Schema Evolution

Background

In this section, we illustrate with a simple example the functioning of schema evolution, contrasting it with the lowest level of schema change support that can be embedded in a database, that is the modality of schema modification (Jensen et al., 1998). Assume that we have a relational database that contains only an AUTHOR relation with the attributes ID (primary key), NAME, PHONE, and COUNTRY. The first state of this database is as shown in Figure 1.

Figure 1.

­

The catalogues store information on the schema S1 of the AUTHOR relation. The table AUTHOR contains two tuples for two authors. Then consider the following schema changes:

ALTER TABLE AUTHOR
DROP COLUMN PHONE;
ALTER TABLE AUTHOR
ADD COLUMN EMAIL CHAR(30);

The schema modification technique allows users to effect changes to the database schema, but neither previous schema nor its underlying data are preserved: the old schema is replaced by the new schema, which is initially empty as data populating the old schema are discarded. The effects in our example would be as shown in Figure 2.

Figure 2.

­