CISE Research Instrumentation: Distributed Computing and Real-Time Networking Research

Vaidya, Nitin Welch, Jennifer Zhao, Wei Amato， Nancy Texas A & M University Texas工程实验站分布式计算和实时网络研究本提案要求为同构工作站集群和ATM-FDDI网络硬件提供资金，以帮助分布式计算和实时网络研究。该设备可用于多个项目，包括：(1)异构网络中的实时通信：该项目开发了在关键任务系统中流行的异构网络中保证时限的技术。所要求的设备能够试验一个由ATM和FDDI组成的网络。(2)分布式系统管理算法：通过形式化分析和实验，开发了用于管理分布式系统中的性能和故障的算法。现实的评估需要一个专用的同质工作站集群，以允许不间断的测试和趋势预测。(3)并行应用程序的故障恢复方案：设计有效的恢复方案是必要的，以尽量减少出现故障时的执行时间。所要求的工作站为不同恢复方案的定量比较提供了必要的同质计算环境。(4)分布式共享内存（DSM）系统设计：所需设备提供工作站集群和高速网络，用于开发能够实现高性能的自适应容错DSM

CDA 9529442 Vaidya, Nitin Welch, Jennifer Zhao, Wei Amato, Nancy Texas A & M University Texas Engineering Experiment Station College Station, TX 77843-3000 Distributed Computing and Real-Time Networking Research This proposal requests funding for a homogeneous workstation cluster and ATM-FDDI networking hardware to aid research in distributed computing and real-time networking. The equipment is useful for several projects, including: (1) Real-time communication in heterogeneous networks: This project develops techniques for guaranteeing deadlines in heterogeneous networks popular in mission-critical systems. Requested equipment enables experimentation with a network consisting of ATM and FDDI. (2) Algorithms for distributed systems management: Algorithms are developed for managing performance and faults in distributed systems, using formal analysis and experimentation. Realistic assessment requires a dedicated cluster of homogeneous workstations to allow undisrupted testing and trend prediction. (3) Failure recovery schemes for parallel applications: Design of efficient recovery schemes is necessary to minimize execution time in presence of failures. The requested workstations provide a homogeneous computing environment necessary for quantitative comparison of different recovery schemes. (4) Distributed shared memory (DSM) system design: Requested equipment provides a workstation cluster and high-speed networking for development of an adaptive fault-tolerant DSM capable of achieving high performance