Re: OODB

"Scotty" <invallid_at_invalid.spam> wrote in message news:j0d4uuk4qhqjeoojutk8g66d2k180m3qp5_at_4ax.com...
> neil wrote:
> >"Scotty" <invallid_at_invalid.spam> wrote in message
> >>
> >> Do they exist or are they a figment of a college lecturers
> >> imagination?
> >
> >Lots exist and are absolutely wonderful things to study. You should avoid
> >them in the real world, however, if the following things are important to
> >you:
>
> That's basically what I was wondering, are they just an academics idea
> of a good time or are they truly implemented in 'Real Life'.

There are examples from real-life and some success stories. However, I come across very few people doing serious large-scale development work with OODBs today. I help run JSIG in London (800+ members from mainly the financial community) and some members have used this technology (e.g. Nomura, JP Morgan, Chase Manhattan), but they have tended to be for very specific problems. In some cases, people have regretted using OODBs afterwards, because of the pain they had to go through to really make them work well. In various jobs I have have held in recent years, I have been fortunate to have got to travel quite a bit and most of the developers I have talked to around the world during this time either have not heard of OODBs or have heard of them, but have not considered using them, because relational is good enough. For jobs in this space, try your favourite recruitment consultancy and see how many positions they have for OODB developers. For textbooks in this space, try your favourite bookshop and see how many books you can find on the product of a specific vendor. It was Roger Sessions that once said about OODBs:

1. Nobody uses them.
2. Nobody is making money from them.
3. They serve no useful purpose.

From my experiences over 10+ years in OODBs (as a researcher, consultant, publisher), I would say 1. and 2. are fairly accurate. I leave 3. for discussion :-)

> My
> college seems to have dropped a large amount of syllabus dedicated to
> OODB's giving only two lectures when previously it was 5 with a large
> content of the final exam, OODB's have been completely dropped from
> the exam and we've mainly concentrated on OLAP, Data Warehousing, Data
> Mining, XML and security, which are all fairly useful study areas from
> my understanding.
>
> >1 data integrity
> >2 ad-hoc querying with SQL
> >3 data-extraction using third party reporting tools that rely heavily
on
> >2 above (E.g. MS-Word)
> >4 data is manipulated using multiple applications
>
> Oh, so pretty much everything a database is needed for :-)
>
> >They are, however, a programmer's dream come true, and can be used to
> >produce prototypes very quickly where users aren't important. My own
> >particular favourite for this sort of prototyping is the ZODB included
with
> >the open source Zope (http://www.zope.org) - a very handy web application
> >development tool.
>
> Yes it fits in perfectly with the OODM, it's a shame it takes no
> account of the business model. My thoughts on OODB's are that you'd
> end up shaping the business structure to fit the database whereas the
> relational model looks at relationships between the data and hence
> fits the business model, sort of like the tail wagging the dog. Would
> you agree or not?

Actually, the idea is that your business objects would map to your programming language and database objects much better, since your are not having to think in terms of relations.

--
Akmal B. Chaudhri
Zone Editor, Special Projects (Eclipse, Grid, Web Services)
IBM developerWorks - http://www.ibm.com/developerWorks/
WebDatabases ------- http://www.btinternet.com/~webdatabases/