Sensitive data are being stored in untrusted cloud locations by many individuals, companies and institutions. Along with technical developments to protect the data against unavailability with partial replication and against sniffing with encryption, new data storage and analysis system designs become necessary. Stealth databases represent one such new design. They work on dispersed and encrypted data to preserve the confidentiality, integrity and availability of data entrusted to them. StealthDB is our prototype which demonstrates how stealth databases are intended to work. It mimics a relational column-store database, but is only superficialy relational. Instead, its strength is in being able to divide each field and run distributed algorithms following the map-reduce and map-carry-reduce paradigms to offload computation securely to its cloud backends.
(jump to top)
StealthDB is being developed inside the Dispersed Algorithms Git repository. It is a Python3 command-line application which makes use of some C libraries. The application is quite adaptive and missing dependencies are gracefully degrading the database system's functionality. The application's functionality can easily be integrated into custom applications by using the 'stealthdb.py' module. Furthermore, a name server and a cloud backend service are available for distributed operation on top of arbitrary storage and compute services. StealthDB's clone URL is git://nubisave.org/git/dispersedalgorithms. The 'db' folder contains a README file with installation instructions.
A screenshot of StealthDB in action with debugging messages (blue) enabled:
(jump to top)