Big Upfront Framework design (BUFD)

Recently I have been thinking is it really worth doing Big Upfront Framework Design(BUFD).
I have taken part in designing several frameworks in my past, but never thought(or didnt get a chance) the best phase to design frameworks. It is agreed by default that frameworks are designed upfront before product development begins. Many highly intelligent architects, designers sit together for days and nights without proper sleep and design meta data driven,XML based,database independent frameworks!!!. But if we carefully look into this, upfront framework design is not really agile, and secondly, ROI is also bad.


* Doing Upfront Framework Design(UFD), would force the designers and developers to design application or develop application to fit the framework, rather than steering the development effort in satisfying the customer's requirements.

* UFD, is not profitable to the customer, because frameworks are generally designed to be as generic as possible keeping future requirements in mind, and due to which, customer would be spending money on "future" requirements in the present!!!

In fact I was trying to think can UFD be applied to Project development scenario, where the project's duration is less than a year ???... Now I feel that it does not suit here too. The reason being, any framework design and development would take atleast 3-4 months, and I feel this effort is not worth spending for a shorter duration project !!

Ok, the best possible solution could be:

* Start thinking about framework from the beginning (dont stop yourself), but dont start the design until you pass the inception pase.

* Start laying the foundation from Elaboration phase.

* As the product development progress, design, tweak the framework iteratively.

* The Framework should evolve over a period of time, and at the same time ensure that is "usable" (need not be complete) before construction phase !!

* keep refactoring, evolving and be agile