Re: Need help to understand difference, and contrast between Relational database model and the Object-Oriented model

From: Corey Brown <>
Date: Thu, 11 Dec 2003 13:39:25 -0500
Message-ID: <AO2Cb.62$>

"Alfredo Novoa" <> wrote in message
> "Corey Brown" <> wrote in message
> > > > Alfredo, why don't you explain to us why you think this answer
> > > > nonsense.
> > >
> > > Because it is evident for anybody with a grasp on data management.
> >
> > This is exactly the type of answer that I would expect from somebody
> > like Bob B. Why must you guys always answer direct questions with
> > inappropriate remarks like this.
> You didn't asked why it is nonsense, you asked why I don't explain
> that. I answered appropiately to your direct question.

    Sorry Alfredo, I thought we were both focused on the broader     question instead of the getting bogged down in the details.

> > If you have the knowledge and the
> > ability to express that knowledge to others, why don't you take a
> > few minutes out to lay down some cold hard facts, instead of just
> > telling us to go educate ourselves?
> Among other things because a few minutes are not enough if you don't
> have a clue, but I gave very good bibliography.

    Ok, I can agree on the bibliography part, but let's calm down     on the "clueless" part for a little bit. You don't know me from     Adam, so please stop making snap judgements about me and     the others who post on this forum.

> > Why can't you step up to the
> > role of teacher and start explaining why you "think" one technology
> > is better than another?
> I can, but I don't want. To know that is the duty of any professional.
> BTW we are talking about models, not about technology.

    Actually, the original poster may NOT have been talking about models but     about physical implementations. No one is confusing the two except for you.

> >
> > >
> > > > I am also pretty sure
> > > > that ODBM
> > > > systems do use direct pointers to relate objects together.
> > >
> > > And I am pretty sure that SQL DBMSes use pointers internally.
> >
> > So your point about network databases being obsolete and discredited
> > doesn't count here? If using internal pointers is so foul, why
> > it
> > apply to your last statement? I know, I know go educate yourself.
> You don't know the difference between the logical and the physical
> levels, you are more ignorant than I thought.

    Yes I do Alfredo, but to date we have not been talking about the differences

    between physical and logical. We have been talking about physical differences

    between ODBMS and RDBMS implementations, at least I have, you seem     to be reading from a different hymnal.

> > > So they are based in a primitive obsolete and discredited approach.
> > > That's all.
> >
> > So what! There are many many examples of technologies that have
> > been eclipsed by better designs. It doesn't mean that the
> > early designs are not practical or useful anymore.
> If the new approach is better in all situations then the old approach
> is not useful anymore.

    I completely disagree. Have you given up the land line in your house     just because cell phones are considered better technology? Have you     switched from antenna and or cable to satellite because HDTV is     available. Will those technologies eventually eclipse older ones, you     bet, but not over night.

> > > Perhaps in very special circumstances when the flaws of the current
> > > SQL DBMSes are more important than the network model inherent flaws,
> > > and the flaws of the concrete OODBMS implementations.
> >
> > I don't think the circumstances are all that special.
> Because you ignore the fundamentals of the data management field.

    Sorry Alfredo, I don't ignore the fundamentals anymore than you do.     But I am fairly grounded in reality though and will continue to choose     the right tool to do the job based on the ENTIRE scope of the work.

> > And I certainly
> > hope
> > that application architects are looking at more than just the flaws
> > associated
> > with specific db technologies
> There are many application architects that ignore the funtamentals of
> data management.

    See my statement above, but yes I do agree with you on this point.     A lot of this boils down to religious beliefs and unfortunately you cannot

    dispose religion with technological fact.

> The implementation flaws are the only reason that could make more
> appropiate a tool based on an inferior approach.

    Excellent, now were grounding out. There are implementation flaws     in ALL technologies. Those that can see and understand those flaws     are not doomed to make the same mistakes over and over again.

> The very first implementations of superior technologies are often
> worse at practice than the older products.


> > > Me too, but I try to base my decisions on accurate information.
> >
> > Ok, but certainly you're not basing your decisions purely on the
> > theoretical
> > disadvantages of an ODBMS over an RDBMS.
> This thread is about a theoretical question: the differences between
> the relational and the OO approaches, but it seems you don't
> distinguish very well between model and implementation.

    There you go again. We weren't talking about models to begin with.     Ok, maybe you were but I was not. So cut me some slack here ok?

> > The whole picture of
> > how the application will be used, how much data will be stored, how
> > will be retrieved, the complexity of the data relationships and the
> > environment that
> > the application must work in must also be taken into account.
> The complexity plays against the network approach.

    I believe that the circumstances surrounding the requirements play a     major factor in this argument. As an example, several years ago I     was responsible for the design and implementation of a modified     Dijkstra's algorithm to be used in automatic network restorations     of core transport facilities at a large telecom company. The requirement     was for us to be able to complelely restore service for the first 100     failed T3 lines within 5 seconds of the actual failure. The network     was comprised of thousands of nodes with hundreds of thousands     of interconnections between them (mesh network). The only way     we could meet the requirement, while using a relational database     as the underlying data store, was to build out a network representation     of the network in memory first. There was just no way to execute     queries against the RDBMS fast enough in order to meet the requirements     for the application. We're talking about a very complex data model here.     Fortunately we were able to implement the solution using a high performance

    object database (Versant) and were easily able to meet our performance     requirements without having to build an in memory representation of the actual network.

    Did we lose anything by moving to an ODBMS, you bet we did. We lost     out ability to run ad-hoc queries against the data. But then again we're talking

    about a very specialized database. This was NOT a shopping cart application

    used for buying CDs on the internet!

> > My own automobile is theoretically and practically inferior to a new
> > hybrid
> > vehicle, but does that mean I have to stop using my car today just
> > because
> > better technology is available?
> No, but if the new technology is actually better we should stop making
> traditional cars.

    Agreed, but again reality gets in the way. There are certain classes of     vehicles that do not lend themselves well to hybrid power plants. Hybrids

    are not known for the kind of power, in terms of torque, necessary to     move an 18 wheel truck up a 10% grade. Electric motors can produce     mountains of torque, even more when coupled to a reducing gear set, but at

    the cost of power consumption. How does this relate to the current     discussion? It's just another example of how alternate and often inferior

    technology can be used to acomplish a given (real) task.

> > I firmly believe that both ODBMS and RDBMS technologies have areas
> > which
> > each may excel over the other.
> And your belief is based on ignorance and inaccurate information.

    Let's lose the whole "ignorance" thing ok? We are all ignorant about a     great many things, but you don't hear me calling you ignorant because you

    can't build jet engines do you?

> > years, so believe me when I tell you that I have seen more than my
> > share of applications
> > where the technology was decided on before the requirements were
> > analyzed, with the
> > end result being a miserable failure.
> Again, The Relational Model and The Network Model are not
> technologies, they are models.

    Great! so let's make sure we're both reading from the same page before     we start in with personal insults, ok?


> Regards
> Alfredo
Received on Thu Dec 11 2003 - 19:39:25 CET

Original text of this message