RE: Death of the database

  • From: "Delmolino, Dominic" <ddelmolino@xxxxxxxxxxxxxxxxxxxx>
  • To: "Oracle-L Freelists" <oracle-l@xxxxxxxxxxxxx>
  • Date: Mon, 31 Oct 2005 12:52:52 -0500

I've seen my work shift out (initial work decline followed by increase
later) as a result of "unstructured data".

 

The first one was where the app developers said they didn't need DBA
help because they were going to store everything as Java Serializable
Objects (JSOs) - so every table was a tuple like (PRODUCT_ID number,
PRODUCT_PROPERTIES blob).  Worked ok for about 3 months (where I had
less work).  Then they came back to my team when queries were slow, and
no one could browse the data, and no one could write reports on it...
And of course we ended up structuring the data back into regular tables
(which ended up with more work for me).

 

The second one was where we received a custom developed invoice
application that made heavy use of XML CLOBs - so every table was a
tuple like (INVOICE_ID number, INVOICE_CONTENTS clob).  Worked ok for 2
months (where I had less work).  Then my management threw out the
product because the text-based indexes were slow, kept needing
re-builds, no one could browse the data, and no one could write reports
on it...

 

Odd that anyone would advocate in "unstructured" XML - good XML has a
DTD/Schema validation attached to it so that it's structured enough to
be useful.  When our developers wanted to store XML in our database I
darn well insisted on at least using XMLTYPE - we caught several errors
when programs tried to store XML documents which didn't even pass simple
validation.

 

The example of the medical data chip in the story is ridiculous - "I'm
sorry your hand with your chip in it was cut off - I guess you're too
hurt for us to consult your medical records... Do you have a backup in
your foot, perhaps?"

 

________________________________

From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Jared Still
Sent: Monday, October 24, 2005 10:40 AM
To: Oracle-L Freelists
Subject: Death of the database

 


Anyone seen their workload reduced due to unstructured data?

Death of the database

As improvements in networking technologies lead to real-time 
connectivity to any data, that data will be best kept closest 
to its natural source rather than at the intersection of a 
database's row and tuple. At last week's Symposium ITxpo, Gartner 
analysts backed up that premise with two examples: an RFID-tag 
equipped can of soup, and a chip embedded in the back of a human 
hand. Must data always be stored -- or cached -- in a database? 
If not, it's time for DBAs and BI vendors to to reinvent themselves.
http://ct.zdnet.com.com/clicks?c=625728-4778725&brand=zdnet&ds=5&fs=0


-- 
Jared Still
Certifiable Oracle DBA and Part Time Perl Evangelist

Other related posts: