Overcoming the Challenges of COTS
BOB LANG

Few organizations today would consider building a system entirely from scratch. Use of commercial-off-the-shelf (COTS) products offers the promise of faster time to market and an opportunity to take advantage of commercial investments in technology to increase the functionality and capability of the system.

But the promise of COTS products is too often not realized in practice. Many organizations find that COTS-based systems are difficult and costly to build, support, and maintain. An important factor in this lack of success is that organizations building COTS-intensive solutions tend either to assume that COTS can simply be thrown together or they fall back on the traditional development skills and processes that they are familiar with—skills and processes that experience has shown not to work in the development of a COTS-based system.

The SEI is working to help organizations overcome the challenges of acquiring and fielding COTS-based systems through the Information Technology Solutions Evolution Process (ITSEP).

Background

Building systems from COTS products is an accelerating trend. A vibrant market delivers COTS software products that range from software development environments to operating systems, database management systems, middleware, and, increasingly, business and mission applications. Because COTS products are developed and refined in the competitive marketplace, it is assumed that they will have improved capability, reliability, and functionality compared to custom-built components. COTS products are also expected to integrate with one another, work in a wide range of environments, and support extensions and tailoring to local requirements.

However, reality is often very different. Experience has shown that successfully using COTS products requires a new way of doing business: new skills, knowledge, and abilities, changed roles and responsibilities, and different processes. But no known work to date provides organizations the level of detail, the tool sets, or the training needed to support this new way of doing business. "The changes extend beyond the traditional development effort," says Lisa Brownsword of the SEI. "Implementing a COTS-intensive solution affects the way the organization will conduct business, so all stakeholders need to undergo a culture change, and these changes are just not happening."

About ITSEP

ITSEP evolved from a need expressed by Col David Bentley, USAF and his MITRE support team for a process designed to meet the challenges of acquiring, developing, and fielding COTS-intensive solutions.
ITSEP is more than a way to select a specific product; rather, it provides a way to develop, field, and support a coherent, harmonious solution set composed of one or more products, any required custom code, and any changes required to end-user processes.

ITSEP relies heavily on ongoing work at the SEI-a framework for developing COTS-based systems-and on the Rational Unified Process (RUP) for its disciplined, risk-based spiral development approach.

Figure 1: Balance Spheres of Influence

As shown in Figure 1, the key to ITSEP is the simultaneous definition and tradeoff of four spheres of influence; business processes and stakeholder needs, architecture and design, offerings from the commercial marketplace, and programmatics and risk.

This contrasts with the more traditional approach, which consists of defining the requirements, then formulating an architecture to meet those requirements, and then trying to fit products into that architecture. Many projects that have tried to use the traditional approach for COTS-based systems have failed. "So rather than trying to get the stakeholder needs fully documented as requirements up front," says Brownsword, "we've found that to be successful, you need to keep the stakeholder needs as fluid as you can until you know what available COTS products can do."

Environment to Support COTS-Intensive Solutions

ITSEP creates an environment that supports the iterative definition of the four spheres over time as decisions converge to systematically reduce the trade space. Initially the trade space is large: there is great flexibility for making tradeoffs between the stakeholder needs and business processes, the architecture and design, the offerings of the commercial marketplace, and programmatics and risk. As an improved understanding of the solution is gained across the stakeholder base, the trade space shrinks: the spheres increasingly overlap as fewer decisions remain in any single sphere that can be made without significant impact to the others.

This understanding includes increasingly detailed knowledge of

• the capabilities and limitations of candidate products, their vendors, and the market segment in which they operate
  
• the stakeholder needs for functional and nonfunctional performance of the system
  
• the implications of the COTS products on the end user's business processes as well as the planning necessary to implement any needed business process changes
  
• the architectural alternatives and integration mechanisms that bind the COTS products together

• the cost, schedule and risk associated with implementing and fielding the solution

In ITSEP, the stakeholders make trades between what COTS products can deliver and how the end users will operate using those products.

Current Status

The ITSEP concepts have been used to implement a number of COTS-based systems. To facilitate institutionalization of these concepts across a broader base, ITSEP has been documented to provide detailed instructions for those who want to implement the process. The process designers are now looking for transition partners and opportunities to pilot ITSEP in both government and commercial settings.

For more information, contact—

Customer Relations

Phone
412 / 268-5800

Email
Ceci Albert
Lisa Brownsword

World Wide Web
http://www.sei.cmu.edu/cbs