Accelerating the Cloud Backup Using GPU Based Data Deduplication

Data deduplication is an efficient technique for data reduction. It has been adopted and widely used in cloud backup service. However deduplication requires high computing capability since it has to calculate the fingerprint of data many times using a hash algorithm. This can be a performance bottleneck for the system. In this paper, we propose to use Graphics Processing Unit (GPU), a high computing architecture, to accelerate the fingerprint generating process in deduplication. Our experiment shows that GPU can significantly improve the performance of deduplication especially when running on large data, where we achieve an acceleration ratio up to 53 times greater compared to with CPU.