Re: Resiliency To New Data Requirements

David Cressey wrote:
> "dawn" <dawnwolthuis_at_gmail.com> wrote in message
> news:1155093201.689872.230930_at_h48g2000cwc.googlegroups.com...
> > that pre-dates me. MultiValued databases pre-date 1NF. I'm definitely
> > not diss'ing the relational model...
>
> I think you ARE dissing the relational model, Dawn.

Yes, I definitely am -- that is the first phrase in the simile

>
> > ... like the RM-advocates have trashed
> > the models that came before it (even though they seemed to have ZERO
> > emperical data to prove their point).
>
> Not true, at least in terms of the material I read when I first read about
> relational databases. The material I read certainly compared the relational
> model with two earleir models, the hierarchical and the network model.
> They certainly pointed out some weaknesses in those two models, ones that
> the writers claimed the relational model overcame. However, it's a gross
> overstatement to say that they "trashed" those two models.

Perhaps it would be more precise to say that they "threw the baby out with the bathwater." I found arguments, but no emperical data, that showed the relational model to be beneficial to those using it. If you are aware of any emperical data, particularly any that shows that the RM is more flexible over time than either MUMPS or PICK that predate it, I would really like to see that.

> Again, I'm
> commenting on the material I read about 20 years ago. The material you read
> may have been different.

Yes, so please pass along any pointers to emperical data comparing data models.

> I also disagree about the lack of empirical data to prove their point. The
> experience of tens of thousands of database builders and users may not come
> up to the standard of "scientific data", but it is certainly "empirical
> data". It is you who intentionally ignore that enormous body of data,
> calling attention only to the failures that have cropped up in that
> experiences and overlooking a massive amount of success.

I'll admit there is marketing data to suggest that on the whole the products based on SQL have been hugely successful. One thing done very well was to have a standard language that the industry tapped into. And I definitely agree that SQL-based products work. My sense is that the marketing of the relational model has been excellent for decades. So, yes, there is data to prove that Windows NT was a better OS than OS/2, from that perspective. So, I will accept your point on that if that is what you mean.

> I certainly have NEVER maintained that the relational model is the end of
> history as far as data modelling goes.

Great, but there are some who still do. I am only saying that it is time to move on. Some are moving on by redefining the model significantly (redefining normalization, for example, suggesting we return to 2-valued logic, etc.).

> But a return to a pre-relational
> model seems to me to be a waste of time.

I agree. If we are going to start somewhere and move forward, we might be well-served to look to what works today outside of the RM (even though it, of course, typically markets itself as relational). Is it less expensive to work with Cache' than Oracle given such and such an environment? If so, why?

> I remain unconvinced by your
> arguments, as far as I got in reading them.

That's fine and I appreciate you reading. I took on too much "real work" and have slowed my writing down. Happily I'm seeing enough movement in the industry since I started delving into this topic that I am optimistic that many of my target areas (NF2, 2VL for starters) will become much more popular in the future. Cheers! --dawn Received on Wed Aug 09 2006 - 15:16:37 CEST