Back in April, we ran a wildly ambitious story by Sean McCown. Entitled Databases Flex their XML, it compared the XML features of DB2, SQL Server, Oracle, and Sybase -- and also made an excursion into Yukon territory. (My contribution was the speculative sidebar on the future of native XML database technology.) Yesterday Microsoft's Michael Rys, a database architect and a co-author of XQuery from the Experts, blogged a lengthy and thoughtful response to Sean's analysis.
To frame his response, Michael develops a taxonomy of XML structures and storage models and says:
It should be clear, that by making this distinction, the terms "shredding," "unstructured," and "structured" are confusing. XML's structure can be highly structured, semi-structured or markup-structured, but it is always structured. And either of these formats can be stored in a way to provide relational, InfoSet or textual fidelity using either relational or blob storage. [Michael Rys]
That's the kind of useful clarification that Michael has been consistently delivering on his blog. I hope this thread will continue. Sean's article was -- as Michael acknowledges -- as good a comparative piece as has ever appeared in the press. But the topic is huge, and will fuel ongoing discussion. We're living through an epochal moment in the history of the industry. The hybridization of SQL and XML will deeply transform the philosophy and practice of data management in ways that I think none of us fully understands. The story will emerge from conversations between practitioner/analysts like Sean, and architects like Michael. Happily, the online realm has become a pretty good place to have those conversations.
Former URL: http://weblog.infoworld.com/udell/2004/06/11.html#a1021