Temporal Relational Databases

Abstract

The evolution of a relational database over time is not captured by the standard relational data model we have presented in the previous chapters. For example, the inventory of items in a warehouse changes over time as items are shifted from and to the warehouse, and the details of employees that work in a company change over time as the database is updated with new employees joining the company and old employees leaving the company. Many other scientific, financial and business applications have a substantial temporal element associated with them including applications that involve time series analysis. Although time can be modelled within the standard relational model, this cannot be done in a straightforward and unified manner, since there is no inherent support for temporal data. Thus due to the importance of recording and manipulating temporal information, there is a need for a cohesive and consistent extension of the standard relational model to handle such temporal data. The research into temporal databases has been an active subarea of relational database theory for well over a decade now. In order to merge and encompass the main proposals for temporal relational database query languages, there has been a recent comprehensive specification of a temporal extension of SQL, termed TSQL2. In this chapter we formalise a temporal extension of the relational model, which provides a basis for understanding the way in which time can be seamlessly incorporated into the data structures, the algebra and the fundamental integrity constraints of relational databases.