How XML Threatens Big Data

[Michael Ludwig]

A. Steven Anderson schrieb:
> How timely....;-)
> http://dataspora.com/blog/xml-and-big-data/

Well, if data is tabular, what benefits can be obtained by storing it as
XML? None that I could see. If you later need XML, it is really easy to
generate. So why would you store it as XML in the first place? Because
you've fallen victim to some hype? That does not mean that "XML fails".
(See Nestor's comment on the blog.)

If your data is full of optional stuff and pretty irregular in general,
XML can mean a significant improvement. A twenty-table-join is not easy
to understand. I've never set up Hibernate, but I think that isn't easy
either - a handy 700 page manual is required to bridge the OO/RDBMS/OO
mismatch. For those situations, objects or document databases may be a
much better match. There's more than one tool in the box.

Also, it never fails to amaze me how people can think that documents
exceeding a certain size - 1 MB, okay: 5 MB - are a good idea. Just
imagine the amount of processing that has to be done before discovering
on parse completion that something's wrong.

-- 
Michael Ludwig