In my past development efforts (in which I've been working alone), I've been
content to gather the classes into a single DLL.  However, in my current
project I'm working with a team of developers and I'm considering the
benefits of separating these logical interfaces into separate DLLs that I'll
install as a single MTS package.  I'm further considering moving my Database
access interface into its own DLL that I'll dynamically link to the MTS
DLLs. My goal is to make the entire architecture more modular and easier to
manage, update and document -- without sacrificing performance.

Again, my past practice has been to keep everything together in a single
component.  Before going forward with this new approach, I'd appreciate
hearing from anyone who has been down this road or has advice about the
advantages/disavantages of this approach.

Thanks,

P. Bernhardt

I have done this in the past and it works well.

try to model your app in such a manner as to minimize the number of objects
you have to instantiate.  This tends to be extremely costly !

I have seen many people attempt to be really Object Oriented Purists with
their design and due to that, their app performed like crud and couldn't
scale for beans.

You also will want to break up your classes into classes that need a
transaction, and those that don't.

--
Cheers,

Greg Jackson MCSD, MCT
Sr. Software Engineer
STEP Technology
PDX, OR
http://www.steptech.com