Software Architecture Metrics

Chapter 7. The Role of Measurement in Software Architecture

Eoin Woods

There are many definitions of software architecture, but in practice, most architecturally significant decisions are related to achieving qualities in the system that satisfy stakeholders, including performance, resilience, and security. Architects often refer to these as quality attributes or nonfunctional qualities.

These important, complex decisions are often difficult to make because they involve significant trade-offs between different qualities. For example, prioritizing resilience may mean reducing performance. Stakeholders often struggle to know what quality attributes they need.

The way architects used to address these complications was to do a lot of “up-front” design and thinking to try to tease out their requirements, consider different trade-offs, make key architectural decisions, and validate them. Today, however, we need to move a lot faster and adapt to change more effectively than this style allows.

A number of approaches, such as continuous delivery,¹ RCDA,² and continuous architecture,³ attempt to make architecture activities happen less up front and more right through the delivery lifecycle. This allows teams to make important decisions later, when more information is available, and support changes that emerge as the system is created.

The difficulty with continuous architecture and related approaches is knowing if you have done enough architecture work and whether you are spending your ...