Re: Newbie question about db normalization theory: redundant keys OK?
Date: Sat, 15 Dec 2007 13:13:52 GMT
"David Portas" <REMOVE_BEFORE_REPLYING_dportas_at_acm.org> wrote in message
> "Brian Selzer" <brian_at_selzer-software.com> wrote in message
> >> You are saying that this is a distributed database but that it lacks a
> >> mechanism for accurately propagating changes out to all its nodes?
> >> Well in my view such a DBMS would be broken. It surely violates Codd's
> >> principle of "Distribution Independence". Let's follow your example to
> >> its conclusion though. The solution is to replace whatever copy of the
> >> Blah relation exists in the application with the new Blah relation
> >> that superceded it. Now all emails reach the correct address and there
> >> is no problem that requires a different key.
> > Forgive me for butting in, David, but where did you come up with the
> > that it is a distributed database? What have disconnected
> > applications--that is, applications that use something akin to
> > disconnected ADO recordsets or ADO.NET datasets--to do with distributed
> > databases?
> > The question is: for how long is the data that was just read out of the
> > database considered to be valid? Until the next update? Or is it stale
> > as soon as its read? Does it have something to do with transaction
> > control or locking? If several updates occur between the reading of one
> > piece of information and the reading of another, how can you be sure
> > any answer that involves both pieces of information is correct? How can
> > you be sure that you haven't read the same information twice?
> I agree that these are important issues of application design. I don't
> they need to affect the database logical design in this case.
They need not afect database design for a single database built in isolation. However, if there is a data architecture that pervades the entire enterprise, and both databases and applications inherit the semantics of the data and to some extent the form, from that pervasive data architecture, then the issues raised become very relevant. Received on Sat Dec 15 2007 - 14:13:52 CET