--- "Thomas B. Passin" <> wrote:
> On the other hand, if you create a well-normalized
> data model to begin
> with - when that is practical - you could probably
> implement it either with
> xml or with a relational database. What are the
> differentiators between
> going one way or the other?
All other things being equal, I'd guess that the XML
approach would be faster and more convenient in
read-only (or heavily weighted toward reading)
applications because you wouldn't have to join the
data back together from multiple tables to recover a
"document." [The old "take the car apart at night and
reassemble it in the morning" analogy that OODBMS and
XMLDBMS advocates like to use and RDBMS advocates just
hate :-) ] But the more updates of existing data,
especially by concurrent applications, the more the
RDBMS approach is likely to be better or even
necessary. XML's lack of a coherent referential
integrity story can bite hard in such situations.
