An object-oriented data model for partition processing in distributed databases

Data replication in a distributed database system can lead to inconsistencies when communication breaks down due to partition failures. Correctness can be ensured if the data model used for representing and manipulating data properly captures real-world semantics. An object-oriented data model which attempts to bridge the gap between real-world semantics and its representation is proposed. Essentially, data is stored in an attributed graph with the nodes representing real-world objects and the arcs representing the relationships between them. A superimposed rule system is employed in the model to enforce consistency and semantic constraints. The model easily incorporates partition processing strategies like Data-patching, Quorum Consensus and Failure mode integrity constraints.