分布系统与分布对象计算

一、分布系统——发展的必然 分布系统是当前计算机系统发展的主要模式。以计算机网为依托,把多个同时工作的分散计算单元、不同的数据库、不同的操作系统联结成一个整体的布式系统.为多种用户提供一个统一的工作环境。这种分布式系统能够把系统的吞吐能力提高     

Checkpointing in Distributed Computing Systems

This paper examines the performance of synchronous checkpointing in a distributed computing environment with and without load redistribution. Performance models are developed, and optimum checkpoint intervals are determined. The analysis extends earlier work by allowing for multiple nodes, state-dependent checkpoint intervals, and a performance metric which is coupled with failure-free performance and the speedup functions associated with implementation of parallel algorithms. The analytic results for synchronous checkpointing without load redistribution are compared to measurements of a synthetic parallel algorithm with user-level checkpointing. Expressions for the optimum checkpoint intervals for synchronous checkpointing with and without load redistribution are used to determine when load redistribution is advantageous.     

分布式系统下的资源安全性计算

主要从理论上研究保证分布式计算中资源安全性的方法。抽象地将每个进程表示为一个数据对,由给定资源和分配给分布式计算的资源两部分组成。在此基础之上,定义了一系列基于指称语义的数学计算,进而将资源安全性问题抽象为数学上的运算。     

Energy-Efficient Scheduling in Distributed Real-Time Computing Systems

Journal of Computer and Systems Sciences International - Scheduling in multichannel distributed information processing systems is studied. The problem is solved for systems-on-a-chip and is...     

新型分布式计算系统中的异构任务调度框架

随着大数据和机器学习的火热发展,面向机器学习的分布式大数据计算引擎随之兴起.这些系统既可以支持批量的分布式学习,也可以支持流式的增量学习和验证,具有低延迟、高性能的特点.然而,当前的一些主流系统采用了随机的任务调度策略,忽略了节点的性能差异,因此容易导致负载不均和性能下降.同时,对于某些任务,如果资源要求不满足,则会导...     

分布式计算机系统中的负荷平衡算法

本文阐述了分布式计算机系统的负荷平衡问题。在能够达到负荷完全平衡和负荷只能近似地达到平衡的两种情况下，本文提出了时间复杂性为Ｏ（ｎ）的负荷平衡算法。最后还讨论了具有不同处理机速度的负荷平衡问题以及求解平衡的算法。     

一种分布计算系统自适应故障侦测方法

面向高可靠智能应用的分布计算系统,首先提出一组故障侦测服务的QoS度量标准,其次给出一种自适应故障侦测方法.该方法使用一个无需统计行为的高度动态的计算方法,动态地估算心跳消息超时时限,并协商改变心跳消息的发送周期,以适应分布计算系统计算节点和网络状态变化,提高故障侦测服务的QoS.模拟实验表明,该方法能够适应分布计算系统状况的变化,在侦测的实时性和正确性上提供较好的平衡.     

On Load Balancing Approaches for Distributed Object Computing Systems

Distributed object computing systems are widely envisioned to be the desired distributed software development paradigm in the near future due to the higher modularity and the capability of handling machine and operating system heterogeneity. Indeed, enabled by the tremendous advancements in processor and networking technologies, complex operations such as object serialization and data marshalling become very efficient, and thus, distributed object systems are being built for many different applications. As the system scales up (e.g., with larger number of server and client objects, and more machines), a judicious load balancing system is required to efficiently distribute the workload (e.g., the queries, messages/objects passing) among different servers in the system. Several such load balancing schemes are proposed recently in the literature. However, while the rationales and mechanisms employed are dramatically different, the relative strengths and weaknesses of these approaches are unknown, making it difficult for a practitioner to choose an appropriate approach for the problem at hand. In this paper, we describe in detail three representative approaches, which are all practicable, and present a quantitative comparison using a real experimental distributed object computing platform. Among these three approaches, namely, JavaSpaces based, request redirection based, and fuzzy decision based, we find that the fuzzy decision-based algorithm outperforms the other two considerably under a wide range of different practical scenarios.     

分布式系统中计算安全问题的一种解决方案

分布式计算系统安全问题分为两大类：分布式计算的安全和主机的安全。其中恶意主机问题较难解决，该文通过对恶意主机问题的形式化分析，提出了一种基于可信域的终端模型(Trust Domain Based Terminal Model, TDBTM)。该模型通过分布式应用参与制定本地访问控制决策来解决分布式系统对信息的机密性和完整性要求。同时提出了基于该模型以及可信计算思想设计的体系结构(Trust Domain Based Terminal Architecture, TDBTA)，以满足远程证实终端系统可信的需要。结合上述两点内容，该方案有望从根本上解决分布式计算系统中恶意主机带来的安全问题。