Secure Deletion: Towards Tailor-Made Privacy in Database Systems

Abstract

In order to ensure a secure data life cycle, it is necessary to delete sensitive data in a forensic secure way. Current state of the art in common database systems is not to provide secure deletion at all. There exist academic demonstrators that address some aspects of secure deletion. However, they are limited to their deletion approach. We argue, due to different data sensitivity levels (probably even on attribute level) and differences in policies (e.g., time when and how a data item has to be deleted), it is necessary to have a standardized, user defined opportunity to enforce secure data deletion in a forensic secure manner. Our literature analysis reveals that most approaches are based on overwriting the data. Thus, in this paper, we examine how it is possible to integrate user-defined overwriting procedures to allow a customizable deletion process based on existing default interfaces to minimize the integration overhead. In general, we propose an extension of SQL and a page propagation strategy allowing the integration of a user defined deletion procedure.