In this chapter, an Automatic Schema Evolution Framework is explored and developed to more effectively manage schema evolution in a FSDBS. This framework provides a Schema Element Dependency Meta-Model, a set of Schema Change Templates, and incorporates view generation, view rewriting, and query rewriting as solutions. These developed methods ensure applications can accommodate schema changes and hence remain valid.
TopIntroduction
The traditional database can be described by schemas that define relations, attributes, etc. A spatial database schema is an extension of the traditional schema to include spatial descriptions and behaviours required for a spatial database (Yeung & Hall, 2007). Spatial database schemas, like traditional schemas, are subject to change or evolution due to changes in perception of reality and application requirements. Changes can also occur as a consequence of integration with other systems, compliance to new regulations and the implementation of new security requirements.
The task of managing schema evolution is to ensure schemas, data and application consistency and integrity after change. Schema changes often result in the applications built against the schemas being invalid.
In a Federated Spatial Database System (FSDBS), in which spatial means geographically spatial, the spatial data are shared by multiple organizations and applications across a number of databases. A local database schema change will affect not only the applications locally but also applications built on the federated schemas. Effective management is problematic and there are several challenges to managing schema evolution in a FSDBS including:
- •
How to evaluate the impact of schema changes on local and global databases;
- •
How to propagate the schema changes across databases in a federated environment; and,
- •
How to automatically rewrite run time queries, and rewrite and generate views (spatial views and traditional views) so that the applications are immune to schema changes.
The objective is to explore and develop an Automatic Schema Evolution (ASE) framework for managing schema evolution in a FSDBS. The main contribution of this chapter is the proposal of semi-automatic methods that, in future work, will lead to the development of fully automatic methods.
Of particular interest is the Shared Land Information Platform (SLIP) built by Landgate (authoritative agency for land information in Western Australia) as a Spatial Data Infrastructure (SDI) enabler (2011). SLIP went into production in 2007 and consists of more than 350 datasets from 20 different agencies. As such it is a federated database and queries are built over the different datasets. Schema updating is needed because the datasets are agency owned and controlled allowing local changes to occur.
For schema evolution, the mismatch between evolved schemas and queries need to be overcome. Queries here include queries generated by applications and stored queries (views and spatial views) as seen by users. The ASE framework defines the components and methods for automatic run time query and view rewriting, as well as new view generation when schemas change.
TopBackground
Before explaining the proposed ASE framework, it is necessary to define various aspects required for the ASE Framework, explore the various issues and concepts involved in the framework, and critique the current state of the art in research into federated and spatial databases.