Overview
The problems EDH is trying to solve are particular
to large complex organizations and thus inherently complex
themselves. The weakness of many architectures in this domain
is the attempt to try and resolve complex problems with
complex solutions. The success of the EDH architecture lies
in a new approach to the classic 'modular decomposition',
of breaking down the overall complex problem into smaller
more manageable pieces.
The result is a set of 'Lego blocks' that
are small, simple but powerful and can be rapidly assembled
to deal with any new problem. By using a Lego based approach,
solutions may be adapted to the changing needs of the organization,
new building blocks may be added and existing solutions
may be modified. Implementing solutions to large complex
problems becomes Childs play.
What is this 'Lego'
approach?
Consider the problem of creating and routing
an electronic purchase requisition. A user entering information
into this document needs to enter information such as 'cost
center', 'price', and 'currency'. 'Supplier', 'delivery
location', 'technical contact'. The classic approach would
be to use a design tool to build a form with the necessary
fields, lookups validations etc. But what if in 6 months
time the problem changes? What if the needs of the organization
change? What if you then need to solve a similar, but slightly
different problem?
The EDH approach is to build a set of generic
building blocks which internally we have called "CBOs" or
"Common Business Objects". These CBOs are the generic building
blocks which contain the specific business knowledge and
validation rules about particular business objects (e.g.
locations, currencies, people etc). So anywhere a 'person'
is required in the purchase requisition (e.g. for technical
contact), then the "person' CBO is used. One may think of
the CBOs as the atomic building blocks, and the solution
to the more complex business problem (in this case the purchase
requisition) is the chemistry of assembling a molecule from
the appropriate atoms.
