Resources Allocation Schemas for Web Information Monitoring

The web is an extremely dynamic world where information is updated even every second. A web information monitoring system fetches information from the web continuously and finds changes by comparing two versions of the same page. The updating of a specific web page is modeled as a Poisson process with parameter to indicate the change frequency. As the amount of computing resources is limited, it is necessary to find some policies for reducing the overall change-detection time. Different allocation schemas are evaluated experimentally to find out which one is the most suitable for the web information monitoring problem. The experimental data shows the runtime characteristics of the overall system performance and the relationship to the total amount of resources.     

相异性容错软件故障模型

针对容错计算领域中由于引入了相异性设计概念,导致传统软件失效模式难以应用的问题,提出了一个用于分析相异性容错软件故障的模型。模型根据各类软件设计故障对系统可靠性的影响,精确定义了相异性容错软件系统中故障的范畴及其划分、以及与系统失效之间的关联关系,在此基础上建立了一个基于功能的故障模型,以描述相异性软件系统的可靠性关系。通过一个通用相异性软件系统为实例,应用故障模型分析了系统可靠性关联关系,建立了与之相对应的故障模式及其可靠性模型,验证了故障模型的实用性。     

Building a Portable File System for Heterogeneous Clusters

Existing in-kernel distributed file systems cannot cope with the higher requirements in wellequipped cluster environments, especially when the system becomes larger and inevitably heterogeneous.TH-CluFS is a cluster file system designed for large heterogeneous systems. TH-CluFS is implemented completely in the user space by emulating the network file system (NFS) V2 server, and is easily portable to other portable operating system interface (POSIX)-compliant platforms with application programming/binary interface API/ABI compliance. In addition, TH-CluFS uses a serverless architecture which flexibly distributes data at file granularity and achieves a consistent file system view from distributed metadata. The global cache makes full use of the aggregated memories and disks in the cluster to optimize system performance.Experimental results suggest that although TH-CluFS is implemented as user-level components, it functions as a portable, single system image, and scalable cluster file system with acceptable performance sacrifices.     

Design and Implementation of a Storage Virtualization System Based on SCSI Target Simulator in SAN

The ideal storage virtualization system is compatible with all operating systems in storage are anetworks (SANs). However, current storage systems on clustered hosts and multiple operating systems are not practical. This paper presents a storage virtualization system based on a SCSI target simulator in a SAN to solve these problems. This storage virtualization system runs in the target hosts of the SAN, dynamically stores the physical information, and uses the mapping table method to modify the SCSI command addresses. The system uses the bitmap technique to manage the free space. The storage virtualization system provides various functions, such as logical volume resizing, data mirroring, and snapshots, and is compatible with clustered hosts and multiple operating systems, such as Windows NT and RedHat.     

可自管理的分布式应用管理覆盖网络

现有的分布式应用管理工具使用集中式算法部署与维护,其扩展性差,且不能很好地应对网络分割错误。为解决这个问题,该文使用epidemic算法,设计了具有自我管理能力的分布式应用管理覆盖网络(SMON)。SMON的对等节点构成一个随机结构覆盖网络,节点通过相互随机探测,能够在一组预定机器上自动部署新的节点,并恢复运行错误的节点。通过使用认证代理技术,节点能够安全地登录到远程机器以完成部署或恢复任务。在Planet-Lab平台上的测试显示,在大规模分布式环境下,SMON具有良好的性能和可扩展性。     

蓬勃发展的中国计算机事业

 中国的计算机事业起步于20世纪50年代中期，与国外同期的先进计算机水平相比晚了约10年。过去60年，经过广大科研人员艰苦卓绝的奋斗，中国的计算机科研能力取得了长足的进步，许多领域达到了国际先进水平，高性能计算机的研制水平达到了国际领先水平。本期《科技导报》组织了信息专题，一方面为读者梳理中国计算机事业发展过程中的重要事件，另一方面介绍中国在多个计算机领域的科研进展，希望能让读者对中国计算机事业有更好的了解。     

面向全系统毁坏后的服务即时恢复技术

论述了一种全新的系统容灾保护方法，它脱离了传统基于数据保护的容灾保护思路，通过复制包括数据状态以及服务运行状态在内的全系统状态，并引入并行恢复的思想，最终达到了无论系统毁坏程度如何，都能保证服务即时恢 复的目的。同时，与传统技术相比，该技术还可独立于具体设备和应用，容易做到容灾资源的共享从而节省了容灾系统的建设成本。     

基于报备池和模版的网格记账溯源服务

提出一种网格记账溯源服务(Accounting Provenance Service,APS),针对网格记账数据的描述和获取,分别采用模版和报备池/动态报备组的方法,并分析APS在ChinaGrid中的潜在应用场景.研究结果表明,APS能将记账数据的采集独立于上层异构的记账策略,灵活地应用于各种记账背景,而且在记录单个网格成员记账行为的同时可以重现多个网格成员参与的分布式流程现场(如工作流作业的执行),提供记账事件的“溯源”功能,防止网格成员的虚假行为.     

基于区域分解和MPI的线性带状方程组归并迭代解法器

线性带状方程组并行解法器往往基于两层迭代的区域分解方法,采用M P I(m essage pass ing in terface)实现,因此导致的总迭代次数太多或者进程通信开销太大都会使解法器效率低下。该文通过研究减少迭代次数和降低进程通信开销的方法,设计了一种适合区域分解和M P I系统的高效的归并迭代并行解法器。这种解法器通过引入全局加速收敛算法,把两层迭代归并为一层迭代,有效减少了迭代求解的总次数,并且采用分块并行技术降低M P I系统上加速收敛算法的进程通信开销。实验证明归并迭代并行解法器能够保证和串行解法器大致相当的总迭代次数,分块并行加速收敛技术能够降低接近1/2的全局进程通信时间。     

Analysis of Factors Affecting Execution Performance of OpenMP Programs

OpenMP is able to develop multithreaded programs with lower cost as a promising way to quickly parallelize a great number of serial programs. However, there are many difficulties when developing OpenMP multithreaded programs. Most importantly, the performance of OpenMP programs cannot be guaranteed, which has greatly affected the application of OpenMP. This paper examines several factors that affect the execution performance of OpenMP programs, including loop transformation, scheduling strategy, the number of threads, and nested parallelism. The results show that one of the key reasons for performance degradation is load imbalance, which can be corrected by reasonable use of these key factors to improve the execution performance of OpenMP programs.