Minimizing Delay by Multi-Hop Routing and Scheduling in Clos Networks for DCNs

被引:0
作者
Zhao, Zhipeng [1 ]
Li, Boyu [1 ]
Hu, Zhenyu [2 ]
Guan, Yan [1 ]
Wu, Bin [1 ]
Dong, Xiaowen [3 ]
Zhu, Guanyu [3 ]
机构
[1] Tianjin Univ, Sch Comp Sci & Technol, Tianjin, Peoples R China
[2] Tianjin Univ, Coll Management & Econ, Tianjin, Peoples R China
[3] Huawei Technol Co Ltd, Data Ctr Technol Lab, Beijing, Peoples R China
来源
2017 4TH INTERNATIONAL CONFERENCE ON SYSTEMS AND INFORMATICS (ICSAI) | 2017年
关键词
Clos networks; data center networks (DCNs); multi-hop routing and scheduling (MHRS); traffic scheduling; DATA CENTERS; SWITCHES; OVERHEAD;
D O I
暂无
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
We study traffic scheduling in optical-electrical switching fabrics in Data Center Networks (DCNs) based on multi-stage Clos networks to minimize data transmission delay and ensure 100% throughput. The existing classical ADAPT scheduling algorithm can minimize the speedup (the radio of the internal transmission rate inside the optical switch to the external line-rate), but still has some idle time-slots (i.e., not-yet-utilized slots), especially for uneven traffic distribution. To solve this issue, we propose a Multi-Hop Routing and Scheduling (MHRS) algorithm by utilizing idle bandwidths without increasing the speedup in multi-stage Clos networks for DCNs. Similar to ADAPT, MHRS divides traffic matrix into quotient and residual matrices. It includes two steps: single-hop scheduling as the first and multi-hop scheduling as the second. The first step is to schedule packets in the residual matrix to the idle time-slots of configuration matrices generated from the quotient matrix. When those packets cannot be directly scheduled in the first step, the second step invokes multi-hop scheduling process by making a detour for transmission. Simulation results confirm that MHRS can achieve a better performance in multi-stage Clos networks as compared with ADAPT.
引用
收藏
页码:1018 / 1023
页数:6
相关论文
共 20 条
  • [1] [Anonymous], 2014, 16 AS PAC NETW OP MA
  • [2] Bai W, 2017, IEEE ACM T NETWORKIN
  • [3] Cai Y, 2009, GLOB WORKSH GC WRKSH
  • [4] Optical Interconnects at the Top of the Rack for Energy-Efficient Data Centers
    Chen, Jiajia
    Gong, Yu
    Fiorani, Matteo
    Aleksic, Slavisa
    [J]. IEEE COMMUNICATIONS MAGAZINE, 2015, 53 (08) : 140 - 148
  • [5] Cheng Y., 2016, GLOBALISATION HIGHER, P1
  • [6] Switch cost and packet delay tradeoff in data center networks with switch reconfiguration overhead
    Fu, Shu
    Wu, Bin
    Jiang, Xiaohong
    Pattavina, Achille
    Wen, Hong
    Yu, Hongfang
    [J]. COMPUTER NETWORKS, 2015, 87 : 33 - 43
  • [7] Exploiting Efficient and Scalable Shuffle Transfers in Future Data Center Networks
    Guo, Deke
    Xie, Junjie
    Zhou, Xiaolei
    Zhu, Xiaomin
    Wei, Wei
    Luo, Xueshan
    [J]. IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, 2015, 26 (04) : 997 - 1009
  • [8] Huang H, 2016, GLOB COMM C GLOBECOM, P1
  • [9] Delay-Differentiated Scheduling in Optical Packet Switches for Cloud Data Centers
    Li Yaofang
    Xiao Jie
    Wu Bin
    Wen Hong
    Yu Hongfang
    Yang Shu
    Xin Shanshan
    Guo Jianing
    [J]. CHINA COMMUNICATIONS, 2015, 12 (08) : 22 - 32
  • [10] NI W, 2014, IEEE OSA J LIGHTWAVE, V32, P1482