虚拟计算环境下的软件自愈机理和方法研究

项目名称： 虚拟计算环境下的软件自愈机理和方法研究

项目编号： No.61300053

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 徐建

作者单位： 南京理工大学

项目金额： 23万元

中文摘要： 本项目旨在关注软件衰退现象，以多服务器虚拟化系统为研究对象，研究主动的、预防性的自愈机制，提高系统性能。探索性能监测、性能分析及性能恢复等技术，重点突破衰退剖面和自愈分析模型构建两个关键问题，并研究上述技术在自愈过程中的综合实施方法，以逐步构造完整的自愈解决方案。以开源的虚拟机监视器XEN平台为载体，构建试验床，开发具备自愈功能的原型系统，开展关键技术实例验证。创新之处在于：1)探索一种多维度的性能监测机理和性能分析方法，克服传统性能监控关注点单一而导致的性能评价片面的缺陷，提高性能分析和量化的准确性；2)探索形式化自愈分析模型，解决自愈决策这一难点问题；3)探索一种多粒度性能恢复方法，通过VMM温重启、同宿主VMs安全状态迁移和跨宿主VM在线迁移，实现用户透明的性能恢复。本课题的研究成果将为研制融合自愈功能的VMM提供可供借鉴的经验，为虚拟化系统性能保障研究提供新的研究方法。

中文关键词： 虚拟机监视器；虚拟化系统；软件衰退；软件自愈；马尔可夫过程

英文摘要： To cope with the phenomena of software in multi-server virtualized system and improve system availability, this project focuses on proactive software rejuvenation mechanism and methods. By studying key technologies, such as performance monitoring, performance analysis, and performance recovery, and solving key issues of constructing aging profiles and rejuvenation analytical models, and accumulating the experience in the application of key technologies, a integrated rejuvenation solution comes into being. Further, taking the open source virtual machine monitor XEN as a test platform, a testbed with rejuvenation is developed to validate the effectiveness of the above key technologies and solution. The main creative contributions include : 1) provides a multi-dimensional performance monitoring and analysis method to improve the analysis precision; 2) put forward a formal analytical model to make rejuvenation decision; 3) provides a transparent multi-level performance recovery method for users, which uses the warm-reboot mechanism for the VMM rejuvenation, the state migration mechanism for the VMs hosted on the same VMM, and the live VM migration mechanism that is a inherent characteristic. The expected findings will be a good reference for developing a VMM with rejuvenation, and provide a new idea to ensure virtua

英文关键词： virtual machine monitor；virtualized system；software aging；software rejuvenation；Markov process