An empirical study of XML data management in business information systems

Due the popularity of XML, an increasingly large amount of business transactions encoded in XML have been exchanged on-line. Currently there are two approaches to process and manage these XML data. One is to store them in relational databases, and the other is to store them in recently-developed native XML databases. There is no conclusion yet as to which approach suits better for contemporary business information systems. Also, the effectiveness of native XML databases used in daily operational systems has not been completely investigated. Therefore, in this paper, we provide: (1) a complete and systematic survey of the current development and challenges of processing XML data in relational and native XML databases, (2) a useful benchmark for IT practitioners who need to process XML data effectively, (3) experimental results and detailed analysis which reveal several interesting tips that can be helpful to XML document designers, and (4) a conclusion, based on the findings of using native XML databases in EDI processes, that it is practical to use native XML databases for daily operations although our experimental results showed that relational database systems outperform native XML databases in processing XML data.