system architecture

solutions are like bandaids, they are temporary. Unless the team steps back and considers a systems engineering approach. The SE approach is not solution driven like design engineering.

The principle we are dealing with is Life Cycle Costing or Total Cost of Ownership which is the associated costs of a system over its life cycle where the design phase is about 10% and the operations phase is 60-70% of the total costs. It is easy to see the opportunity to spend a little more effort in the design phase in order to create a true enterprise solution that will reduce the overall life cycle costs of the system.

Design engineers will plague a project by jumping straight to the answer and plugging the first solution within their experience that comes to mind. This limited approach is highly rewarded and satisfying to a poorly-managed program that is behind schedule and over-budget. With each knee-jerk solution, all of the adverse effects have to be accepted or managed. This can mean unexpected customization to interfaces and business methods. A better approach would take a moment to understand the constraints and trade-offs to make a balanced decision.

Since the SE approach considers the entire life cycle of the implementation, it will always be worth the scientific sanity check to go back to square #1, the customer's requirements. Understanding the endstate that will most satisfy the customer will help the team to properly weight the trade off criteria will provide the most value and satisfaction to the customer. Each one of the performance parameters must be prioritized and traded among cost and schedule concerns as well. With objective decision criteria, discovery shall be made to find possible alternative approaches or solutions that can be traded.

How far that you have to go back is up to you and how well documented your architecture is. Ideally, systems architecture will have a concept of operations (CONOPS) that describes all of the system components, capabilities, and the end-users (actors). The up-front investment of creating good architecture will consistently be referenced to validate the incremental system development. It ensures that the business is modeled and prevents the business from having to tailor around technology. Beware that a CONOPS can be a quick description (or marketing presentation) of a solution or it can be a visionary future-state that aligns all future interfaces.

Engineers and Scientists are like lawyers, they are happy to create more problems than they solve; its called job security. You can save the world once or keep rescuing the team from the long-term affects of every solution.