Collaborative Research: Software and Hardware Support for Efficient Monitoring of Program Behavior

提案编号：CCF-0541080、CCF-0541075和CCF-0541108 Alessandro Orso、Johnathan Cook、Yan Solihin标题：协作研究：对程序行为有效监控的软件和硬件支持由于当今的软件越来越动态，并且在异构环境中运行，因此很难在它们执行的实际上下文之外评估软件系统。这种情况激发了人们对生产环境中软件系统动态行为的有效运行时监控的兴趣。 然而，大多数现有的运行时监控方法有三个局限性：它们是临时的，难以扩展/适应;它们施加太多的开销而不实用;它们仅在一个计算层（硬件，O/S，VM或软件）内定义。这个项目将研究和定义一个通用的，高效的运行时监控方法，以协同的方式利用软件和硬件功能。在这种方法中，将使用提供适当抽象的语言来指定监控任务。 然后，这些规范将被自动分析，监控任务将分布在软件和硬件层，以最大限度地减少性能开销。该项目将通过以下方式提高最新技术水平：（1）为目前因不切实际而被忽视的新监测活动提供硬件和软件支持;（2）开发可供研究界使用的基础设施和工具;（3）通过检测和应对异常、不稳定或恶意软件行为来提高整体软件质量。

Proposal Numbers: CCF-0541080, CCF-0541075, and CCF-0541108Alessandro Orso, Johnathan Cook, Yan SolihinTitle: Collaborative Research: Software and Hardware Support for Efficient Monitoring of Program BehaviorBecause today's software is increasingly dynamic and runs in heterogeneous environments, it is difficult to assess software systems outside the actual context in which they execute. This situation has spurred much interest in efficient runtime monitoring of software systems' dynamic behavior in production environments. Most existing runtime monitoring approaches, however, suffer from three limitations: they are ad-hoc and hard to extend/adapt; they impose too much overhead to be practical; and they are defined within only one computational layer (hardware, O/S, VM, or software). This project will investigate and define a general, efficient runtime monitoring approach that leverages software and hardware capabilities in a synergistic way. In this approach, monitoring tasks will be specified using a language that provides suitable abstractions. These specifications will then be automatically analyzed and the monitoring tasks distributed across software and hardware layers to minimize performance overhead. This project will improve the state of the art by (1) providing hardware and software support for new monitoring activities currently neglected as impractical, (2) developing infrastructure and tools that will be made available to the research community, and (3) improving overall software quality through detection of and reaction to anomalous, erratic, or malicious software behavior.