6.1 Development of solutions
17 17.6.1 Development of solutions
C orrective and
Solutions to identified causes of recurrent software systems faults are required to:
Eliminate recurrence of the types of faults detected
Contribute to improved efficiency by enabling higher productivity and shorter schedules.
preventive action
Several directions for solutions are commonly taken: ■
Updating relevant procedures. Changes may refer to a spectrum of pro- cedures, from those related to specific stages of software development or maintenance (e.g., changes in style of software comments, changes of contract review procedure in clauses dealing with proposals for small projects) to procedures of a general nature (e.g. changes of employee recruitment procedures, changes of the maximum and minimum number
of participants in a formal design review).
Changes in practices, including updating of relevant work instructions (if any exist).
Shifting to a development tool that is more effective and less prone to the detected faults.
Improvement of reporting methods, including changes in report content, frequency of reporting and reporting tasks. This direction is expected to improve prospects for identification of software system faults and their earlier detection, both resulting in substantial reductions in damages.
Initiatives for training, retraining or updating staff. This direction is taken only in cases when the same training deficiencies are found in sev- eral teams.
It is worth noting that: ■
In many cases, the recommended solutions combine several action items, from one or several directions.
Changing and updating of procedures and work instructions need to be discussed and approved by the bodies assigned to their development and maintenance.
Returning to our example, the “3S” case displays six instances of CAPA: ■
Updating of existing procedure (recommendations (1), (2), (3) and (6))
Replacement of development tools of low efficiency and effectiveness by
Example A: High percentage of severe defects 357 Analysis of software quality metrics for the Development Department of “Peak Performance Software Ltd” identified a high proportion of high-sever-
17.6 Devlopment
ity software defects in the projects completed by two of its six teams. It was also found that the resources these teams required to correct the defects were substantially higher in comparison to other teams.
The analysis was based on documented information related to the two teams’ current as well as former projects, in addition to projects performed by the four “healthy” teams. The findings revealed that the characteristic common to most of the faulty modules was the presence of algorithms of
of
medium to high complexity. Inquiries related to the SQA tools applied by all
solution
the teams revealed a meaningful difference in the number of applications inspected, especially in the analysis and design stages. While the “healthy” teams treated inspection as a more-or-less standard procedure for the more complicated modules, the other teams used inspections rather sparingly. The
and their impl
recommended CAPA solution was to introduce definitions of the module
types requiring inspection within the inspection work instructions. The second example illustrates how a CAPA process can produced unex- pected findings and recommendations.
Example B: Increase in help desk calls that require service at the ement customer’s site The “Perfect Programming Company” regularly operates two help desk teams to support users of its two most popular software products: Team 1
ation
specializes in point of sale (POS) packages, Team 2 in accountancy packages. The Help Desk Unit’s management devised some new quality metrics to sup- port control of the teams’ effectiveness and efficiency. These new metrics emphasized control of services performed at the customer’s site, due to their high costs, and kept track of two variables (metrics), namely, percentage of customer site visits and average technician time per site visit. The quarterly metrics reports for the two help desk teams are shown in Table 17.1.
The report for the fourth quarter ignited a warning signal among com- pany management. Whereas Team 2 showed stability in its performance, a
Table 17.1: Help desk quarterly report – fourth quarter emphasized The HD team
Quality metrics I II III IV Quarter Quarter Quarter Quarter
Team 1 Number of packages installed
POS packages % of customer’s site visits
Average technician time per site
visit (hours)
358 dangerous change in Team 1’s performance was observed. Management was very concerned by the substantial increase in the percentage of customer’s
17 site visits and average technician time per site visit. A corrective and preven-
C tive action team (CAPA team) headed by an SQA unit staff member was orrective and
appointed. The CAPA team held three long meetings devoted to interviewing the help desk team leaders, reviewing a sample of their customer’s site visit reports and examining their detailed monthly statistical reports. The team also observed the help desk teams at work during one afternoon.
The CAPA team discovered that while the previous year was one of con- servative operations for Team 2, displaying some regression in their
preventive action
efficiency, it had been a year of major changes in the operations of Team 1. During the first and second quarters, the team had invested substantial efforts in improvement of the user interface of the POS package and added several helpful error messages. In addition, a revised user manual had been issued. All these improvements were included in the new Version 6.4 that replaced the Version 6.3 of the POS packages that had served the company for the last 20 months. Version 6.4 had been installed by most users during
the third quarter. Analysis of the monthly operations statistics revealed that the currently used quarterly reports were misleading. Unexpectedly, it soon became obvi- ous that in the last two quarters Team 1 had actually achieved a substantial reduction of total help desk efforts, as measured in hours of help desk serv- ice per customer. Moreover, a dramatic decrease in the number of user calls was observed, evidently as a result of the new friendlier and more proficient version of the packages. The increase in average time spent at a customer’s site visit was due to the higher percentage of services now given to new cus- tomers. The CAPA team based its conclusions on a revised, extended quarterly report, presented in Table 17.2. Application of the revised help desk quarterly report for Team 2 figures revealed a constant decrease in the efficiency and effectiveness of the team’s HD services.
Two corrective actions were proposed by the CAPA team: (1) To replace the currently used quarterly report by a more comprehensive
one, based on the lines of Table 17.2. (2) An inquiry into the practices implemented by Team 2 was suggested to
achieve a substantial improvement in the team’s performance.