Sorry I missed this thread, I was out of the country! I've been
wondering about this stuff for some time.

It seems to me that the major alternative to an EJB-based application
server is having each client maintain its own business objects that
talk directly to a database through JDBC (using any kind of O/R
framework). This achieves many of the requirements of a highly
scalable and maintainable system, and allows you to have a lot of
flexibility in terms of client-based connection pools, configurations
for databases, etc.

The main argument for having an EJB-based container, as opposed to
business objects on each server within a farm, is pooling and reuse
of object instances; I don't find the other reasons very compelling.
That said, it's not clear to me how these caching strategies would
work. The consensus seems to be that the database is probably the
best place for a lot of these kinds of tasks, and I haven't found any
consistency in the caching strategies implemented by EJB containers.
Trying to optimize your transactioning approach for your particular
container defeats the point of a standard.

??? Any thoughts? Without caching, what's the point of EJB? It's
slightly easier to update business objects, I'll grant, in one place
rather than several, but I don't find this particularly compelling.
There's a *lot* of overhead to using EJB objects rather than business
objects of some other kind (assuming you have a decent tool), and I
haven't seen a CMP solution (short of the stuff proposed in EJB 2.0)
that looks at all comparable to what we can do with our O/R tools.

Which, again, puts all the onus on caching. I'd like to see it defended.

Regards,
Ben Flaumenhaft

At 3:24 PM -0800 12/20/00, Beauchemin, Bob wrote:
>As long as the subject has been brought up, what is anyone's opinion on
>Persistence Power Tier? They make some amazing claims and seem to have
>addressed most of the CMP concerns like facade pattern, data chunking,
>stored procedure utilization, relationships, cache management and failover.
>Opinions? War stories? The patent for parts of the architecture is on file
>and accessible through the web and makes interesting reading.
>
>IMHO, CMP entity beans don't seem all that useful without a cache
>architecture to make it worthwhile to shuttle all that data back and forth
>between database and middle tier, with the gobs of network roundtrips that
>implies. Unless the EJB container has a better SQL engine than the RDBMS or
>90% of the access is row-at-a-time (and, pre-relationship support,
>table-at-a-time). Am I missing something obvious? Or are performance and
>scalability not major concerns?
>
>Bob Beauchemin
>[EMAIL PROTECTED]
>
>
>-----Original Message-----
>From: Chip Wilson [mailto:[EMAIL PROTECTED]]
>Sent: Wednesday, December 20, 2000 1:48 PM
>To: [EMAIL PROTECTED]
>Subject: Re: O/R Mapping Tools
>
>
>I've used it on three different projects.  I don't have the ability to
>support all the following claims right now - so take them for what they are;
>unsupported opinions.  TOPLink is bloated, overly-complex, and innapropriate
>for an EJB architecture.  The transaction API and implementation have to be
>shoe-horned into a JTA environment.  It doesn't support 2PC, or at least it
>didn't a year ago.  Debugging problems is a nightmare.  Developing and
>maintaining mappings takes longer than writing individual broker/mapper
>classes for each persistent object.  The ramp-up time for developers is
>extremely long.  TOPLink is really, really slow.  The cache is useless in an
>EJB environment.  TOPLink is really expensive.
>
>I have written, helped write, designed and used countless O/R frameworks in
>C++, Smalltalk and Java.  I even helped develop a commercial competitor to
>TOPLink, but it is no longer supported.
>
>In an EJB environment, I favor a minimalistic O/R framework with
>broker/mapper classes that execute JDBC calls and map between the schema and
>the domain model.  Entity beans access the persistence framework through a
>facade.  I typically prefer to hard-code the mapping logic, as this is the
>easiest to develop, learn and maintain.  If you must do some sort of
>meta-data-based mapper, keep it simple and allow the ability to do it
>manually in specific brokers.  A simple framework like this can be designed,
>implemented and tested (minus all the individual broker classes) in 40
>man-hours.
>
>>From: "B. Flaumenhaft" <[EMAIL PROTECTED]>
>>To: [EMAIL PROTECTED], [EMAIL PROTECTED]
>>Subject: Re: O/R Mapping Tools
>>Date: Wed, 20 Dec 2000 11:55:20 -0800
>>
>>
>>Can you give us some details? I'd be very interested to learn more. I
>>looked at TOPLink some time ago and was very impressed by its feature
>>set and customizability. It was simply too expensive, and I had some
>>doubts about performance and the runtime model.
>>
>>Why should we run away?
>>
>>Thanks,
>>Ben Flaumenhaft
>>
>>At 8:13 AM -0600 12/20/00, Chip Wilson wrote:
>>>Run away from TOPLink as fast as you can.  Run far away.
>>>
>>>>From: Thomas Bökamp <[EMAIL PROTECTED]>
>>>>Reply-To: A mailing list for Enterprise JavaBeans development
>>>><[EMAIL PROTECTED]>
>>>>To: [EMAIL PROTECTED]
>>>>Subject: O/R Mapping Tools
>>>>Date: Tue, 19 Dec 2000 23:37:50 -0800
>>>>
>>>>Hi,
>>>>
>>>>Parts of my current work is to evaluate O/R Mapping Tools products. Is
>>>>there
>>>>such a thing a report to study different O/R Mapping Tools . Currently we
>>>>are looking at Cocobase and Toplink. Can somebody help with the following
>>>>questions:
>>>>
>>>>- Anybody know about Cccobase or Toplink the quality and Performance?
>>>>- Any other O/R Mappers worth evaluate? Especially we would like to know
>>>>if
>>>>there is a product that the C++-part of the project can use too.
>>>>
>>>>Thanks  in advance
>>>>
>>>>Thomas.
>>>>
>>>>_________________________________________________________________________
>>>>Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
>>>>
>>>>=========================================================================
>==
>>>>To unsubscribe, send email to [EMAIL PROTECTED] and include in the
>>>>body
>>>>of the message "signoff EJB-INTEREST".  For general help, send email to
>>>>[EMAIL PROTECTED] and include in the body of the message "help".
>>>>
>>>
>>>_________________________________________________________________
>>>Get your FREE download of MSN Explorer at http://explorer.msn.com
>>>
>>>==========================================================================
>=
>>>To unsubscribe, send email to [EMAIL PROTECTED] and include in the
>>>body
>>>of the message "signoff EJB-INTEREST".  For general help, send email to
>>>[EMAIL PROTECTED] and include in the body of the message "help".
>>
>>
>
>_________________________________________________________________
>Get your FREE download of MSN Explorer at http://explorer.msn.com
>
>===========================================================================
>To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
>of the message "signoff EJB-INTEREST".  For general help, send email to
>[EMAIL PROTECTED] and include in the body of the message "help".
>
>===========================================================================
>To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
>of the message "signoff EJB-INTEREST".  For general help, send email to
>[EMAIL PROTECTED] and include in the body of the message "help".

===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST".  For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".

Reply via email to