Efficient Cross User Client Side Data Deduplication in Hadoop

Abstract

Hadoop is widely used for applications like Aadhaar card, Healthcare, Media, Ad Platform, Fraud Detection & Crime, and Education etc. However, it does not provide efficient and optimized data storage solution. One interesting thing we found that when user uploads the same file twice with same file name it doesn't allow saving the same file. But when user uploads the same file content with different file name Hadoop allows uploading that file. In general same files are uploaded by many users (cross user) with different name with same contents so this leads to wastage of storage space. So we provided the solution of above problem and provide Data Deduplication in Hadoop. Before uploading data to HDFS we calculate Hash Value of File and stored that Hash Value in Database for later use. Now same or other user wants to upload the same content file but with same content, our DeDup module will calculate Hash value and verify it to HBase. Now if Hash Value is matched so it will give message that "File is already exits". Experimental analysis demonstrates (i.e. Text, Audio, Video, Zip files etc.) that proposed solution gives more optimized storage acquiring very small computation overhead and having optimized storage space.

‏An Enhanced Approach to Improve the Security and Performance for Deduplication

“…This is a waste of storage space and reduced processing efficiency on devices. This paper [47] proposes DeDup approach to eliminating duplication data by calculating the hash value at the file level before uploading it to HDFS and comparing it with existing files. When download the file by any user, check to Hbase through the hash account to see if the files download the same content or not.…”