Re: the relational model of data objects *and* program objects

Bernard Peek wrote:
> In message <426507d3$1_at_news.fhg.de>, Alexandr Savinov
<savinov_at_host.com>
> writes
> >mountain man schrieb:
> >> "Bernard Peek" <bap_at_shrdlu.com> wrote:
> >>
> >>>It really is a good idea to have some things centralised.
> >> I agree. The efficiency of management, coordination and
> >> change-management of a thing increases with centralisation.
> >> Do counter-examples exist at all?
> >
> >It absolutely clear that a centralized system is easier to create
and
> >operate so it is the best choice.

>
> I dispute your assertion that centralised systems are necessarily
easier
> to create and/or operate.

How can you? With a centralized system, you have more design options (you can do a monolithic app if you like, though you don't have to), you simplify your deployment, you have a single point of failure, etc. Operating easily depends on the underlying system (e.g. the server), but I still have no doubts it's easier than a distributed system (which could involve a web server, app server, DB server, communications between them, potentially different O/Ss and platforms, etc. etc.).

> There are systems that work better with central control and systems
that
> work better with decentralised control.

I agree that some systems can't be centralized, at least without violating some requirements; that doesn't mean that most systems are that way, or that it's not easier to do a centralized system WHEN POSSIBLE (e.g. when it meets requirements).

99.9% of mainstream software development discussions surround distributed systems, but while some distribution is necessary in light of application and organization requirements, distribution has simply become orthodoxy.

> You are making broad assertions
> and we can see exceptions to them.

You have yet to name one, however.

> You need to focus your argument so
> that you can define precisely what type of system will definitely
> benefit from decentralisation.

Likewise.

> Until then I don't think you have an
> argument that is worth listening to.
> I already know that some systems
> benefit from decentralisation. I need you to tell me how to
> unambiguously identify exactly which ones.

Uh sure, "unambiguous" advice like this is likely to be forthcoming. Can you do the same?

To sound even more juvenile: you brought it up, you go first. Name some criteria.

• erk