Currently, database systems do not provide effective means for archiving the database history and support queries on past snapshots and the temporal evolution of the database. Indeed the flat structure of relational tables and the limited expressive power of SQL have frustrated past attempts to extend SQL with temporal capabilities. These problems can be solved by using XML, whereby the current trend of publishing database tables in XML is extended to publishing the history of such tables. Indeed, ArchIS uses a temporally grouped data model to represent the history of relational tables in XML, achieving a more natural representation that significantly reduces the need for coalescing when column tables are eliminated. Complex temporal queries can then be expressed on these views using the standard XQuery language—whereas they require special temporal extensions in SQL.

The history of the database can be managed by a native XML DBMS or by an RDBMS. The second approach allows complete integration with current databases, but poses several technical challenges, including temporal clustering and indexing, compression of archived data, and mapping XQuery statements into equivalent SQL/XML queries. ArchIS has developed general and efficient solutions to these problems; said solutions can be used to add transaction-time support to any relational database system, as demonstrated by our ArchiDB2 and ArchiATLaS prototypes.


Two related projects focus on (i) representing the history of the database schema evolution and coping with it, and (2) archiving and querying the version history of arbitrary documents. These are the DEIS project and the ICAP project.

กก

Project Links

http://stromboli.cs.ucla.edu/archis/

กก