Traceability Based Dynamic Metrics as a Tool for Evaluating Partitioning Decisions in Object-Oriented Architecture

本文研究了一种评估面向对象软件体系结构预测性能特征的新方法。由于场景代表了系统预期的运行时行为，因此可以在设计的早期阶段使用它们来定量预测该行为的各个方面。所提出的方法可以比其他现有方法（如模拟和可执行设计）更早应用，并且不需要特殊的分析技能。早期评估是一种重要的风险缓解技术，它可以对构建分布式系统的任务做出重大贡献，并对其按要求执行的能力有更大程度的信心。该研究包括正式定义一组基于场景的度量和相应的系统范围动态度量模型。定义了强类型可追溯性方案，以支持性能需求、场景和体系结构组件之间关系的定义和自动查询。在典型使用模式的上下文中指定场景调用率，并调查场景之间的关系，以确定何时应该将性能需求从一个场景传播到另一个场景。开发了一个支持工具，用于捕获度量，并根据从各种分布式系统获取的运行时和模拟测量来验证它们。

This research investigates a new method of assessing predicted performancecharacteristics of Object-Oriented software architectures. As scenariosrepresent the anticipated runtime behavior of a system, they can be used ata very early stage in the design to quantitatively predict aspects of thatbehavior. The proposed approach can be applied earlier than other existingmethods such as simulations and executable designs, and does not requirespecial analytical skills. Early assessment is an important risk mitigationtechnique, that can contribute significantly to the task of building adistributed system with a greater degree of confidence in its ability toperform as required.The research involves formally defining a set of scenario-based metrics anda corresponding system-wide dynamic metrics model. A strongly typedtraceability scheme is defined to support the definition and automatedquerying of relationships between performance requirements, scenarios, andarchitectural components. Scenario invocation rates are specified withinthe context of typical usage patterns, and the relationship betweenscenarios is investigated to determine when performance requirements shouldbe propagated from one scenario to another. A supporting tool is developedand used to capture metrics and validate them against runtime and simulationmeasurements taken from a variety of distributed systems.