From: Cade Bryant
To: Editor
Date: 22 Jun 2005
Here is an excerpt (much abridged) from an industry rep that
you might find entertaining. It was
written by Matt Nunn, a senior product manager with Microsoft's SQL Server
team, for the June 2005 issue of SQL Server magazine. I have capitalized the offensive parts:
---------------------------------------------------------------------------
Let's look at another feature set that will BLUR THE
TRADITIONAL BOUNDARIES OF APPLICATION AND DATABASE DESIGN: native XML support.
XML has become the standard format for transporting data over
the Internet and has also found its way into other application-design areas
(e.g., DATA STORAGE). ... Another distinguishing feature of XML data is the
structure you STORE it in. Rather than attempting to duplicate a traditional
relational data model—in which complex relationships between data items are
defined along with the data—XML uses a hierarchical structure in which data is
arranged as a set of nodes in a tree. Each data item can have its own
associated set of data, described as sub-nodes for the item, and one XML
structure can contain many top-level data items ... This STORAGE structure is
different from a relational STORE consisting of tables, columns, and rows ...
... not having to predefine what data items will contain or
restrict data or data types before reading them has significant advantages.
When you combine XML data with relational data, you can create more flexible
application designs. For example, imagine a table that stores data about
magazine articles. Each article is associated with data items such as title,
text, author, and editor. In a purely relational model, you'd need to create an
Articles table containing fields for each piece of data. But what if some
articles have multiple authors or editors, have a subhead in addition to the
title, or are actually a series of articles? Suddenly, the relational structure
becomes more complex. You now need an Articles table containing an extra field
for subheads, an Authors table, an Editors table, and a table for article
series. Along with these extra tables, you likely need many linking tables
because editors and authors can be associated with many articles. Although such
complex relational models are common, XML might provide a better STORAGE
mechanism. You can represent all possible article data in one XML file ...
... Ideally, a STORAGE mechanism would let you combine the
flexibility of XML with the power, speed, and efficiency of the relational
STORE ... The XML data type lets you create tables that store only XML or store
both XML and relational data. This flexibility lets you make the best use of
the relational model for structured data and enhance that data with XML's
semi-structured data. In our articles example, you'd retain the relational
model for storing articles, then add an XML column to that data to store
information that's not required or could change, such as subtitles ...
... The native XML data type in SQL Server 2005 lets you
better model your REAL-WORLD DATA STRUCTURES. In life, nothing is certain; but
by combining XML with relational data, you can now account for this inherent
uncertainty, which will help your systems better react to changes and give them
a longer life. Next month, I'll look at how native Web Service support in SQL
Server 2005, combined with XML features, further CLOSES THE GAP BETWEEN THE
DATABASE AND THE APPLICATIONS THAT USE IT.
----------------------------------------------------
It's obvious that this writer doesn't understand the
difference between the logical and the physical. He keeps referring to the relational model as a method of
(physical) storage. Furthermore, as a
DBA, I have always fought for a GREATER separation between application
development and database design, yet according to this author, Microsoft (like
many other vendors) are trying mightily to undermine this separation and bring
us back to the days of application-managed data.
Ed. Comment: None
warranted.
Posted 9/3/06
