Stealth Databases: Ensuring User-Controlled Queries in Untrusted Cloud Environments

Sensitive data is increasingly being hosted online in ubiquitous cloud storage services. Recent advances in multi-cloud service integration through provider multiplexing and data dispersion have alleviated most of the associated risks for hosting files which are retrieved by users for further processing. However, for structured data managed in databases, many issues remain, including the need to perform operations directly on the remote data to avoid costly transfers. In this paper, we motivate the need for distributed stealth databases which combine properties from structure-preserving dispersed file storage for capacity-saving increased availability with emerging work on structure-preserving encryption for on-demand increased confidentiality with controllable performance degradation. We contribute an analysis of operators executing in map-reduce or map-carry-reduce phases and derive performance statistics. Our prototype, StealthDB, demonstrates that for typical amounts of personal structured data, stealth databases are a convincing concept for taming untrusted and unsafe cloud environments.