Every company runs certain business processes, either documented or not. These processes can be simple or complex, depending on how big and complex is the company. Collection of processes comprises certain structure. To name this structure 'architecture' or not, is a matter of taste.
Suppose you decided for a short trip outdoor and intend to pitch a tent for a night. With some skill and common sense, you can arrange pegs and pins, spread a canvas and have a reasonable shelter for a while. Should you hire an architect for this purpose? Probably, not.
Suppose now that you decide to build a house. You can affordably buy some bricks, tiling and cement. With a bit of luck and advise you can compose these into a sort of walls and a roof. However, there will be a risk that the construct will eventually collapse right on your head. Detailed architecture and professional builders are always advised to avoid a trouble.
Enterprise architecture is a comparable challenge. If you keep a stall, drawing an architecture for it might be an overkill. But running a decent corporation without well developed process architecture will eventually lead into disaster. Remarkably many skyscrapers on a business landscape collapse just by missing their architectural foundation.
There always exists a discrepancy between a model of business process, however well designed and accurate, and real execution of this process in a business environment. The reason for this gap is an unforeseen depth and hidden details inherent to any real process. Real business model of organization is ultimately unlimited in its depth. Going from highest management levels, it descends to individual departments, client relations, production units, technical code of equipment and controllers etc. In vast majority of cases, it is impossible and senseless to build a complete model covering all and every fine detail of the business. Omitted lower layers of the model create (pseudo) random fluctuations during execution of the model. Real execution paths of a process never follow its model exactly. However, in case of the correct model, we can expect to see that an ensemble of execution paths statistically converges to the model as to its average path over a significant set of observation...
Comments
Post a Comment