The paper presents key aspects of fragmented data processing (FDP), an overall approach to the reliable processing of confidential information. FDP aims to provide dependability with respect to the following attributes: availability, reliability and security. The author presents the problem statement, which is an antagonism between usual fault tolerance techniques allowing high reliability/availability, and intrusion prevention techniques allowing high security. Then emphasis is placed on the fine grain fragmentation, that enables him to define a global framework for a quantitative evaluation of confidentiality preservation by entropy calculation. Basic operators are evaluated and comply with the a priori intuition: the finer the fragmentation, the better the confidentiality obtained.
展开▼