RE: XML versus Relational Database

On Thu, 1 Feb 2001, Caroline Clewlow wrote:

> Isn't the issue that XML is not designed as a mechansim 
> to store data, but a mechanism to allow data to be provided 
> in a common format that can be understood by disparate systems ?
>
> ( I know this is a criminally simplistic view of what XML is 
> about but I just wanted to make the point  :-)

There are obvious benefits to storing data in a format that can be
understood by disparate systems, and which is powerful enough to represent
data from disparate systems. If the XML view of the world is the one that
everyone needs when they unify data, and all your programs are moving things
into and out of XML, why not just store it in XML?

There are a few reasons.

1. Query Language and Transformations

SQL has always maintained that no one view of the data is the "real" view,
and that users will often need to restructure the data. In other words,
transformation is often a crucial part of a query. Note that even relational
database vendors are embedding XSLT in their systems.

The W3C XML Query Language allows many kinds of transformations and joins.
(There is not yet a public draft of the language, but see
http://www.almaden.ibm.com/cs/people/chamberlin/quilt.html for information
on an earlier language, Quilt, which is similar to the XQuery language).


2. Performance - depending on what you are doing

For some things, a native XML database can offer superior performance. For
others, especially where many joins are involved, relational databases offer
superior performance. The performance you get depends a lot on the product
you use, how it is configured, etc. For many applications, XML databases can
offer very good performance, and improving performance for other kinds of
applications is clearly something that native XML database vendors are
working on.

I do not claim that native XML databases will replace relational databases
anytime soon. On the other hand, I do believe that a native XML database,
combined with XQuery, will offer a very convenient general purpose database
with a great deal of expressive power.

Jonathan