In editie 24.2 van I/O Vivat, het magazine van studievereniging Inter-Actief, zijn het mijn antwoorden die in de vaste rubriek “drie vragen aan” prijken.
Archive for ◊ December, 2008 ◊
The project description for Joost Diepenmaat’s Msc project has been finalized. The project is being supervised by me and Jan Flokstra.
Object XML Mapping
Database Management Systems (DBMSs) play an important role in application development. Almost every information system stores its information in a DBMS. Nowadays, the application domain requires, more and more, specific and detailed storage structures for complex objects. This application domain ranges from document standards we use every day (e.g. OpenXML for word processing) to specific business process documents (e.g. the Universal Business Language for purchase, orders, invoices, etc.). Due to these complex, but useful standards, there is a stronger need for agile adoption techniques within the development frameworks.
It would be easier to access XML data using mechanisms that fit in the application development language better. Object Relational Mappings/Domain Models currently support this for the relational SQL world. As they manage interconnected objects, where each object represents something meaningful within the application domain. A Domain Model in an application contains a whole layer of objects that model the business you’re working in. These objects mimic the data in the business and/or capture the business rules.
The thesis studies the possibilities of an Object XML Mapping with XML/XPath as a base language for persistent storage. We explore and describe the fundamentals of Object XML Mappings. Various database back-ends (with or without build in XML support) will be studied. A ruby-based prototype implementation with support for build in XML database (XPath Accelerator) or a native databases will show the capabilities of the Object XML Mapping.
Today I accepted an invitation to serve on the program comittee of the International Conference on Data Engineering (ICDE 2010).
MSc student Luuk Peters received “green light” for his concept thesis “Battle of the bulk – Corporate XMLDB vs. Research XMLDB”. The defence is expected to take place in January 2009. This MSc research was conducted at FINAN, a company that specializes on software for financial analysis for risk management. Luuk performed extensive experiments with Oracle trying to find out how to apply its XML-support for scalable querying of XBRL documents (XBRL is an XML standard for financial reporting). Moreover, Luuk compared performance and scalability of Oracle with MonetDB/XQuery. MonetDB/XQuery clearly beats Oracle on raw performance, but when the collection of documents becomes larger than 80,000 documents (1.7GB), MonetDB/XQuery starts to give stability problems while Oracle continues to (albeit slowly) produce answers.
Today I gave a demonstration of the “EPrints Clickable Views” prototype during the CWI meeting (Committee on Scientific Information). EPrints is the faculty’s publication management system. The prototype allows end-users to define views on the publication database and see the results of their changes in real-time and WYSIWYG. The prototype is based on the MonetDB/XQuery XML database and demonstrates its power and scalability for applications like these. The committee decided that it is definitely worthwhile to pursue this effort further. A project will be started to develop a production version in 2009.