A Hierarchical Framework for Organizing a Software Development Process

• 成果类型: Article
• 署名作者: Iravani, Foad; Dasu, Sriram; Ahmadi, Reza
• 署名单位: University of Washington; University of Washington Seattle; University of Southern California; University of California System; University of California Los Angeles
• 刊物名称: OPERATIONS RESEARCH
• ISSN/ISSBN: 0030-364X
• DOI: 10.1287/opre.1120.1111
• 发表日期: 2012
• 页码: 1310-1322
• 关键词: PRODUCT DEVELOPMENT; lagrangian-relaxation; maintenance; allocation; projects; MODEL

摘要

Every year, companies that produce commercial tax preparation software struggle with thousands of state and federal changes to tax laws and forms. Three competitors dominate the market with its short selling season, and release delays slash profits. Tax authorities issue updates August December, and all changes must be processed and incorporated before year end. Systematic resource allocation and process manageinent are crucial yet problematic due to the volume and complexity of changes, brief production time frame, and feedback loops for bug resolution. A leading tax software provider asked us to propose systematic approaches for managing process flow and staffing development stages with the goal of releasing the new version on time at minimum cost. To that end, we developed deterministic models that partitioned tax forms into development groups and determined staffing levels for each group. Partitioning forms into groups simplified workflow management and staffing decisions. To provide a range of resource configurations, we used two modeling approaches. Numerical experiments showed that our models capture the salient features of the process and that our heuristics perform well. Implementing our models reduced company overtime hours by 31% and total workforce cost by 13%. Subject classifications: product development; software development; tax software; workforce management; resource allocation grouping index; integer programming. Area of review: OR Practice. History: Received February 2011; revisions received October 2011, May 2012; accepted June 2012. Published online in Articles in Advance November 20, 2012.