme has read amplification problem and consumes more network bandwidth while downloading a file. To overcome these issues, in this chapter, a data Pre-fetching scheme is proposed with primary level data deduplication. The proposed scheme uses the linear regression technique and data Pre-fetching together to analyze the I/O disk access with client piggybacking information. The proposed scheme is useful to clients with less computing resources and limited power. The objective of the proposed work is to reduce the client data download time and reduce the data access process. The performance analysis shows that the proposed scheme improves storage server access efficiency by resolving read amplification problem and minimizes network latency, network traffic and energy consumption as compared to the existing scheme.