Continuous and Staged, a Choice of CMMI Representations
SANDY SHRUM

What is a CMMI model representation? The answer requires an explanation of the structure of CMMI models. The basic building blocks in every CMMI model are called "process areas." A process area does not describe how an effective process is executed (e.g., entrance and exit criteria, roles of participants, resources). Instead, a process area describes what those using an effective process do (practices) and why they do those things (goals).

To illustrate why an organization might choose one representation over the other, imagine two companies, Foo Toys and Widget Toys. Both companies manufacture software-intensive toys and, until now, have not pursued process improvement.

The management of Foo Toys wants to improve how the company handles risks and integrates product components. Management is happy with how the company’s other processes are operating and so decides to focus on those two process areas only. Foo Toys’ management chooses the continuous representation. Using that representation, Foo Toys will concentrate on only those process areas that relate to risk management and the integration of components. When Foo Toys achieves both the specific goals for a process area and the general goals associated with all levels equal to or less than a particular capability level, it achieves the capability level for that process area. (Goal achievement is determined by a review of the practices associated with the goal.) So, if Foo Toys successfully achieves both the specific goals for product integration and all the capability level 2 and 3 goals, it could be said that, "Foo Toys is level 3 in product integration."

The management of Widget Toys, on the other hand, wants to improve the company’s overall development capability and sees many areas requiring attention. Recognizing the many interdependencies across process areas, Widget Toys’ management chooses the staged representation. Using that representation, Widget Toys will concentrate on the process areas at maturity level 2, thus establishing its project management processes. When Widget Toys performs the practices in these process areas successfully, it also achieves the corresponding goals. When Widget Toys achieves all of the goals of a process area, the process area is satisfied. For Widget Toys to successfully achieve a maturity level, it must satisfy all of the process areas through that level. If Widget Toys satisfies all of the process areas through maturity level 2, it could be said that "Widget Toys is maturity level 2."

By design, the granular information contained in the two CMMI model representations is virtually identical. However, each of the representations provides benefits that will be valued differently by organizations.

In CMMI models, process areas describe key aspects of such processes as configuration management, requirements management, product verification, systems integration, and many others. Let’s examine the two representations in more detail.

Continuous Representation

In the continuous representation of a CMMI model, the summary components are process areas. Within each process area there are specific goals that are implemented by specific practices. Also contained in the continuous representation of a CMMI model are generic goals that are implemented by generic practices.

Specific goals and practices are unique to individual process areas, whereas generic goals and practices apply to multiple process areas. Each practice belongs to only one capability level. To satisfy capability level 2 for a process area, Foo Toys must satisfy the specific goals and level-2 practices for that process area as well as the level-2 generic goals for that same process area.

Staged Representation

In the staged representation, the summary components are maturity levels. Within each maturity level there are process areas, which contain goals, common features, and practices. For Widget Toys, the practices serve as guides on what to implement to achieve the goals of the process area.

In a staged representation of a CMMI model, practices are categorized into common features:

1. Commitment to perform includes practices that ensure that the process is established and will endure. It typically involves establishing organizational policies and leadership.
2. Ability to perform includes practices that establish the necessary conditions for implementing the process completely. It typically involves plans, resources, organizational structures, and training.
3. Activities performed includes practices that directly implement a process. These practices distinguish a process area from others.
4. Directing implementation includes practices that monitor and control the performance of the process. These typically involve placing designated work products of the process under configuration management, monitoring and controlling the performance of the process against the plan, and taking corrective action.
5. Verifying implementation includes practices that ensure compliance with the requirements of the process area. These typically involve reviews and audits.

Capability Levels Versus Maturity Levels

The continuous representation consists of capability levels, while the staged representation consists of maturity levels. The main difference between these two types of levels is the representation they belong to and how they are applied:

• Capability levels, which belong to a continuous representation, apply to an organization’s process-improvement achievement in individual process areas. There are six capability levels, numbered 0 through 5.
• Maturity levels, which belong to a staged representation, apply to an organization’s overall process-improvement achievement using the model. There are five maturity levels, numbered 1 through 5. Each maturity level comprises a set of goals that, when satisfied, improve processes. Maturity levels are measured by the achievement of the goals that apply to a set of process areas.

Table 1: Capability Levels and Maturity Levels

When Widget Toys uses the staged representation, it will evaluate its progress using the same basis as all other organizations that use the same model with the staged representation. Although Widget Toys can pursue process improvement at any pace it wishes, the basis for evaluating its progress will be exactly the same.

Using the staged representation, Widget Toys can identify the maturity levels through which it can evolve to establish a culture of engineering excellence. Each maturity level forms a necessary foundation on which to build the next level.

Using the continuous representation, Foo Toys can produce a capability level profile (i.e., a list of process areas and their corresponding capability levels). Types of capability level profiles include the following:

• An achievement profile represents the current achieved capability level in selected process areas at Foo Toys.
• A target profile represents the capability levels that Foo Toys wishes to achieve.

Maintaining capability level profiles throughout the process-improvement life cycle enables the engineering process group at Foo Toys to demonstrate its progress to management as well as guide its process-improvement activities.

A target profile can reflect the unique needs of the organization (called target staging) or it can reflect the levels used by the staged representation (called equivalent staging). Equivalent staging permits benchmarking of progress among projects, organizations, and other enterprises.

Selecting a Representation

When making the decision about which architectural representation to use for process improvement, Foo Toys and Widget Toys would consider the comparative advantages of each approach as represented in Table 2:

 Table 2: Advantages of Using Each Model Representation

Foo Toys decided to choose the continuous representation because it wanted to focus improvement efforts in two predefined areas.

Widget Toys decided to choose the staged representation because it wanted a clear path to process improvement that provides an easy comparison to competitors that use the same model.

About the Author

Sandy Shrum is a member of the CMMI product-development team and has been a senior writer/editor at the SEI since 1995. Before joining the SEI, she spent eight years with Legent Corp., where she was a senior information developer, a member of a software-development team, and a member of Legent’s IT organization. She has an MA in professional writing from Carnegie Mellon University and a BS in business administration and marketing from Gannon University.