Re: RM's Canonical database

AndrewMackDonna wrote:
> Marshall wrote:
> > AndrewMackDonna wrote:
> >> Marshall wrote:
> >>> In other words, this approach doesn't scale.
> >> Hmmm.. I'd have to say I disagree, having seen it scale.
> >>
> >> An example I think.....(off the top of my head)
> >>
> >> http://www.google.com/apis/
> >
> > I was referring specifically to scaling in the size and complexity
> > of the development organization, not with the number of
> > connected clients. Sorry for the ambiguity.
> >
> > Marshall
>
> I didn't see any ambiguity, I thought thats what you meant.

I misunderstood your counterexample; sorry. Generally when people bring up Google they do so as an example of scalability of number of connections or data set size; I mistakenly thought that was what you were doing.

> I still don't see any either, the sheer size and complexity of those
> development teams using the Google API should far outnumber the usual
> number found in medium sized corporates*

But maps isn't a good example for illustrating where constraints go, because it's not an application that supports online editing. It's essentially entirely read-only, with a monolithic, separate, offline update process, rather than having a large number of small updates.

In general, search engines are not good examples to use when thinking about OLTP database management, because of this very reason.

Marshall Received on Mon Jul 03 2006 - 23:10:19 CEST