For sure Oracle has done little to win me over with *anything* other than
the database. Even their jdbc drivers are broken in places. What's up with
that? Now, more to the point of hard data...
Last winter I was the lead architect on an e-business web site that used the
front half of j2ee, namely jsp/servlets and simple Java beans. Rather than
using ejb for the enterprise data I opted for a home spun o-r mapping
solution that squirted Oracle db data into Java objects via
introspection/reflection. The project manager wanted to use Oracle's ejb
server, JServer, and Oracle's business objects (ejb) but from the looks of
their integration with JDeveloper I chose to dodge that bullet (CMM calls em
risks, fyi). The overall archtecture we went with looked like this:
browser ... jsp/servlet->java beans->java business objects->oracle 8
For our insane time constraints I believe leaving JServer (and their totally
weird o-r mapping junk!) out of the picture was right smart of me; although
had I had jboss then I think we could have done ejb successfully --
deployment is sooo quick and easy at 10-1 compared with weblogic. The
project was a success, though slightly over schedule. Why? Java? No! Our web
designer kept flaking out and didn't show up with html when we needed to
move on. I was quite happy with JRun as a jsp/servlet engine. I thought it
was very snappy, although I'll say that the package came confiured in a
strange way and it kept us guessing for a few days (required paid calls to
tech support). We deployed on NT. The site is still running and being
extended, although the sponsor's marketing plan has failed to draw the
number of users desired. Otherwise I am sure they would LOVE to have
scalability problems and the NEED to move my "java business objects->oracle
8" to "ejb->oracle 8". I will refrain from identifying the site or company
to protect the guilty :-)
<suggestion>
If you doubt ejb, deploy using the above architecture and then back fill
with ejb as scalability demands (read, if your product sales succeed). If
you do the above architecture correctly, plugging in ejb later should not be
a very big deal at all... just replace the "java business objects" with
equivelent ejbs.
</suggestion>
Vaughn
> Jeremy I'm not sure how much you've looked around, but you may want to
look
> at Orion if you're looking for something thats more affordable and is more
> featured as a commercial product. I cringe at the idea of using a
databases
> vendors internally developed appserver - I've been fighting this on my end
> on a project I'm interested in. Sure its nice to say 'well the database
> supports that', but just as with anything else once you go down that dark
> road there is usually no coming back from the embedding of LOTS of
> proprietary code into your system. I think Oracle makes a good database,
but
> their tools and support systems are of the devil :)
>
> -----Original Message-----
> From: Jeremy Clymer [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, August 08, 2000 2:51 PM
> To: jBoss
> Subject: RE: [jBoss-User] Case studies
>
>
> > Come on! I need "real life" we have so much IT people on the list with
> > ongoing projects. Can we hear your failures and your successes?
>
> In my department, we have been using Oracle 8i's JServer for EJBs with
> terrible results. They force you to use too much proprietary code and its
> not easy to develop on. We chase obscure errors and need to spend time
> working around different oracle problems that come up. I love their
> database, but they are so far off the original intention of the EJB spec.
>
> Thats why I've been looking into JBoss. A free app server thats written
to
> the latest spec. The only problem that I have with JBoss is that it won't
> be complete in time for my applications to go live. Therefore I have
> recommended that we go with Weblogic. It comes with its own problems,
such
> as high cost, but the fact that it is proven in the industry to provide
> scalable solutions to large projects gives the project managers the
> confidence they need.
>
> JBoss on the other hand is not yet a complete solution. I'm sure that
when
> it is done, it will compete with the best app servers on the market. But
> for my needs, its development won't be complete in time.
>
> Jeremy
>
>
>
> --
> --------------------------------------------------------------
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Problems?: [EMAIL PROTECTED]
>
>
> --
> --------------------------------------------------------------
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Problems?: [EMAIL PROTECTED]
>
--
--------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Problems?: [EMAIL PROTECTED]
