Fine-Grained Vs Coarse-Grained Architecture

I got a chance to go through Telvent Software Architecture today and I really liked it. If you take a look at both Telvent and ESRI Object Models you will notice the difference.


Here is the Telvent  Object Model  for all their Products, and this is ESRI's.


ESRI ArcObjects are always referred to as "Fine-grained" while Telvent use the term Coarse Grained. The difference?  its just how much you chop of your design to smaller components.

The smaller your components, the finer it gets, you isolate your code, increase usability to the max. However because your components are so small now you will end up with ALOT of couplings. This will make your architecture very dependant on other objects and it will be very difficult to read as you can see in ESRI object model. Moreover, it is almost impossible to show a fine-grained architecture to upper management since it is too detailed.

Telvent's on the other hand does not go and break down their software components very far, instead they have a little bigger components, yes they cannot reuse all sub components but that makes the architecture decoupled and easily readable and can be presented to someone without architecture skills.