计算机科学 ›› 2017, Vol. 44 ›› Issue (Z6): 563-566.doi: 10.11896/j.issn.1002-137X.2017.6A.126

• 综合、交叉与应用 • 上一篇    下一篇

OpenStack云平台的高可用设计与实现

罗兵,谯英,符晓   

  1. 西南石油大学计算机科学学院 成都610500,西南石油大学计算机科学学院 成都610500,西南石油大学计算机科学学院 成都610500
  • 出版日期:2017-12-01 发布日期:2018-12-01
  • 基金资助:
    本文受四川省教育厅专项基金018-数字油田监控(020402000018)资助

Design and Implementation of High-availability Based on OpenStack Cloud Platform

LUO Bing, QIAO Ying and FU Xiao   

  • Online:2017-12-01 Published:2018-12-01

摘要: 实现高可用性是OpenStack云计算管理平台研究的重要问题之一。针对OpenStack云计算管理平台的相关服务组件运行在单节点上易导致单点故障(SPoF)的问题,结合现有多种系统高可用性解决方案,提出一种基于Pacemaker+Corosync+HAProxy+Ceph的解决方案以实现OpenStack云计算管理平台的高可用。该方案将Active-Active的双活模式、Active-Passive的主备模式及集群技术3种高可用设计模式融合在一起,通过软硬件冗余和服务实例故障转移等方式实现OpenStack云计算管理平台的高可用性。实验证明,在少量节点或链路中断的情况下OpenStack云计算管理平台仍然能够稳定运行,该高可用性方案具有可行性。

关键词: 云计算,OpenStack,高可用性,单点故障,云平台

Abstract: Achieving high availability is one of the most important issues in the study of the OpenStack cloud management platform.In order to solve the problem that related services for OpenStack cloud management platform components running on a single node can cause single points of failure (SPoF),we combined with existing high-availability solutions,put forward a method based on Pacemaker+Corosync+HAProxy+Ceph to realize the high availability of OpenStack cloud management platform.This solution combines with Active-Active main/standby mode,Active-Passive Double live mode and cluster technology together,and then uses the method of hardware and software redundancy and transfer failure of service instance,finally,comes up with the high availability of OpenStack cloud management platform.Experiments show that in the case of the system only having fewer nodes or lose of link,the method is effective.

Key words: Cloud computing,OpenStack,High-availability,SPoF,Cloud platform

[1] 程宏兵,赵紫星,叶长河.基于体系架构的云计算安全研究进展[J].计算机科学,2016,43(7):19-27.
[2] 宋俊锋.基于MILP的云计算数据中心扩张策略优化模型[J].湘潭大学自科学报,2015,37(4):105-110.
[3] CORRADI A,FANELLI M,FOSCHINI L.VM consolidation:A real case based on OpenStack Cloud[J].Future Generation Computer Systems,2014,32(1):118-127.
[4] 张帆,李磊,杨成胡,等.基于Eucalyptus构建私有云计算平台[J].电信科学,2011,27(11):57-61.
[5] KESSACI Y,MELAB N,TALBI E G.A multi-start local search heuristic for an energy efficient VMs assignment on top of the OpenNebula cloud manager[J].Future Generation Computer Systems,2014,36(3):237-256.
[6] MILOJIˇIC′ D,LLORENTE I M,MONTERO R S.OpenNebula:A Cloud Management Tool [J].Internet Computing IEEE,2011,15(2):11-14.
[7] 于飞.基于openNebula云平台实验及性能评估[D].北京:北京邮电大学,2013
[8] PARADOWSKI A,LIU L,YUAN B.Benchmarking the Performance of OpenStack and CloudSack[C]∥ IEEE,International Symposium on Object/Component/Service-Oriented Real-Time Distributed Computing.IEEE,2014:405-412.
[9] 陈星,张颖,张晓东,等.基于运行时模型的多样化云资源管理方法[J].软件学报,2014,25(7):1476-1491.
[10] 马友忠,慈祥,孟小峰.海量高维向量的并行Top-k连接查询[J].计算机学报,2015,38(1):86-98.
[11] 马友忠,孟小峰.云数据管理索引技术研究[J].软件学报,2015,26(1):145-166.
[12] ANGLANO C,CANONICO M,GUAZZONE M.FC2Q:exploiting fuzzy control in server consolidation for cloud applications with SLA constraints[J].Concurrency and Computation:Practice and Experience,2015,27(17):910-915.
[13] TROPE R L,RESSLER E K.Mettle Fatigue:VW’s Single-Point-of-Failure Ethics[J].IEEE Security & Privacy Magazine,2016,14(1):12-30.
[14] 杨祥.无线传感器网络无标度容错拓扑的连锁故障诊断算法[J].计算机应用研究,2016,33(2):549-551.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!