Design and testbed evaluation of RDMA-based middleware for high-performance data transfer applications

被引:6
|
作者
Ren, Yufei [1 ]
Li, Tan [1 ]
Yu, Dantong [2 ,3 ]
Jin, Shudong [1 ,4 ]
Robertazzi, Thomas [1 ,5 ]
机构
[1] SUNY Stony Brook, Stony Brook, NY 11794 USA
[2] Brookhaven Natl Lab, Upton, NY 11973 USA
[3] Brookhaven Natl Lab, Grid Comp Grp, Upton, NY 11973 USA
[4] SUNY Stony Brook, Ctr Wireless & Informat Technol, Stony Brook, NY 11794 USA
[5] SUNY Stony Brook, Dept Elect & Comp Engn, Stony Brook, NY 11794 USA
基金
美国能源部;
关键词
Distributed systems; Middleware; Remote Direct Memory Access;
D O I
10.1016/j.jss.2013.01.070
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Providing high-speed data transfer is vital to various data-intensive applications supported by data center networks. We design a middleware layer of high-speed communication based on Remote Direct Memory Access (RDMA) that serves as the common substrate to accelerate various data transfer tools, such as FTP, HTTP, file copy, sync and remote file I/O. This middleware offers better end-to-end bandwidth performance than the traditional TCP-based alternatives, while it hides the heterogeneity of the underlying high-speed architecture. This paper describes this middleware's function modules, including resource abstraction and task synchronization and scheduling, that maximize the parallelism and performance of RDMA operations. For networks without RDMA hardware acceleration, we integrate Linux kernel optimization techniques to reduce data copy and processing in the middleware. We provide a reference implementation of the popular file-transfer protocol over this RDMA-based middleware layer, called RFTP. Our experimental results show that our RFTP outperforms several TCP-based FTP tools, such as GridFTP, while it maintains very low CPU consumption on a variety of data center platforms. Furthermore, those results confirm that our RFTP tool achieves near line-speed performance in both LAN and WAN, and scales consistently from 10 Gbps Ethernet to 40 Gbps Ethernet and InfiniBand environments. (C) 2013 Elsevier Inc. All rights reserved.
引用
收藏
页码:1850 / 1863
页数:14
相关论文
共 50 条
  • [1] Middleware Support for RDMA-based Data Transfer in Cloud Computing
    Ren, Yufei
    Li, Tan
    Yu, Dantong
    Jin, Shudong
    Robertazzi, Thomas
    2012 IEEE 26TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM WORKSHOPS & PHD FORUM (IPDPSW), 2012, : 1095 - 1103
  • [2] RDMA-Based Apache Storm for High-Performance Stream Data Processing
    Ziyu Zhang
    Zitan Liu
    Qingcai Jiang
    Junshi Chen
    Hong An
    International Journal of Parallel Programming, 2021, 49 : 671 - 684
  • [3] RDMA-Based Apache Storm for High-Performance Stream Data Processing
    Zhang, Ziyu
    Liu, Zitan
    Jiang, Qingcai
    Chen, Junshi
    An, Hong
    INTERNATIONAL JOURNAL OF PARALLEL PROGRAMMING, 2021, 49 (05) : 671 - 684
  • [4] Design and Performance Evaluation of NUMA-Aware RDMA-Based End-to-End Data Transfer Systems
    Ren, Yufei
    Li, Tan
    Yu, Dantong
    Jin, Shudong
    Robertazzi, Thomas
    2013 INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS (SC), 2013,
  • [5] High Performance RDMA-based Design of HDFS over InfiniBand
    Islam, N. S.
    Rahman, M. W.
    Jose, J.
    Rajachandrasekar, R.
    Wang, H.
    Subramoni, H.
    Murthy, C.
    Panda, D. K.
    2012 INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS (SC), 2012,
  • [6] Evaluating High Performance Data Transfer with RDMA-based Protocols in Wide-Area Networks
    Kissel, Ezra
    Swany, Martin
    2012 IEEE 14TH INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING AND COMMUNICATIONS & 2012 IEEE 9TH INTERNATIONAL CONFERENCE ON EMBEDDED SOFTWARE AND SYSTEMS (HPCC-ICESS), 2012, : 802 - 811
  • [7] DArray: A High Performance RDMA-Based Distributed Array
    Ding, Baorong
    Han, Mingcong
    Chen, Rong
    PROCEEDINGS OF THE 52ND INTERNATIONAL CONFERENCE ON PARALLEL PROCESSING, ICPP 2023, 2023, : 715 - 724
  • [8] Employing an RDMA-based file system for high performance computing
    Velusamy, V
    Skjellum, A
    Kanevsky, A
    2004 12TH IEEE INTERNATIONAL CONFERENCE ON NETWORKS, VOLS 1 AND 2 , PROCEEDINGS: UNITY IN DIVERSITY, 2004, : 66 - 70
  • [9] High performance RDMA-based MPI implementation over InfiniBand
    Liu, JX
    Wu, JS
    Panda, DK
    INTERNATIONAL JOURNAL OF PARALLEL PROGRAMMING, 2004, 32 (03) : 167 - 198
  • [10] High Performance RDMA-Based MPI Implementation over InfiniBand
    Jiuxing Liu
    Jiesheng Wu
    Dhabaleswar K. Panda
    International Journal of Parallel Programming, 2004, 32 : 167 - 198