[xquery-talk] XQuery and SQL opinion

Michael Rys mrys at microsoft.com
Thu Feb 12 13:06:10 PST 2004


I have not yet read Jim's piece, but knowing him, I am sure it will be a
delightful read.

Let me however comment on your statements below.

> -----Original Message-----
> From: talk-bounces at xquery.com [mailto:talk-bounces at xquery.com] On
Behalf
> Of Ronald Bourret
> Sent: Thursday, February 12, 2004 10:02 AM
> To: talk at xquery.com
> Subject: Re: [xquery-talk] XQuery and SQL opinion
> 
> I pretty much agree with what Jim says. A couple of comments:
> 
> 1) I think Jim is right that big RDBMS companies will dominate the
> XQuery field. There are several reasons for this. First, those
companies
> have a lot of money and expertise to throw at both implementation and
> marketing. Second, they have large, well-heeled customer bases who
would
> be better served by staying with an existing product than moving to a
> native XML database. And third, they are in a better position to
> integrate relational data and XML than the native XML databases, both
> because of the fact that they own the SQL engines and because their
> customers own the data.

[Michael Rys] I think this is an important observation and what
ultimately will make so-called native XML databases end up next to
OODBMSes.

> (This prediction assumes that the XML data type in SQL/XML will
> eventually be implemented by native XML technology -- it's currently
> implemented in both IBM and Oracle as a CLOB. I can't think of any
> reasons that the big RDBMS companies won't eventually do this --
> performance will be horrible otherwise.)

[Michael Rys] SQL Server Yukon is storing its XML Datatype in a non-CLOB
format (more information I currently cannot give). However, even that
format probably will evolve further down the road and I see some
interesting combinations of relational and XML data items in the future.

> 2) I think that the statement that new XQuery applications will
overtake
> new SQL applications by 2010 or 2012 is credible, but perhaps a bit
> misleading. The implication is somehow that SQL is going away. Given
the
> durability of, say, COBOL, I doubt this will happen. So while this
might
> be true for *new* applications, I don't think it will be true for
total
> applications. That is, SQL is a good fit for most of what it is used
for
> today and those applications will be around for a long time to come.

[Michael Rys] I am a bit more skeptical about this prediction, since it
heavily depends on the type of data you will mainly query at that time
and how efficient XQuery can be implemented on data that is relational
in its base structure.

 > (An interesting question is what conditions would make the total
number
> of XQuery applications overtake the total number of SQL applications.
It
> implies either that people do everything through XML or that the set
of
> data we are using with databases changes considerably to favor
> semi-structured and unstructured data over relationally structured
data.
> I'm hard-pressed to believe the first scenario for performance reasons
> -- I think SQL will always outperform XQuery -- but the second
scenario
> would posit a very interesting future indeed.)

[Michael Rys] Exactly.
 
> 3) I find it curious that Jim seems to lament the cessation of
> significant growth in SQL. Does software that stops growing actually
> die? It certainly loses its sex appeal -- people do tend to flock to
> what is new -- but big, sober customers like banks tend to stick with
> what works. Look at the continued use of hierarchical databases,
ASCII,
> and COBOL, for example.

[Michael Rys] Correct. However, if your company makes most of its
revenue on SQL and growth is as important to economic success today as
it is, then obviously a cessation of growth is bad :-).

> -- Ron
> 
> Jason Hunter wrote:
> >
> > http://otn.oracle.com/pub/columns/melton_sql.html
> >
> > An opinion piece by Jim Melton.  For the XQuery-related content,
jump
> > ahead to the "SQL and XML: Glimpse into the Future" section toward
the
> end.
> _______________________________________________
> talk at xquery.com
> http://xquery.com/mailman/listinfo/talk



More information about the talk mailing list