跨数据中心虚拟网络技术设计与实现

发布时间：2018-01-17 19:36

  本文关键词：跨数据中心虚拟网络技术设计与实现　出处：《国防科学技术大学》2013年硕士论文　论文类型：学位论文

  更多相关文章： 数据中心 虚拟网络技术 虚拟交换机

【摘要】：跨数据中心虚拟网络技术可以较好的解决数据中心集群互联问题。跨数据中心大二层虚拟网络技术实现在多个数据中心间的虚拟机无缝迁移和资源池的最大化利用。研究大二层网络技术具有重要的理论意义和实用价值。当前,国内外已推广应用的跨数据中心虚拟网络技术主要有OVS-GRE和OTV技术。本文针对这两种技术进行分析研究,总结这两种技术,或多或少都存在着一定的局限性:(1)OVS-GRE能来解决虚拟机和物理网络之间的通信问题,但是该方案只能创建点到点的隧道,且效率较低。(2)OTV技术能够在分散的二层数据中心之间实现跨广域网的二层连接,但是该技术属于思科的私有技术,且转发机制复杂、需要其他设备支持组播、运维成本高,尚未实现标准化,部署成本较高等缺陷。鉴于上述分析,提出了一种新的跨数据中心的报文转发机制ODV。文中对ODV技术的总体架构和实现进行了深入分析。与OTV相比,ODV它简化了层次架构,把数据中心双层架构改为单层。缩短了数据传输过程,原来的传输过程的17个步骤缩减为7个步骤。减少封装次数,重新定义传输接口,把原来需要的四种接口简化为本地、跨越两种接口,重新定义了数据封装格式,提出了一种新的数据包发送接收规则。通过一种新的BRUDP报文收发模式,通过源与目的端交换DONE信号与错位位图,通告未达数据报信息,来保证未达报文重传,确保报文传输的可靠性,减小了传输开销。在转发策略的选择上提出新的转发选择算法。在功能验证和性能测试中,验证了ODV技术的可行性,表明该技术可以实现数据中心之间的二层网络的互联互通,具有较好的应用前景。
[Abstract]:Cross-data center virtual network technology can better solve the problem of data center cluster interconnection. The virtual network technology can realize seamless migration of virtual machine and maximization of resource pool among multiple data centers. It is of great theoretical significance and practical value to study the technology of the second layer network. OVS-GRE and OTV are the main technologies of cross-data center virtual network which have been popularized and applied at home and abroad. In this paper, the two technologies are analyzed and summarized. More or less, there are some limitations. The OVS-GRE can solve the communication problem between virtual machine and physical network, but the scheme can only create a point-to-point tunnel. And the OTV technology of low efficiency can realize the two-layer connection across WAN between the distributed two-layer data centers, but this technology belongs to the private technology of Cisco, and the forwarding mechanism is complex. Other devices are needed to support multicast, the cost of operation and maintenance is high, the standardization has not been realized, and the cost of deployment is high. In view of the above analysis. A new packet forwarding mechanism across data centers, ODV, is proposed. In this paper, the overall architecture and implementation of ODV technology are analyzed in depth. Compared with OTV, it simplifies the hierarchical architecture. The data center double layer architecture is changed to single layer. The data transfer process is shortened, the 17 steps of the original transmission process are reduced to 7 steps, the number of encapsulation is reduced, and the transfer interface is redefined. This paper simplifies four kinds of interfaces into local, spans two interfaces, redefines the data encapsulation format, proposes a new packet sending and receiving rule, and uses a new BRUDP message sending and receiving mode. Through the exchange of DONE signal and dislocation map between source and destination, the information of unreachable Datagram is notified to ensure the retransmission of unreachable message and ensure the reliability of message transmission. The transmission overhead is reduced. A new forwarding selection algorithm is proposed in the selection of forwarding strategy. The feasibility of ODV technology is verified in function verification and performance testing. It is shown that this technique can realize the interconnection and interworking between the two layers of data centers, and it has a good application prospect.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP393.01;TP308
，

本文编号：1437701