SCORD: Shuffling Column-Oriented Relational Database to Enhance Security

Column-oriented database systems have drawn a lot of attention in recent years because of their performance advantages in terms of data querying and computation on columns. Since databases often contain sensitive and valuable information, protecting the confidentiality, integrity, and availability of the database is of utmost importance. In addition to encryption, shuffling can also be used to secure the database storage files. In this paper, we present a novel approach called SCORD to prevent the leakage of confidential data stored in the database management system by applying the shuffling technique on column-oriented databases. SCORD can protect column-oriented databases in both offline storage mode and online running mode. SCORD also bundles semantically or statistically associated attributes, such that the data is shuffled in a way that appears deceptively authentic and indistinguishable to potential attacks. We implement a prototype of SCORD and evaluate it using a set of real-world data. The experiment results demonstrate the effectiveness of our approach and show that the processing overhead is acceptable.