• DocumentCode
    3275828
  • Title

    The research of performance optimization technology in the high-speed network

  • Author

    Binfeng Wang ; Lin Chen ; Yaping Liu

  • Author_Institution
    Sch. of Comput., Nat. Univ. of Defense Technol., Changsha, China
  • fYear
    2013
  • fDate
    23-25 May 2013
  • Firstpage
    405
  • Lastpage
    409
  • Abstract
    RDMA (Remote Direct Memory Access) has spread rapidly with its unique advantages in the high-speed cluster and server area network and becomes an important supporting technology of InfiniBand(IB) network. Analyzing the supporting mechanisms and semantics of RDMA, this paper proposes a centralized control measurement method called CenCtlM (Central Control Measure), which can automatically generate measurement configuration strategy combined with the network topology. To further optimize the transmission performance of RDMA in real IB network, the network performance was evaluated and validated in terms of different message ranges and MTUs (Maximum Transmission Unit). The experiment shows that the measurement method proposed in this paper can be fine-grained, low overhead to discover and locate the performance bottleneck of the IB network, which has an important guiding significance in the research of the optimizing transmission performance of the IB network.
  • Keywords
    optimisation; storage management; CenCtlM; IB network; InfiniBand network; MTU; RDMA; centralized control measurement; high-speed network; maximum transmission unit; network topology; performance optimization technology; remote direct memory access; Kernel; Read only memory; Reliability theory; Semantics; Switches; Topology; InfiniBand; RDMA; network bandwidth; network latency; network measurement;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Software Engineering and Service Science (ICSESS), 2013 4th IEEE International Conference on
  • Conference_Location
    Beijing
  • ISSN
    2327-0586
  • Print_ISBN
    978-1-4673-4997-0
  • Type

    conf

  • DOI
    10.1109/ICSESS.2013.6615335
  • Filename
    6615335