Avoiding schema change & a framework for non-disruptive schema evolution

Abstract

It is a well-known fact that a significant share of the cost of information systems is spent on maintenance. Some researchers estimate this share to be around 80% of the lifetime cost of information systems. A significant share of adaptive maintenance work of a business information system starts from a change in the data models it is based on. Since, almost all business information systems are based on relational databases, it is clear that relational database schema changes account for a significant share of the total maintenance cost. This is aggravated by the fact that a change in database schema has the potential of driving existing applications out of date. Modifications to such applications cost a considerable of time and money to business organisations. Possible solutions to minimize the costs associated with this problem are essentially technical in nature. This dissertation explores a couple of possible solutions in relational database technology, which could help organisations to keep their system maintenance cost under check. The main solutions explored are in the area of data model patterns and schema evolution. A new branch of schema evolution called “Non Disruptive Schema Evolution” is explored as well.