On model-driven data migration
- 15:30 13th March 2009 ( Hilary Term 2009 )479
Traditionally, data modeling for Information System (IS) development has been based on entity-relationship (E/R) diagrams. More recently object models have been widely used, given object-orientation rich semantic support for classification, association and specification of constraints.
Although object-oriented approach brings several benefits to IS data modeling activity, it introduces a post-implementation maintenance challenge. Every change to the model describing the database represents a new version of the system. If the older version of the system was used to collect and persist data, an important question facing IS maintainers is how to make the data collected by the previous version of the system available in the new version.
In many cases, the answer to this question is developing data migration programs to move the data from one model version to the other as a result of the data model evolution. This development task could be a complex, time- consuming and error-prone.
This talk aims to look into this data migration problem from a model-driven perspective and attempt to outline some potential support that can be offered to maintainers confronted with this challenge. The talk is based on �work-in- progress� that has not realized concrete results yet. The focus will be on motivating the topic, outlining related literature and proposing a high-level model-driven solution.
Share this:
