期刊文献+

基于最小接入保证带宽的数据中心网络带宽分配机制 被引量:3

Minimum access guaranteed bandwidth allocation mechanism in data center network
下载PDF
导出
摘要 针对云计算数据中心(DC)网络中多租户间易相互干扰导致应用性能不可预测,预留带宽资源难以保证高网络利用率,从而导致云网络供应商收益损失等问题,提出了基于最小接入保证带宽的数据中心网络带宽分配(MAGBA)机制。MAGBA机制在发送端通过加权公平队列调度租户虚拟机(VM)流量,在接收端基于检测到的接收速率调节TCP流接收窗口,从而为租户VM提供最小接入保证带宽,同时充分利用闲置带宽资源。通过NS2平台仿真,MAGBA机制相比静态资源预留方式能够灵活分配带宽资源,网络吞吐量提升25%。当某些租户产生大量TCP流时,MAGBA机制中其他租户分得的带宽高于现有基于TCP流的带宽分配机制。仿真结果表明,MAGBA机制能够为租户VM提供最小接入带宽保证,避免租户应用受到其他租户通信模式的干扰。 In Data Center Network (DCN), multiple tenants may interfere with each other, which leads to unpredictable application performance. Reserving bandwidth resources can hardly guarantee high network utilization, which may result in revenue loss of cloud providers. To address above problems, a Minimum Access Guaranteed Bandwidth Allocation (MAGBA) mechanism for data center networks was proposed. To provide the minimum bandwidth guarantee and make full use of idle bandwidth for tenants, the MAGBA scheduled traffic of Virtual Machine (VM) through Weighted Fairness Queuing at the sending side, and it adjusted TCP flow's receiving window at the receiving side. Through simulations on NS2 ( Network Simulation version 2) platform, compared with the static resource reservation method, MAGBA mechanism was more flexible in bandwidth allocation and it could improve the network throughput by 25%. When some tenants sent a lot of TCP flows, the other tenants in the MAGBA obtained higher bandwidth than that in the existing bandwidth allocation mechanism based on TCP flows. The simulation results show that the MAGBA mechanism can provide minimum access guaranteed bandwidth for VMs and it can avoid interferences from other tenants.
出处 《计算机应用》 CSCD 北大核心 2017年第7期1825-1829,共5页 journal of Computer Applications
基金 国家自然科学基金资助项目(61301119)~~
关键词 云计算 数据中心网络 多租户 资源共享 带宽分配 性能隔离 cloud computing Data Center Network (DCN) multi-tenant resource sharing bandwidth allocation performance isolation
  • 相关文献

参考文献2

二级参考文献9

  • 1Kim H, Jo H, Lee J. XHive: efficient cooperative caching for virtual machines [ J ]. IEEE Transactions on Computers, 2011,60(1) : 106-119.
  • 2Chowdhury N M M K, Boutaba R. Network virtualization: state of the art and research challenges [ J ]. IEEE Communications Magazine, 2009,47(7) :20-26.
  • 3Mohammad A, Alexander L, Amin V. A scalable commodity data center network architecture [ C ] //SICWDMM' 08. Se.att[e: ACM, 2008:63-74.
  • 4Li D, Guo C, Wu H, et al. FiConn: using backup port for server interconnection in data centers[C]//INFI2)CX)M 2009. Rio de Janeiro: IEEE, 2009..2276-2285.
  • 5Guo C, Wu H, Tan K, et al. DCelh a scalable and faulttolerant network structure for data center [C]//SIGCOMM' 08. Seattle: ACM, 2008:75-86.
  • 6Guo C, Lu G, Dan L, et al. BCube: a high performance, server-centric network architecture for modular data centers [ C]//SIGCOMM' 09. Barcelona: ACM, 2009 : 63-74.
  • 7Rajanna V S, Shah S, Jahagirdar A, et al. XCo: explicit coordination for preventing congestion in data center ethernet [C]//The 6th IEEE International Workshop on Storage Network Architecture and Parallel I/Os. Incline Village: IEEE, 2010:81-89.
  • 8Nick M, Tom A, Karl I3, et al. Openflow: enabling innovation in campus networks [ EB/OL]. (2008-03-14 ) [2011-04-06]. http://www, openflow, org/documents/ openflow-up-latest, pdf.
  • 9刘晓茜,杨寿保,郭良敏,王淑玲,宋浒.雪花结构:一种新型数据中心网络结构[J].计算机学报,2011,34(1):76-86. 被引量:30

共引文献113

同被引文献29

引证文献3

二级引证文献19

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部