首页> 外文会议>2015 IEEE 4th International Conference on Cloud Networking >The design and implementation of high-speed interconnection network for SDC-based cloud server
【24h】

The design and implementation of high-speed interconnection network for SDC-based cloud server

机译:基于SDC的云服务器高速互联网络的设计与实现

获取原文
获取原文并翻译 | 示例

摘要

A cloud server with high performance/cost, performance/Watt, high-density and high scalability has attracted more attentions from academy and industry field. How to efficiently share and aggregate resources (e.g., computing, storage, I/O bandwidth) is the design focus of cloud server. This paper proposed a 2 D-torus cloud server architecture bas e d on SDC (S oft ware Defined Cluster), in which CPU can share the network and storage resources on demand. In a cloud server, there are two communication types (i.e., local mode and remote mode). To meet these communication demands, this paper proposed a kind of two-level interconnection networks in which the first-level is corresponding to the local communication and the second-level is aiming at the remote communication. For local communication, an Ethernet switch and a Shared Storage Controller (SSC) were proposed to solve the problem. For remote transmission, this paper implements an Interconnection Fabric Controller (IFC) to support the remote data exchanging. The IFC focuses on resolving the router computing, port ar biter, link flow control, retransmission and high speed link transfer protocol. Furthermore, we demonstrate a cloud server with 4 ×4 CPUs based on FP GA. Results of simulation and experiments demonstrate that the above-mentioned designs are valid and feasible.
机译:具有高性能/成本,性能/功率,高密度和高可伸缩性的云服务器吸引了学术界和行业领域的更多关注。如何有效地共享和聚合资源(例如,计算,存储,I / O带宽)是云服务器的设计重点。本文提出了一种基于SDC(软件定义集群)的2 D-torus云服务器架构,其中CPU可以按需共享网络和存储资源。在云服务器中,有两种通信类型(即本地模式和远程模式)。为了满足这些通信需求,本文提出了一种两级互连网络,其中第一级对应于本地通信,第二级针对远程通信。对于本地通信,提出了以太网交换机和共享存储控制器(SSC)来解决该问题。对于远程传输,本文实现了一个互连结构控制器(IFC)以支持远程数据交换。 IFC专注于解决路由器计算,端口仲裁器,链路流控制,重传和高速链路传输协议。此外,我们演示了基于FP GA的具有4×4 CPU的云服务器。仿真和实验结果表明,上述设计是有效可行的。

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