Caching of large database objects in Web servers

The popularity of the World Wide Web has been increasing at an exponential rate of late. As such growth was unanticipated, the infrastructure is increasingly experiencing problems. The combination of increased network bandwidth demand and overloaded servers results in increased data retrieval latency for the end-user. Caching data at appropriate points in the Web helps alleviate this problem. Almost all previous and existing Web servers use a flat file approach to store data, with use of database management systems (DBMSs) rudimentary, if extant at all. Storing pages in a file system may result in faster retrieval, but storing them in a DBMS gives the user greater administrative control. The use of a DBMS in a Web server, and the concomitant implication of frequently changing data, complicates the caching problem an Web-based applications. The Illustra Object Relational DBMS provides a flexible and user-friendly environment for building Web applications where all the server data is stored in the DBMS. The authors develop a caching scheme for large objects in the Web DataBlade module of the Illustra ORDBMS. Implementation details and preliminary performance results are presented.