Temporality in Object Database Management Systems

  • Author(s) / Creator(s)
  • Technical report TR98-04. Conventional databases represent the state of an enterprise at one particular point in time. That is, they contain only current data. As a database changes, out-of-date information, representing past states of the enterprise, is discarded. However, temporal support is a requirement posed by many database applications, such as office information systems, engineering databases, and multimedia systems. Most of the research on modeling time has concentrated on the definition of a particular temporal model and its incorporation into a (relational or object-oriented) database management system. This research has led to the definition and design of a multitude of temporal models. Many of these assume a specific set of temporal features, and therefore do not incorporate sufficient functionality or extensibility to meet the varying temporal requirements of today's applications. Instead, similar functionality is re-engineered every time a temporal model is created for a new application. This suggests a need for combining the diverse features of time under a single infrastructure that allows design reuse. In this thesis, an object-oriented framework that provides such a unified infrastructure is presented. An object-oriented approach allows one to capture the complex semantics of time by representing it as a basic entity. Furthermore, the typing and inheritance mechanisms of object-oriented systems directly enable the various notions of time to be reflected in a single framework. The framework can then be tailored to accommodate the temporal needs of different applications, and existing temporal models can be derived by making a series of design decisions through subclass specialization. The framework can also be used to derive a series of more general temporal models that meet the needs of a growing number of emerging applications. Furthermore, the framework can be used to compare and analyze different temporal object models with respect to the design dimensions. This helps identify the strengths and weaknesses of the different models according to the temporal features they support. The framework is then used to instantiate a single temporal object model which has multiple facets of time. There have been many temporal object model proposals. These models differ in the functionality that they offer, however as in relational systems, they assume a set of fixed notions of time. The temporal object model developed in this thesis consists of an extensible set of primitive time types with a rich set of behaviors to consistently and uniformly model the diverse features of time. The model is one possible implemention of the temporal framework and provides concrete and consistent semantics for the different temporal features which is necessary for their coexistence. The establishment of a temporal object model provides a foundation from which investigations are carried out on using temporality to model other database features such as schema evolution. The issues of schema evolution and temporal object models have traditionally been considered to be orthogonal and handled independently. This is unrealistic because to properly model applications that need incremental design and experimentation (such as CAD, software design process), the evolutionary histories of the schema objects should be traceable. In this thesis, a method for managing schema changes by exploiting the functionality of the temporal object model is proposed. The result is a uniform treatment of schema evolution and temporal support for many object database management systems applications that require both. | TRID-ID TR98-04

  • Date created
  • Subjects / Keywords
  • Type of Item
  • DOI
  • License
    Attribution 3.0 International