Re: Data Redundancy

From: dawn <dawnwolthuis_at_gmail.com>
Date: 16 Feb 2006 19:02:11 -0800
Message-ID: <1140145330.945617.288490_at_g14g2000cwa.googlegroups.com>


Marshall Spight wrote:
> Jan Hidders wrote:
> >
> > Just out of curiosity, in which MV DBMSs would it be possible to specify
> > a declarative integrity constraint that said that one relationship

There are no declarative constraints in MV. What does it matter if they are declarative, procedural, OO, functional or whatever type of constraints?

> > (emp->dept) is the inverse of the other relationship (dept->emp)? Or how
> > else would you prevent inconsistency in your database?

Just as it is prevented in other systems -- with hand-coded software associated with these data. In the case of an RDBMS that might be written by a dba using a declarative language, where with MV, it would be written by the software developers in services used for updating the database. An application developer would then write code using these services rather than directly accessing the database. Oddly enough, this services approach is becoming popular now even when using an RDBMS. These "return links" (which are admittedly redundant data) are very common, incredibly common. I'm guessing there are times when problems occur that toss things out of whack where a repair routine would need to be run, but no such issues ever bubbled up to me as I recall. I never worried that payroll would not go out or students would not get registered because the return-links were out of synch, for example.

>
> That's the money question. How is this done in MV?
>
> To my mind,
> SQL handles this case just about perfectly. It was exactly the
> example that made me want to investigate SQL in the first place,
> after I had been pointing at it and laughing for quite some time.

And here I went the other direction, assumed it deserved my respect (and it does) and promoted it then experienced that it just wasn't quite up to par on some important fronts like developer productivity and cost of ownership. If it isn't as practical in these respects as other approaches, even if it has some elegance, then I would rather keep it in the lab (or perhaps competitors' shops) until it is. --dawn Received on Fri Feb 17 2006 - 04:02:11 CET

Original text of this message