Benchmarking Framework for Transparent Data Encryption Systems

Abstract: In the digital world of today, information is always at risk regardless of its state, at rest or in transit. Cryptography is the technology that promises to address the security issues that emerge. Hence, it was a reasonable consequence to introduce cryptography to databases. However, manually encrypting and decrypting data along with the key management is a burden for the regular user of a database. The need for removing this burden gave birth to Transparent Data Encryption (TDE).   TDE technology is widely available nowadays and a number of vendors have developed their own solutions for protecting data at rest in a transparent way to the end user. However, cryptographic operations are resource intensive and introduce an overhead to the computational operations. The burden of cryptographic operations has drawn the interest of both academia and the industry for a long time before TDE appeared on the horizon. Hence, a lot of research has been done to measure the performance impact of those operations.   Despite the extensive study for the performance of cryptographic algorithms, the performance of the TDE systems and the add-on computational burden for the introduced encryption has not yet been studied thoroughly. As a result, the current Thesis project tries to develop a theoretical benchmarking framework that evaluates the performance of Transparent Data Encryption systems. The study is conducted utilizing the Design Research methodology.   The developed benchmarking framework focuses on the basic performance metrics of TDE systems, Elapsed time, CPU time and Hard Disk memory consumption. These metrics are calculated for varying key lengths, encryption algorithms and table sizes.  The framework follows a five - step procedure that includes the creation of topology - lab environment, creation of databases and definition of scenarios, activation of TDE feature, sequencial execution of scenarios and analysis of the results. The developed framework is evaluated by applying it on real TDE systems.