NEWS AT SEI
This library item is related to the following area(s) of work:Software Product Lines
This article was originally published in News at SEI on: September 1, 1999
Software developed as a product line promises to be a dominant development paradigm for the new century, one that the Department of Defense (DoD) can leverage when acquiring software-intensive systems. This article discusses the advantages of product lines, uncovers some of their pitfalls, and shows by example the kinds of successes that the organizations can enjoy.
Imagine turning out a 1.5 million-line Ada command and control system for a Navy frigate warship. The system is hard real-time, fault-tolerant, and highly distributed, running on 70 separate processors on 30 different local area network nodes scattered all over the ship. It must interface with radar and other sensors, missile and torpedo launchers, and other complicated devices. The human–computer interface is complex and highly demanding. In this application, quality is everything: The system must be robust, be reliable, and avoid a host of performance, distribution, communication, and other errors.
Now suppose that you have not one of these systems to build but several. Your marketing department has succeeded beyond your wildest dreams. Navies from all over the world have ordered your command and control system. Now, your software must run on almost a dozen different ship classes including a submarine, and the systems are drastically separate in numerous ways:
For all their differences, however, quality remains crucial in every system.
Suppose you are the manager for this megaproject. Do you panic? Do you resign? Run to a third-world country? What if you could produce each one of the systems for a fraction of the cost and in a fraction of the time that one would normally expect? And what if you could do it so that quality was improved and reliability and customer satisfaction increased with each new system? What if creating a new ship system were merely a matter of combining large, easily tailorable components under the auspices of a software architecture that was generic across the entire domain (in this case, of shipboard command and control systems)?
Is this a fantasy? No, it is not. It is the story of CelsiusTech Systems AB, a long-time European defense contractor. In the 1980s, CelsiusTech was confronted with the dilemma outlined above. It had to build two large command and control systems, each larger than anything that the company had attempted before, and it had barely enough resources to build one. Because necessity stimulates invention (and determination implements it), CelsiusTech realized that its only hope was to build both systems at once using the same assets and resources. And in a visionary stroke, CelsiusTech knew that its future lay in exploiting these assets for not only the first two systems but also for a whole family of products that the company hoped and expected would follow.
In short, CelsiusTech launched a software product line. A product line is a set of products that together correspond to a particular market segment or fulfill a particular mission. Product lines promise to become the dominating production-software paradigm of the new century. Product flexibility is the new anthem of the marketplace, and product lines fulfill the promise of tailor-made systems built specifically for the needs of particular customers or customer groups. What makes product lines succeed from the vendor’s (and developer’s) point of view is that the commonalities shared by the products can be exploited to achieve economies of production.
Product lines are nothing new in manufacturing. Boeing builds one, so does Ford, IBM, and even McDonald’s. Each of these companies exploits commonality in different ways. Boeing, for example, developed the 757 and 767 transports in tandem, and the parts lists of these two decidedly different aircraft overlap by about 60 percent.
But software product lines based on interproduct commonality are a relatively new concept, and the community is discovering that this path to success contains more than its share of pitfalls.
The Software Engineering Institute has a technical program to identify and promulgate the best practices for product line production and help organizations negotiate the hurdles of adopting a product line approach. The Product Line Systems Program focuses on the following essential technology areas for product line production:
The result is a technology infrastructure that can produce large custom systems quickly and reliably by checking out components from the asset repository, tailoring those components for their particular application (CelsiusTech uses compile-time parameters to instantiate different versions of a component), and beginning the integrate-and-test cycle as in normal system development.
Once the product line repository is established, consider what is saved each time a product is ordered:
Product lines enhance quality. Each new system takes advantage of all of the defect elimination in its forebearers; both developer and customer confidence rise with each new instantiation. The more complicated the system, the higher the payoff for solving the vexing performance, distribution, reliability, and other engineering issues only once for the entire family.
Clearly, product lines benefit the developing organization, but they also benefit acquirers of systems as well. Acquiring a family of related systems using a product line acquisition approach (as opposed to acquiring each system separately and independently) clearly falls within the realm of Department of Defense (DoD) reuse initiatives and policies, and it promises to accrue significant benefits for the DoD, including
It takes a certain maturity in the developing organization to successfully field a product line. Technology is not the only barrier to successful product line adoption. Experiences observed by the Product Line Systems Program show that it is equally vital to master organization, process, and business issues.
For instance, traditional organizational structures that have one business unit per product are generally not appropriate for product lines. Who will build and maintain the core reusable assets—the architecture, the reusable components, and so forth? If these assets are under the control of a business unit associated with one product or one large customer, the assets may evolve to serve that business unit, that product, and that customer to the exclusion of the others. On the other hand, to establish a separate business unit to work on the core assets but be divorced from working on individual products carries the danger that this unit will produce assets that emphasize beauty and elegance over practicality and utility. In either case, producing and managing the reusable assets means establishing processes to make the assets satisfy the needs of all of the business units that use them. This is a crucial role that requires staff skilled in abstraction, design, negotiation, and creative problem solving. The question of funding the core asset development is crucial.
Customer management becomes an important product line function. Customers interact with a product line organization in a different way. Marketers can no longer agree to anything customers want but must instead nudge customers to set their requirements so that they can be fulfilled by a version of the product line within the planned scope of variation.
Contrary to intuition, this often makes the customer much happier than before. Under the new paradigm, the marketer can point to specific requirements that would put the customer’s new system outside the scope of the product line, which would increase the cost and delivery time, lower the system’s reliability, and keep that customer out of a community of customers to which the vendor pays a lot of attention. Thus, the customer could clearly (and probably for the first time) see the real cost of those "special" requirements and make an informed decision about their real value. If the customer decides that a variant of the “standard” or product line system will suffice, so much the better. If not, the customer can still order a system to satisfy particular requirements but with a better idea of where the risks may be hiding.
The customer community should not be underestimated. In CelsiusTech’s case, the naval customers around the world banded together to form a users’ group. They did this in their self-interest—to provide a forum in which they could jointly derive new requirements for their evolving systems and drive CelsiusTech to supply new systems more economically than it otherwise might. But it does not take much to realize how beneficial this is to CelsiusTech as well: Its customer base is jointly defining the next generation of products and is effectively buying in to CelsiusTech’s approach, thus guaranteeing the vitality of the product line for years to come.
The users’ group provides a clear lesson for DoD acquisitions: It pays to collaborate (or at least communicate) when it comes to commissioning or purchasing similar systems.
The successful transition to product line technology requires a careful blend of technology, process, organization, and business factors improvement. The Product Line Systems Program is attempting to codify these practices and understand how they vary with the type of organization involved and the kind of systems being built. Through a series of workshops, case studies, and collaborative engagements, the SEI is helping to build a community of organizations interested in moving to a product line approach for their software products.
We believe that product lines will be the predominant software paradigm at the beginning of the new century. The history of programming can be viewed as an upward spiral in which the abstractions manifested by components have grown larger and more application meaningful, with resulting increases in the reuse and applicability of those components. From subroutines in the 1960s to modules in the 1970s to objects in the 1980s to component-based systems in the 1990s, software product lines will perpetuate the upward spiral by accomplishing previously unheard-of levels of reuse from system to system.
If the pitfalls are successfully negotiated, the result is an enviable capacity to deliver extremely large systems on time and within budget.
You can download the full report that details the CelsiusTech product line case study, which includes data about the company’s dramatic results in time-to-market, levels of reuse, and required staffing, at
Paul Clements is a senior member of the technical staff at the Software Engineering Institute. A graduate of the University of North Carolina and the University of Texas, he is a project leader in the SEI’s Product Line Systems Program. His work includes collaborating with organizations that are launching product line efforts. He is a co-creator of the Software Architecture Analysis Method (SAAM), which allows organizations to evaluate architectures for fitness of purpose. He and others are working on an extension to SAAM, which will allow analysis of quality attribute tradeoffs at the architectural level. He is co-author of Software Architecture in Practice (Addison-Wesley-Longman, 1998) and more than three dozen papers and articles about software engineering.
For more information
Please tell us what you
think with this short
(< 5 minute) survey.