High Performance Parallel and Distributed Systems: Resource Management and Middleware

以合理的成本获得计算系统和通信网络正在增加并行和分布式系统的使用。分布式系统，特别是基于互联网的系统正在成为现代社会日常生活的一部分，并跨越多个领域，包括(I)按需向企业和科学应用提供计算和存储资源的通用云，(Ii)基于无线传感器网络的系统，如智能家居和患者监护系统，以及(Iii)用于管理智能关键基础设施的系统，如配备传感器的桥梁和风力涡轮机。对资源的有效管理对于利用这种分布式系统的潜在能力至关重要。 在这三个新兴和相互关联的主题中，提出了设计资源管理技术的研究。在应用程序设计、操作系统和部署在外地的系统的中间件方面需要研究的技术。这项研究的结果将推进最先进的技术，并有助于实现系统用户所需的服务质量、服务提供商的足够收入以及公众使用的安全和经济维护的基础设施。培训学生是该计划不可或缺的一部分，该研究计划将有助于将加拿大确立为全球市场的领先者。 申请人从事资源管理研究已有25年。他的工作包括对多程序并行系统上的并行作业调度的开创性研究，以及处理大量异类资源的新技术，以及容易出错的用户对云上请求运行时间的估计。他的贡献通过最佳论文/演示文稿奖和他被选为IEEE计算机协会杰出参观者而得到认可。申请人在过去六年中的活动体现在9篇期刊论文、3篇书籍投稿、34篇参考会议论文、6篇特邀演讲、在各种会议计划和指导委员会的服务以及2次国际会议的主持。

The availability of computing systems and communication networks at a reasonable cost is increasing the use of parallel and distributed systems. Distributed systems in general and internet-based systems in particular are becoming part of the daily life in a modern society and span domains that include (i) general purpose clouds that provide computing and storage resources on demand to enterprise and scientific applications, (ii) wireless sensor network based systems such as smart homes and patient monitoring systems and (iii) systems for management of smart critical infrastructures such as sensor equipped bridges and wind turbines. Effective management of resources is crucial for harnessing the underlying power of such distributed systems. Research is proposed for devising techniques for resource management in these three emerging and inter-related topics. Techniques that will be investigated are needed in the context of application design, operating systems and middleware for systems deployed in the field. The results of this research will advance the state of the art and contribute to the achievement of the desired quality of service for system users, adequate revenues for the service providers as well as safe and economically maintained infrastructures used by the public. Training of students is an integral part of the plan and the research program will aid in establishing Canada as a leader in the global market. The applicant is involved in research on resource management for 25 years. His work includes pioneering research on parallel job scheduling on multiprogrammed parallel systems and novel techniques for handling a large heterogeneous set of resources and error-prone user estimates of request run times on clouds. His contributions have been recognized through best paper/presentation awards and his selection as a Distinguished Visitor for the IEEE Computer Society. The applicant's activities in the past six years are reflected in 9 journal papers, 3 contributions to books, 34 refereed conference papers, 6 invited presentations, service in various program and steering committees for conferences and chairing of 2 international conferences.