AJORAmerican Journal of Operations Research2160-8830Scientific Research Publishing10.4236/ajor.2012.21004AJOR-17827ArticlesPhysics&Mathematics Empirical Analysis for High Quality Software Development aomiHonda1*ShigeruYamada2*Department of Social Management Engineering, Graduate School of Engineering, Tottori University, Tottori, JapanIT Software Operations Unit, NEC Corporation, Tokyo, Japan* E-mail:n-honda@ay.jp.nec.com(AH);yamada@sse.tottori-u.ac.jp(SY);1403201202013642January 18, 2012February 20, 2012 February 29, 2012© Copyright 2014 by authors and Scientific Research Publishing Inc. 2014This work is licensed under the Creative Commons Attribution International License (CC BY). http://creativecommons.org/licenses/by/4.0/

It remains important for a development organization to configure a software process that enables it to develop software products with the least possible number of defects after shipment. A development organization of CMMI level 5 has, over three years, been strived to improve those software projects that had been noted as having many defects after shipment. In this paper, we discuss our organization’s improvement (Kaizen) activities, to analyze the important matters of software process to be considered when developing a software product with the least possible number of defects after shipment. Our results are identified by three important points; 1) early ensured quality by defect detection during design or code review; 2) quality assurance for both process quality and product one; and 3) quantitative management by which data of the appropriate resolution can be collected at an appropriate timing.

Software Process; Review; Quantitative Management; Process Quality; Product Quality1. Introduction

Software must take on the role of managing the systems that support the infrastructures on which society depends. Given this degree of dependence, society has the right to demand the development of software products that are shipped with as few defects after shipment as possible. The venders that supply software products have difficulty in satisfying this requirement, however [1].

In response to the above, several different development and management techniques for improving the quality of software have been proposed. Among them, CMMI is widely applied worldwide. Unfortunately, even though a high level of CMMI is attained, satisfactory quality of software cannot always be obtained [2]. In fact, some development organizations with CMMI level 5 exhibit many defects after shipment.

The CMMI level 5 organization has spent three years working on improving software products that had exhibited many defects after shipment, based on the results of benchmarking other organizations with the same level, as part of its activities to reduce the number of defects after shipment. This paper describes the results. Based on the results, this paper discusses the software process conditions required to develop software products that have few defects after shipment. In this paper, a CMMI level 5 organization was selected as a case study because the loss of process areas itself has little effect on the number of defects after shipment, due to well-established software process of the organization.

2. Overview of Organization

This paper presents two CMMI level 5 organizations as a case study. These are called Organizations A and B. Organization A’s product exhibits a smaller number of defects after shipment than that produced by Organization B. Organization B planned improvement (Kaizen) measures by benchmarking A, and conducted Kaizen activeties. This paper outlines organizations A and B, and discusses the number of defects after shipment of the products produced by the organizations.

2.1. Organizations Characteristic

Organizations A and B belong to the same company and develop general-purpose IT-related software products in their different business areas. These organizations’ customer groups are basically in the same enterprise area. Organization A has almost the same shipment volume, development amount, and number of engineers as B. The two organizations applied almost the same software process and accomplished CMMI level of 5 in the early 2000s. Usually, they would apply V-shaped model and implement V&V. Their development techniques, such as those for design and testing, are almost the same. Both adopt the “Quality accounting” internally created and developed as a quality management method [3].

2.2. Number of Defects after Shipment

Figure 1 shows the number of defects after shipment in the products produced by Organization A and those of Organization B (see Organizations A and B before Kaizen). On the basis of the mean number of defects after shipment, assuming that the number of defects in the products of Organization A is 100, the number of defects in those of Organization B will reach 283. It can be shown that the number of defects after shipment by B is almost three times bigger than that appearing in the products of A. Organization B clearly had a problem with the number of defects after shipment in its products. This triggered Kaizen activities in Organization B.

ReferencesW. S. Humphrey, “Winning with Software: An Executive Strategy,” Pearson Education, Inc., New Jersey, 2002.N. Honda, “Beyond CMMI Level 5—Comparative Analysis of Two CMMI Level 5 Organizations,” Software Quality Professional, Vol. 11, No. 4, 2009, pp. 4- 12.N. Honda, “Software Quality Accounting—Quality Assurance Technology That Supports High Quality Software Development at NEC,” JUSE Press, Tokyo, 2010.B. Boehm, “Software Engineering Economics,” Prentice- Hall, Englewood Cliffs, 1981.T. Gilb and D. Graham, “Software Inspection,” Addison-Wesley, Reading, 1993.SQuBOK Study Team, “SQuBOK Guide—Software Quality Body of Knowledge (in Japanese),” Ohmsya, Tokyo, 1997.JUSE, “Research Report for CMMI Level 5 Organizations,” Tokyo, 2009. http://www.juse.or.jp/software/86/attachs/kikaku_1.pdf