A scalable relational database model for cloud computing.

Abstract

Relational databases introduce transitive dependencies between the various tables from the perspective of a particular table as a result of database normalization and these dependencies prevent one from achieving parallel dynamic on demand horizontal scaling of data in hot spots of the database using database sharding. Cloud databases address this problem by modeling databases as non relational and hence allow for it to support dynamic scaling in a parallel manner, this research was undertaken to show how we can use a hybrid relational and non relational database in cloud computing with each model supporting a subset of transactions where by reads are executed off the horizontally scalable non relational model and writes on the relational model. This research shows how the Binary First Search algorithm could be used on a directed graph representation of a relational database model to derive a horizontally scalable non relational database model which can be used by cloud applications that require data storage, the database will still retain the relational structure when executing writes so as to ensure that the data stored conforms to data integrity rules and is hence reliable while the non relational database will support reads resulting in a hybrid database. The Binary First Search algorithm was chosen since it has been proven to visit all nodes in this case all tables provided it is reachable from the root node and terminate once it has visited all the nodes in a logically correct and complete manner hence ensuring that all reads would reflect the correct state of the relational writes.