a request for some experienced input.....

Hi all-

We are leaning toward using postgresql for a current project we are
expecting at least 10 million data points.  Currently the design has at
least 20 tables (and we are expecting at least 5 more).  Probably one
table will hold at least 500,000 tuples (probably 5 columns the larges
field being varchar2(250) ).  Sorry-I don't have better descriptions of
the data at this point.  Honestly, we are especially new to the whole db
admin issues from a project of this size.

Could someone tell me the state of the referential integrity
implementation.  I have been reading the newsgroups for about 3 weeks
and have seen various references to these issues.  I have just read the
todo list and they mention that 'MATCH PARTIAL' and 'PENDANT' are yet to
be done....what is this????

Frankly, I am concerned when I read that there may be 'issues' with
referential integrity especially since some of our data will be fleeting
(i.e. there will be no files outside of the database from which we could
re-capture the data).  I am especially paranoid about losing this type
of data (as anyone would be!)

Since we are just starting this...we would be in a position to upgrade
to 7.1 when it is released (although we do have concerns about having to
upgrade when every possible release comes available to make sure that we
have the strongest implementation of referential constraints).

Although these admin issues may be mitigated by the standpoint that we
hear that Oracle is a bear for administration....since we have never
done this we don't really know what this means....Can anyone comment on
this?

Our other option would be to go with Oracle in which we 'think' that we
could just use once single release (maybe we are naive here?)

I would be extremely thankful if someone could  please discuss these
issues....and how worried I should be about referential integrity issues
and about upgrade issues?


Thanks a lot-
Beth