Why XML Over the Relational Model?

Simon St.Laurent simonstl at simonstl.com
Tue Jan 5 04:07:46 GMT 1999

At 10:50 PM 1/4/99 -0500, Paul Butkiewicz wrote:
>But using XML for persistence in this case presents its own perils.
>Consider the case of modifying the database which is now implemented as an
>XML document.  If I add an attribute, a node, or make some text longer, I
>may very well need to re-write the *entire* file to accommodate the changes
>I've just made and make them persistent.  Of course, I could pad various
>parts of the document with spaces to get around this, but given that XML has
>no notion of field lengths this certainly isn't a sure-fire fix --- I'm
>still going to need to rewrite the whole document now and then, especially
>if I'm adding nodes ad hoc.  And in addition to the overhead of actually
>writing out the whole document (which *must* be done as part of the
>transaction), when writing that document I would need to lock the *entire*
>document.  This wouldn't be acceptable at all in a situation where many
>concurrent writes are going to data in the same document, even if on
>different nodes.
>And then there's the obvious problem that if I want to go directly to
>Article 14, Section 3, Subsection 41, Chapter 27, Paragraph 6, and I don't
>have that part of the document cached in memory, I may have to iterate
>through the entire document get that item.
>Maybe this part of the discussion should be XML vs RDBMS vs ODBMS.  :)

I think we're in agreement here, mostly because I'd like to see XML
documents stored in ODBMS-like systems rather than in the traditional
filesystem goop we have today.  When I talk about using XML in this way, I
really see XML as a much friendlier face to an ODBMS rather than a big
chunk of characters in sequence.  On the other hand, I'd rather not have to
address the ODBMS directly, ever - everything should be addressable through
XML-based mechanisms.  The problem with the view of XML presented above, in
my mind, isn't XML - it's the crummy file systems we have to work with at

This wouldn't be a great solution for data that really fits well into an
RDBMS.  The 'natural' fit of the data structures is still important.

See my (old) essay Building the File System into the File
(http://www.simonstl.com/articles/filesyst.htm) for more on this.

Simon St.Laurent
XML: A Primer / Cookies
Sharing Bandwidth 
Building XML Applications (February)

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)

More information about the Xml-dev mailing list