Home > Articles > Software Development & Management > Architecture and Design

So Now I'm A Software Architect. What Do I Actually Do?

  • Print
  • + Share This
  • 💬 Discuss
A key part of the software architect's job is producing an architectural description of the system that defines the architecture's key functions, features, and characteristics for its stakeholders. Where do you start? What do you need to know? Nick Rozanski and Eóin Woods provide detailed answers to these questions, with useful suggestions on how to attack this fundamental document that underpins any development project.

The practice of architecture has a solid theoretical foundation that goes back at least to Roman times, when Marcus Vitruvius Pollio wrote the first-ever architectural handbook, de Architectura libri decem—"Ten Books on Architecture." Vitruvius described the architect's role and required skills and provided a wealth of material on standard architectural structures. Over the next 2,000 years, this definition was refined, extended, and underpinned by a solid theoretical, practical, and regulatory framework that today guides the work of every architect.

Unfortunately, everything in the previous paragraph refers to building architects. The position of the software systems architect is scarcely comparable to that of our counterparts in construction. We work in an industry, information and communications technology, which is scarcely over 50 years old, and it wasn't until the 1990s that we seriously started to think about how the elements of a computer system fit together into a coherent whole—the system's architecture.

While it's rare today for a large systems development project not to include at least one architect, and many organizations have a whole department responsible for developing architectural strategy and vision for the enterprise, there is very little consensus over what the job actually involves:

  • Who are our clients?
  • To whom are we accountable?
  • What are we expected to deliver?
  • What is our involvement once the architectural design has been completed?
  • Where are the boundaries between requirements, architecture, and design?

This situation makes things very difficult for the newly appointed architect and causes problems even for the more experienced architect. It also confuses our colleagues, our management, and the other people with whom we work.

If you've read the standard books on software systems architecture, you know that a key part of your job is to produce an architectural description (AD) of your system. The AD documents your architecture in a way that people can understand, and helps to demonstrate that the architecture has met their concerns. Your developers will use it to guide their design and coding work (they're certainly going to blame you if something doesn't work), and many other people are going to be interested in it.

What the books don't really tell you is how you should go about producing your AD. Where do you start? Who do you talk to? What do you put into it? What can you safely leave out? How do you validate it and get it signed off? There are no simple answers to these questions, but you have a much greater chance of success if you focus your attention on three activities:

  • Identifying your stakeholders and working with them to create an architecture that meets their complex, overlapping, and often conflicting needs
  • Using views and viewpoints to help you define and document the structure of your architecture, making what is a huge problem manageable
  • Using perspectives to help understand the "quality properties" of your system—such as performance, availability, and security—and to ensure that your system meets its nonfunctional goals

Let's look at each of these activities in a bit more detail.

Identifying and Engaging Your Stakeholders

Traditional software development has been driven by the need of the delivered software to meet the requirements of users. However, the people affected by a software system are not limited to those who use it. That is, software systems are not just used: They have to be built and tested, they have to be operated, they may have to be repaired, they are usually enhanced, and of course they have to be paid for.

Each of these activities involves a (possibly significant) number of people in addition to the users. Each of these groups of people has its own requirements, interests, and needs to be met by the software system. We refer to these people collectively as stakeholders.

IEEE Standard 1471-2000 defines a stakeholder in a software architecture as a person, group, or entity with an interest in or concerns about the realization of the architecture. Stakeholders (explicitly or implicitly) drive the whole shape and direction of the architecture, which is developed solely for their benefit and to serve their needs. Stakeholders ultimately make or direct the fundamental decisions about scope, functionality, operational characteristics, and structure of the eventual product or system—under the guidance of the architect, of course. Without stakeholders, there would be no point in developing the architecture because there would be no need for the system it will turn into, nor would there be anyone to build it, deploy it, run it, or pay for it.

Architectures are created solely to meet stakeholder needs. A good architecture is one that meets the objectives, goals, and needs of its stakeholders. It follows that if a system doesn't meet the needs of its stakeholders, it can't be considered a success—no matter how well it conforms to good architectural practice. In other words, architectures must be evaluated with respect to stakeholder needs as well as abstract architectural principles.

  • + Share This
  • 🔖 Save To Your Account

Discussions

comments powered by Disqus