In the model-driven approach to software development, system implementations are generated automatically from abstract models of structure and behaviour. This could greatly facilitate systems evolution: a new version of a system may be produced simply by updating the system model and repeating the generation process. However, an information system may hold data of considerable value and complexity, and this must be safely migrated at each evolutionary step. This paper shows how this problem can be solved through a formal, model-driven approach: platform-specific data migration functions are generated automatically from a formal model of system changes, and the applicability of these functions is calculated in advance, ensuring that they may be safely applied to existing data. © 2011 Springer-Verlag.

Original publication

DOI

10.1007/978-3-642-21759-3_28

Type

Journal article

Journal

Lecture Notes in Business Information Processing

Publication Date

01/01/2011

Volume

81 LNBIP

Pages

383 - 397