Separation of Powers in the Cloud: Where Applications and Users Become Peers

We challenge the widely accepted practice that web appli-cations must be trusted with user data. We present an al-ternative model based on logic programming, where users and applications are equal peers in a shared cloud environ-ment. User data is represented as a set of facts. The ap-plication is represented as a set of rules defining how user data is to be processed, but is not given direct access to the data. This way, end users remain the owners of their own data, and are able to determine who can see it and who can modify it. For concreteness, we define a data representa-tion and query language, named CloudLog, for a new family of deductive databases, named NoDatalog. We add access control to the language for guaranteeing that the rules pro-vided by the application cannot change the choices made by users. We demonstrate how business logic can be expressed in CloudLog, and discuss how an efficient CloudLog-based database can be implemented.