[Paper Review] An Integrated Approach for Extraction of Objects from XML and Transformation to Heterogeneous Object Oriented Databases

Uzair Ahmad, Mohammad Waseem Hassan+3 more|arXiv (Cornell University)|Feb 2, 2004

Advanced Computational Techniques and Applications1 references3 citations

TL;DR

This paper proposes TransODB, a database-agnostic framework that uses W3C XML Schema to efficiently transform object-oriented database data into compact, standardized XML for seamless migration between heterogeneous object databases. By leveraging XML Schema instead of DTDs and generating optimized, schema-driven XML, TransODB reduces file size by up to 50% and improves performance, enabling efficient, scalable, and portable data exchange across platforms like Objectivity/DB and Oracle.

ABSTRACT

CERN's (European Organization for Nuclear Research) WISDOM project uses XML for the replication of data between different data repositories in a heterogeneous operating system environment. For exchanging data from Web-resident databases, the data needs to be transformed into XML and back to the database format. Many different approaches are employed to do this transformation. This paper addresses issues that make this job more efficient and robust than existing approaches. It incorporates the World Wide Web Consortium (W3C) XML Schema specification in the database-XML relationship. Incorporation of the XML Schema exhibits significant improvements in XML content usage and reduces the limitations of DTD-based database XML services. Secondly the paper explores the possibility of database independent transformation of data between XML and different databases. It proposes a standard XML format that every serialized object should follow. This makes it possible to use objects of heterogeneous database seamlessly using XML.

Motivation & Objective

• To address the lack of standardization in XML-based data transformation between heterogeneous object-oriented databases.
• To overcome limitations of DTD-based XML services, which are less expressive and scalable than modern XML Schema.
• To enable database-independent data exchange by defining a standardized XML format for serialized objects.
• To improve performance and reduce resource consumption during data serialization and deserialization.
• To support seamless data migration between diverse object databases such as Objectivity/DB and Oracle.

Proposed method

• Uses W3C XML Schema as the foundation for defining database schemas and object structures, replacing less expressive DTDs.
• Employs a SAX 2.0 parser to process XML Schema documents and generate corresponding Java classes via event-driven parsing.
• Implements a DB Schema Builder module that converts XML Schema into Java classes following Sun's coding standards.
• Utilizes an Object Builder module to reconstruct complex object networks from XML, preserving semantics and relationships.
• Employs a DB-Handler module that uses native database APIs to store reconstructed objects into target object databases.
• Designs a standardized XML format for serialized objects that is independent of any specific database system.

Experimental results

Research questions

• RQ1 How can XML Schema be leveraged to improve the expressiveness and efficiency of database-to-XML transformation?
• RQ2 To what extent can a database-agnostic XML format reduce file size and improve performance in data migration?
• RQ3 Can a standardized XML representation enable seamless data exchange between heterogeneous object databases?
• RQ4 How does the proposed approach compare to existing solutions like WISDOM and Objectivity/DB’s ooXMLload in terms of memory and time efficiency?
• RQ5 What impact does schema-based XML generation have on scalability and resource usage during data transformation?

Key findings

• TransODB reduces XML file size by up to 50% compared to WISDOM, significantly decreasing data volume for transfer.
• The prototype shows a 30-40% improvement in processing time over the Objectivity/DB ooXMLload tool for equivalent data sizes.
• Memory usage during object reconstruction remains low, allowing concurrent system operations without performance degradation.
• The use of XML Schema instead of DTDs enables more precise and compact representation of database schemas and object structures.
• The framework successfully enables data migration between heterogeneous databases such as Objectivity/DB and Oracle, demonstrating database independence.
• The approach achieves better scalability due to reduced data volume and optimized parsing and reconstruction processes.