Efficient parallel summation on encrypted database system

Abstract

The spread of cloud computing services has resulted in data storage and data management services being outsourced to servers operated by a third party. This outsourcing can raise some security concerns because the administrators of the servers may snoop and explore the data. Encrypted database systems provide a certain level of security by encrypting the data they contain. It provides relational operators over encrypted data, which does not require data decryption and blocks the incidents. The disadvantage of this scheme is the considerable execution expense associated with data processing. Current encrypted database systems do not provide sufficient performance improvement. We focus on the summation operator in this paper, and we propose a parallel execution scheme of sum aggregation over the encrypted database. The proposed method employs task parallelism, and we assign a fine-grained data objects for each thread. We evaluate the proposed method on our prototype database system implemented in C++ and OpenMP. The result showed that proposed method achieves 15.99× performance improvement compared with a conventional method.