[论文解读] Measuring Cost of Quality (CoQ) on SDLC Projects is Indispensible for Effective Software Quality Assurance

Parvez Mahmood Khan, M. M. Sufyan Beg|arXiv (Cornell University)|May 19, 2014

Software Engineering Research参考文献 12被引用 4

一句话总结

本文主张在软件开发生命周期（SDLC）项目中衡量质量成本（CoQ）对于有效的软件质量保证至关重要。通过将制造业中传统的CoQ框架——预防成本、评估成本和故障成本——适配至软件工程领域，作者证明，追踪这些成本可实现缺陷的早期发现，减少交付后的故障，并提升项目绩效，尤其对中小型软件企业而言益处显著。

ABSTRACT

It is well known fact that was phrased by famous quality scholar P.B. Crosby that it is always cheaper to do the job right the first time. However, this statement must be reconsidered with respect to software development projects, because the concept of quality and associated costs measurements in software engineering discipline is not as matured as in manufacturing and other fields of the industry. Post delivery defects (i.e. software bugs) are very common and integral part of software industry. While the process of measuring and classifying quality cost components is visible, obvious and institutionalized in manufacturing industry, it is still evolving in software industry. In addition to this, the recommendations of British standard BS-6143-2:1990 for classifying quality-related costs into prevention costs, appraisal costs, and failure costs have been successfully adopted by many industries, by identifying the activities carried out within each of these categories, and measuring the costs connected with them, software industry has a long-way to go to have the same level of adoption and institutionalization of cost of quality measurements and visibility. Cost of Quality for software isn't the price of creating a quality software product or IT-service. It's actually the cost of NOT creating a quality software product or IT-service. The chronic affliction of majority of software development projects that are frequently found bleeding with cost overruns, schedule slippage, scope creep and poor quality of deliverables in the global IT industry, was the trigger for this research work. Lessons learnt from this study offer valuable prescriptive guidance for small and medium software businesses, who can benefit from this study by applying the same for their quality improvement initiatives using CoQ-metric, to enhance the capability and maturity of their SDLC-project performance.

研究动机与目标

• 解决尽管在制造业中已相当成熟，但在软件工程中质量成本（CoQ）测量仍缺乏制度化的问题。
• 强调软件项目中质量问题带来的财务与运营后果，包括成本超支、进度延迟和范围蔓延。
• 为中小型软件企业提供一个可操作的框架，以实施CoQ指标来提升质量。
• 倡导在软件开发中采用BS-6143-2:1990质量成本分类模型——即预防、评估和故障成本。
• 证明质量的真实成本并非创造质量的成本，而是未采取行动所导致的成本。

提出的方法

• 将BS-6143-2:1990框架适配至软件质量成本分类，划分为预防、评估和故障成本。
• 分析真实世界的SDLC项目，识别并量化开发生命周期中各阶段的CoQ组成部分。
• 采用定性与案例研究分析方法，将质量相关活动映射至相应的成本类别。
• 提出一种结构化方法，用于追踪和报告CoQ指标，以支持软件项目中的决策制定。
• 强调利用CoQ数据指导质量实践投资，降低发布后的缺陷修复成本。
• 借鉴制造业质量管理体系的经验，应用于软件开发，以提升流程成熟度。

实验结果

研究问题

• RQ1 尽管CoQ在制造业中已被证明具有显著价值，为何在软件开发项目中其测量仍未能充分制度化？
• RQ2 如何有效将传统的CoQ框架（预防、评估、故障）应用于SDLC项目？
• RQ3 在软件项目中不测量CoQ会产生哪些财务与运营影响？
• RQ4 中小型软件企业如何通过在SDLC流程中实施CoQ指标获益？
• RQ5 早期缺陷检测与预防在降低长期软件质量成本方面发挥何种作用？

主要发现

• 未创建高质量软件所导致的成本——即故障成本——远超预防与评估活动的投资成本。
• 交付后的缺陷在软件项目中普遍存在且代价高昂，反映出CoQ测量与管理的薄弱。
• 尽管在其他行业已证明成功，BS-6143-2:1990 CoQ分类模型在软件工程中的采用仍有限。
• 测量CoQ使组织能够识别效率低下之处，减少返工，并提升项目在成本、进度和范围方面的可预测性。
• 中小型软件企业通过实施CoQ指标来改进质量，可显著提升SDLC绩效。
• 本研究证实，首次就正确完成软件开发在经济上是可行且必要的，尤其当故障成本被量化并主动管理时。