[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: XML storage

• To: "'Helena'" <helenaforsberg@h...>,<xml-dev@l...>
• Subject: RE: XML storage
• From: "Michael Kay" <michael.h.kay@n...>
• Date: Fri, 29 Aug 2003 15:49:10 +0100
• Importance: Normal
• In-reply-to: <1062145623.472574c0helenaforsberg@h...>
• Reply-to: <michael.h.kay@n...>

> 
> I want to store a large number (somewhere around 50000) of 
> data-centric XML documents. I'm a total newbie at this, but a 
> little research has revealed that there is three main ways to 
> go, store them as BLOBs in a RDBMS, map them up into 
> relations, or use a native database. The documents will all 
> conform to the same DTD, and will vary in size between 9 and 
> 50 KB. The DTD contains many optional elements, no mixed 
> contents, no processing instructions, CDATA or comments. The 
> structure is not really important to preserve, and there is 
> no need for full-text querying. However, I'm afraid the 
> relational approach will result in too many relations for 
> efficient querying. The DTD contains 225 elements. 
> 

This sounds like a very good fit to the kind of things which a Native
XML database like Tamino is designed to handle.

The problems with the RDBMS approach is that typically you either have
to shred the document into tiny pieces, which gives you an
incomprehensibly complex database and a lot of cost on storage and
retrieval, or you store a raw CLOB, in which case you are getting very
little value out of the ability of a database to store structured data.
There are cases where storing XML in relational databases makes sense
(like when you've got one handy and don't want to get another) but this
looks to me like a Native XML scenario. I'm biased, of course.

Michael Kay
Software AG

• Follow-Ups:
  • Mixed Content
    • From: Alejandro Lopez <lopgaral@y...>

• References:
  • XML storage
    • From: "Helena" <helenaforsberg@h...>

• Prev by Date: Re: XSLT and default namespaces
• Next by Date: RE: Question about UTF-8
• Previous by thread: Re: XML storage
• Next by thread: Mixed Content
• Index(es):
  • Date
  • Thread