ITR: Architectural Support for Programming-in-the-Small-and-Many

在过去的几十年里，软件研究人员和实践者提出了各种方法、技术和工具，用于开发针对传统工作站环境的更大、更复杂的系统。这些努力的结果具有许多共同的特征：系统大小和复杂性，跨桌面平台的可能分布，在实现之前关注建模和分析，伴随的开发环境，明确的软件体系结构，等等。由此产生的软件开发范例被称为大型编程（PitL）。随着小型、廉价、异构、资源受限、可能是嵌入式、高度分布式和高度移动的计算平台的激增，出现了一系列新的挑战。虽然许多单独的挑战可能与PitL所解决的问题相似，但它们的组合和整体的新颖性可以更恰当地描述为“小而多的编程”（Prism）。拟议的项目旨在解决棱镜计划的挑战。这个项目的重点是在软件应用工程的层次上。也就是说，该项目将利用较低级软件技术的进步（操作系统、优化编译、（无线）网络和数据压缩），因为它解决了开发理解、原则、框架、方法、机制和工具的目标，这些目标将支持高效、可扩展、可靠、健壮，以及最重要的是，在Prism设置中具有高度适应性的软件应用程序。该项目将侧重于四个关键目标：1。开发有效的设计方法，配套的软件模型，灵活的，增量的软件分析技术；2 .开发技术，使设计方法能够在异构、高度分布、移动和资源受限的环境中实现；3 .开发执行时解决方案，以支持（持续的）动态系统演进、（重新）部署、移动性和优雅的降级；通过一系列实验来量化其关键方面，以及通过在三个特定问题领域中的应用来验证该方法。该项目将建立在新兴的软件体系结构研究的基础上，为小而多的软件开发提供一个全面的方法。特别地，它将调查软件架构风格（跨系统重复出现的结构、行为和交互模式）在为设计方法和增量架构分析提供基础方面的潜在角色和影响。基于组件的框架和软件连接器（体现组件交互的显式体系结构抽象）将构成体系结构实现和执行解决方案的可能基础。由于假设Prism设置中的所有应用程序都具有相同的建模和实现需求是不现实的，因此该项目将提供可配置的体系结构风格和实现框架，并提供将这些框架实例化为特定风格和基于风格的实现/执行平台的指导方针。

ABSTRACT0312780Nenad MedvidovicOver the past several decades software researchers and practitioners have proposed various approaches, techniques, and tools for developing ever larger, more complex systems targeted at the traditional workstation environment. The results of these efforts have shared a number of traits: system size and complexity, possible distribution across desktop platforms, focus on modeling and analysis before implementation, accompanying development environments, explicit software architectures, and so forth. The resulting software development paradigm has been referred to as programming-in-the-large (PitL). A new set of challenges has arisen with the proliferation of small, inexpensive, heterogeneous, resource-constrained, possibly embedded, highly-distributed, and highly-mobile computing platforms. While a number of the individual challenges may bear similarity to those addressed by PitL, their combination and overall novelty can be described more appropriately as programming-in-the-small-and-many (Prism).The proposed project intends to address the challenges of Prism. The project's focus is at the level ofsoftware application engineering. That is, the project will utilize advances in lower level software technologies (operating systems, optimized compilation, (wireless) networking, and data compression) as itaddresses the goal of developing understandings, principles, frameworks, methods, mechanisms, and toolsthat will support efficient, scalable, reliable, robust, and, above all, highly adaptable software applications in the Prism setting. The project will focus on four key objectives:1. develop an effective design methodology, accompanying software models, and flexible, incremental softwareanalysis techniques;2. develop technology to carry the design approach forward to implementation in a heterogeneous, highlydistributed, mobile, and resource constrained setting;3. develop execution-time solutions in support of (continuous) dynamic system evolution, (re)deployment,mobility, and graceful degradation;4. validate the approach through series of experiments to quantify its key aspects, as well as through applications in three specific problem domains.The project will build upon the emerging body of research on software architectures in providing a comprehensive methodology for software development in-the-small-and-many. In particular, it will investigate the potential role and impact of software architectural styles (recurring structural, behavioral, and interaction patterns across systems) in providing a basis for the design methodology and incremental architectural analysis. Component-based frameworks and software connectors (explicit architectural abstractions embodying component interactions) will form the likely basis of the architecture implementation and execution solutions. Since it is unrealistic to assume that all applications in the Prism setting will have the same modeling and implementation needs, the project will provide configurable architectural style and implementation frameworks, with guidelines for instantiating these frameworks into specific styles and style-based implementation/execution platforms.