Organization responsible: 4TEC

People involved:

• Paul Dechering, 4TEC

Project description:

Cascading Referential Integrity is a concept well known in server based database architectures. The research undertaken was to find approaches and constraints that would support distributed, peer to peer databases and yet retain real-time characteristics. The approach has been to first have a proof of principle for promising solutions in a stand-alone situation and secondly convert the most promising approach to a distributed environment. The stand-alone approach realizes the constraints on a stand-alone database that is not shared across the network. The distributed phase developed principles to manage these constraints in shared databases.

Without primary key updates (and secondary key updates) the ON UPDATE constraint is not interesting and has trivial semantics. For a sound set of constraints it was necessary to also realize cascading primary key updates. Since data sets are typically identified by their primary keys, this needs a thorough design in general and in distributed environments in particular. In latter case there is the problem of having related tables distributed over the network, while the consistency enforced by constraints between these tables remains in effect.

Support of cascading referential integrity in a distributed real-time architecture is quite unique and required a substantial effort. To the best of our knowledge, having such a capability in a pear to peer commercial database product would be unique, once implemented.

Publications: (none)