Name Your Here Final Examination OMSE 532, Software Architecture, Spring 2009
Please email your exam answers as [e-mail] no later than 11:59 PM, Wednesday, June 10th. Your answers should be kept concise and to the point. If you have any questions about what is being asked for or what you are allowed to assume, please ASK! Likewise, you should state any assumptions you are making if you aren't sure from the question. In general, am looking for you to apply the principles, methods, and conceptual structures from your text, readings or the lecture notes to answer each question. While your work experience or material from other courses may be applicable to the problems described, your answer should first demonstrate an understanding of the relevant material from this course (i.e., I need to know how to relate your answer to the course material however eccentric you may find it). Your answer should be provided as a Word document (e.g., insert your answer below the question) or as plain text.
1. Throughout the course, we have argued that it is important to capture the architectural design decisions in written form (for example, in the views and other architectural documentation). This is what the book refers to as a "transferable abstraction". I suggest that the reason it is important to document the architecture is to assist in maintaining adequate control over "the products and processes" of the software development. Considering the architectural business cycle, briefly answer the following: 1.1. Typically more than one abstract representation (view) of the architecture is produced. Why would we need more than one view of the architecture? How should the architect decide which view(s) to provide? Start of answer...
1.2.
How do accurate representations of the architectures help the development team maintain control over the current software development from detailed design through deployment and maintenance? Briefly describe three different possible uses of the architectural representation in helping maintain control i.e., describe who would use the architectural representation and what they would do with it. (Remember that "maintaining control" means that we provide software that meets its requirements, on time, and within budget). Start of answer...
1.3.
The usefulness of the architectural representation extends beyond the current development and deployment cycle As we've discussed in class, all systems have an architecture whether it is explicitly represented or not, and most systems go through multiple development cycles. Under what circumstances and in what ways are subsequent development cycles adversely affected by a failure to adequately capture the
-1-
architecture in a communicable form? Give at least two examples of specific development activities (i.e., design, coding testing maintenance) in subsequent cycles where the lack of documentation could lead to a loss of control. For each, briefly explain how the activity could be affected and why. Start of answer... 2. Your company is in the process of acquiring a major piece of software to handle the company's internet storefront and customer support (the specific application is not important). Management is considering products from two different vendors. You have been asked to evaluate the products with respect to both fit-to-purpose and total cost of ownership. Both products (call them "A" and "B") provide the same functionality for about the same cost. However, there are people in the company who are interested in other properties of the software. In particular: a. Management expects the number products, number of customers, and volume of sales to rise over time. There is concern that the system be able to handle much larger volumes without decline in response time. b. Marketing expects to expand the storefront to add additional services. It is important that the vendor be responsive in adding capabilities to the system in a timely manner. Both vendors say they are willing, but are they able? c. The internet storefront is expected to be the primary source of company sales and its interface to its customers. It is critical that the vendor be able to fix problems and make small changes quickly. Your task is to analyze the candidate systems and provide an objective evaluation of the strengths and weaknesses each of relative to your company's needs. You plan to have the vendor's representatives come in to go over the system architecture. Relative to that task, answer the following: 2.1. What information would you need to obtain before meeting with the vendors? Briefly describe what kind of information you need, where you would get it, and how you would record it for subsequent use in the vendor analysis and comparison. Start of answer... 2.2. Which views of the architecture would it be appropriate to ask the vendor to provide and why? Start of answer... 2.3. Briefly describe what you would ask the vendor to explain or show about their architecture(s) or their design approach to convince you of the superiority of their product. Start of answer...
-2-
2.4.
How might you go about quantifying the relative suitability of the two products (i.e., providing a "metric of goodness") that would give management some idea of how well the vendor products meet the company's needs? (Assume that for obscure reasons, management wants a single number, sort of like a golf score. Use materials we have covered in class, not other methods like the CBAM.) Start of answer...
2.5.
Could you adapt the ATAM approach to support your comparison? If so, briefly describe what parts of the process you would use for your evaluation. If not, describe why not. Start of answer...
3. The class lectures have made a distinction between developing architectures for families of systems and developing software product lines. One of the reasons we argue for this distinction is that developing an architecture for a program "family" can be appropriate and cost effective regardless of whether one ever intentionally develops a "product line" or not.. Considering that discussion and your experience with the buoy problem, answer the following:
3.1.
Briefly explain why it often makes sense to develop an architecture for a "family" even when we only plan to deploy one version of the system (which we will maintain). In particular, in what way is a (supposedly) single system actually a family of systems? What distinguishes the family members? Start of answer...
3.2.
If it makes sense to view a development as a family, I argue that it is often cost effective to build a common family architecture whether or not one ever seeks to develop concurrent versions of the system (e.g., a product-line). In such a case, what quality attribute(s) can one achieve by building such an architecture and what is the payback (i.e., what can you do more cost effectively given such an architecture)? If you identify multipl...