首页> 外文会议>Global Data Networking, 1993. Proceedings >Shaping the distributed end-system for global distributedapplications
【24h】

Shaping the distributed end-system for global distributedapplications

机译:为全球分布式系统打造分布式终端系统应用领域

获取原文

摘要

Advances in optical networking technology provide the network withunique storage capabilities enabling the creation of a physical sharedmemory in a wide area distributed system. This has implications on theoverall systems' architecture as well as on the host. From the userperspective the end-system will be shaped by the application and becomea global distributed shared-memory multiprocessor. In this paper wepropose a memory building block based on optical networking technologythat we further use in devising a multiprocessor host architecture. Ahierarchical memory architecture that ensures scalability and allowsmultiple copies of a shared object is adopted. The architecture favourshardware solutions for enforcing cache coherence by exploiting theintrinsic broadcast capability of the new memory module. Theinterprocess communication system suitable for such an environment isanalysed in order to achieve functional transparency for applicationsported from a multicomputer to a multiprocessor, and issues related toperformance transparency when dealing with long light-propagationlatency are addressed. With this processing-communication paradigm, thehost-network interface simply provides a direct path between thehost-memory and the external network-memory
机译:光网络技术的进步为网络提供了 独特的存储功能可创建物理共享 广域分布式系统中的内存。这对 整个系统的架构以及主机上的架构。来自用户 透视图,最终系统将由应用程序决定并成为 全局分布式共享内存多处理器。在本文中,我们 提出基于光网络技术的内存构建块 我们将在设计多处理器主机体系结构中进一步使用它。一种 分层内存架构,可确保可伸缩性并允许 共享对象的多个副本被采用。体系结构的青睐 通过利用以下方面来增强缓存一致性的硬件解决方案 新内存模块的固有广播功能。这 适用于这种环境的进程间通信系统是 分析以实现应用程序的功能透明性 从多台计算机移植到多处理器,以及与 长时间传播时的性能透明 延迟得到解决。通过这种处理通信范例, 主机网络接口只是提供了之间的直接路径 主机内存和外部网络内存

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号