APPLICATION: Software Process Improvement Program for Boeing Wichita Site -- With 22,250 employees on site at The Boeing Company's Wichita Division, Boeing's Wichita Information Systems (I/S) team faced quite a challenge in standardizing the organization's software engineering practices. Faced and conquered the challenge, that is. In October 1997, the Wichita division received a Software Engineering Institute Capability Maturity Model (SEI/CMM) Level 3 rating. Boeing Wichita is one of the largest and most diverse sites to achieve this rating.>/P>

Located in Wichita, Kan., a city with a rich history of aviation, the unit manufactures commercial jetliner assemblies and components, and is the state's largest private employer. Boeing Wichita operations are divided between commercial and military work: Boeing Commercial Airplane Group, Wichita Division; and Boeing Defense and Space Group, Product Support Division. Wichita produces part of every commercial jetliner, including 75% of the 737, the best-selling jetliner in history. Commercial work in Wichita also includes modification and maintenance, and a special conversion program where passenger 747s are modified into cargo airplanes.

The Wichita Division's Level 3 rating follows a December 1995 attainment of a Level 2 rating. A Level 4 rating is on the current agenda.

"In general, we are not much different than most businesses in today's age," said William Hammers, of the Boeing Wichita Software Engineering Process Group (SEPG). "We were looking to improve our process, to make it more cost-effective, to achieve better customer satisfaction, to add value to our product and to improve the quality and efficiency of our process."

The immediate benefit realized by Boeing in the Wichita site's efforts to improve its software processes is the achievement of a SEI/CMM Level 3 rating of the entire Wichita site. This is critical to Boeing's business since many federal and military programs require a Level 3 rating for contracted software work, said John Huffman, senior principal engineer, information systems.

"Of course, the Capability Maturity Model (CMM) was our primary tool for the software process improvement activities," Huffman said. "We used the Productivity Plus (P+) software development methodology for DMR as our standard life-cycle methodology."

"As far as productivity tools go, we mostly use our own 'home-grown' tools. Of course, we use popular commercial software (like Microsoft Office) for many of analysis and reporting utilities," Huffman said.

Prior to this effort, software engineering at the Wichita site was approached in mostly an ad hoc fashion. Individual projects would define their own methods for software project management, software quality assurance, software configuration management, peer reviews, requirements management, measures, metrics, and the like.

Development team members at the site attribute the accomplishment to "both the commitment and tenacity of the 400-plus team members over the last four years, and their innovative approach top overall software process improvement."

Through use of the SEI/CMM framework for software process improvement and adoption of Productivity Plus as Boeing's standard software development project methodology, the Wichita site has established reliable methods and processes for software that truly reflects the practices associated with an engineering discipline, according to Huffman. Productivity Plus is a collaborative approach to software projects that time project management activities and software development activities through a management-by-deliverable philosophy.

"The CMM was a road map we believed we could use to help us," Hammers continued. "It's an initiative the entire organization supports."

Software system development at the Wichita site is now a full-fledged engineering task requiring project team members to have full depth of software and systems engineering skills. The increased reliability of the software engineering practices has improved the reliability of the software systems produced, and has thus had a positive impact on aircraft products produced.

Normally, Boeing Wichita's I/S developers work in distinct teams to develop specific software systems. "Here, we have brought along the whole group rather than a single project. That's the difference," Hammers said.

In defining, piloting and deploying reliable software engineering methods and processes throughout the Wichita site, the Wichita Information Systems team has created an infrastructure in which quantitative software improvement can be measured, according to Huffman. The CMM effort has reduced pre-release and post-release software defects, reduced software system cycle time, allows for better estimates for deliverables produced through its software projects and has produced high levels of customer satisfaction for software products and services.

With the great number of employees on site, officials say that a primary feat in the CMM effort was the extraordinary task of "getting people to change." Susan Casper, SEPG project manager at Boeing's Wichita site, recalled that the greatest challenges of the project were "attitudinal and behavioral." Added Hammers: "It's not to say these people didn't have good practices, they were just very individualized."

John Hostetler, SEPG member, added that the CMM model did indeed give the Wichita site a road map, and the team concentrated on improving the process in specific areas. "Now with resources at a premium, there is an infrastructure in place," Hostetler continued. "The process was about trying to come up with an infrastructure that supports customers. We work at helping them to achieve their goals."

The major technology challenge faced during the process improvement effort was integrating the various in-house utilities used to plan, manage and control software projects, and the updating of component CASE tools to reflect new deliverable standards adopted through SEI/CMM and Productivity Plus. Additionally, some online decision support is necessary to help new users navigate the various productivity tools used for the deliverable management and configuration management pieces of the process. The SEPG teamed with members of the Wichita Information Systems' Manufacturing Decision Support Systems group to create knowledge-based productivity tool to aid analysts in the creation of project deliverables. Also several sitewide analysis utilities were created to measure and track the measure from all software projects.

Besides tracking pre-release and post-release defects, customer satisfaction rating, system cycle time, variance to plan and product size across all projects, certain quality metrics have been devised, including pre-release defect removal rate, effectiveness and efficiency, variance to schedule, cost per element created/modified and reuse rate, according to Huffman. Many more will be identified for the statistical process control requirements for Level 4 and Level 5 of the SEI/CMM.

- Denise M. Dubie