Collaborative Research: SHF: Small: Technical Debt Management in Dynamic and Distributed Systems

庞大且不断发展的软件行业正在迅速采用分布式计算范式。这些范例得到健壮的基于云的基础设施以及较新的编程语言和框架的支持。虽然这带来了许多好处，但也带来了新的问题。最紧迫的问题是，就像传统的单片系统一样，这种现代软件的结构也会因为实现捷径的积累而衰退，这是一种技术债务的形式。这种情况在整个软件行业都很普遍。这种债务在很大程度上不为人注意地增长，直到它影响到软件稳定性、维护成本，并最终影响到公司的财务业绩。但是很少有项目能够偿还债务，因为他们没有办法精确定位它、监控它、可视化它或量化它，特别是对于现代分布式（例如，基于微服务的）系统，或者使用依赖关系隐式和多样化的动态语言构建的系统。软件行业面临的关键问题是：如何有效地度量这些当代软件系统的复杂性和风险，以及如何确定它们的设计债的来源？考虑到这些现代系统的隐式和分布式特性，为传统单片系统定义的现有复杂性度量和反模式不再适用。本项目将通过以下重点来应对这一挑战。第一个重点是创建一个统一的模型来捕获异构的和分布式的实体和关系，使用各种技术，如类型推断。第二部分从经验上探索并定义了一组反模式，这些反模式可以捕获动态类型和分布式系统的体系结构/设计债务，构建在第一部分中创建的统一模型之上。第三个要点探索并正式定义适合这些现代系统的复杂性度量。这些新的反模式和度量将基于软件设计的基本原则来定义，并使用大量开源和工业项目进行经验验证。这个项目在庞大且快速变化的软件行业中处理紧迫和广泛的问题，并将使设计师和管理人员能够可视化他们的设计债务，量化这种债务的影响，并做出明智的决定来提高他们产品的价值。该项目具有交付重大工业影响的潜力：改进驱动美国经济的软件产品和开发过程。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The vast and ever-evolving software industry is rapidly adopting distributed computing paradigms. These paradigms are supported by robust cloud-based infrastructures and by newer programming languages and frameworks. While this brings many benefits, it also brings new problems. The most pressing issue is that—as with traditional monolithic systems—the structure of this modern software also decays from an accumulation of implementation shortcuts, a form of technical debt. This situation is pervasive throughout the software industry. Such debt grows, largely unnoticed, until it impacts software stability, maintenance costs, and, ultimately, the financial performance of a company. But few projects pay down the debt, because they have no way to pinpoint it, monitor it, visualize it, or quantify it, especially for modern distributed (e.g., microservice-based) systems, or systems built using dynamic languages where dependencies are implicit and diverse. The critical issue facing the software industry is: how to effectively measure the complexity and risk of these contemporary software systems and how to pinpoint the sources of their design debt? Given the implicit and distributed nature of these modern systems, existing complexity measures and anti-patterns defined for traditional monolithic systems no longer fit. This project will address this challenge through the following thrusts. The first thrust creates a unified model to capture heterogeneous and distributed entities and relations, using various techniques such as type inference. The second thrust empirically explores and defines a set of anti-patterns that capture architecture/design debts for dynamically typed and distributed systems, building on the unified model created in the first thrust. The third thrust explores and formally defines complexity metrics suitable for these contemporary systems. These new anti-patterns and metrics will be defined based on fundamental principles of software design, and empirically validated using a large number of open source and industrial projects. This project addresses pressing and widespread issues in the enormous and rapidly changing software industry, and will enable designers and managers to visualize their design debt, to quantify the impact of such debt, and to make informed decisions to improve the value of their products. This project has the potential to deliver significant industrial impact: improving the software products and development processes that drive US economy.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.