基于能耗感知的虚拟网络映射模型与优化算法研究
发布时间：2019-05-19 16:37
【摘要】：随着云计算的兴起,互联网业务呈爆炸式增长,使得传统的互联网结构面临许多挑战,使其僵化现象日趋显著。如果采用全新的互联网架构来解决僵化问题将会遇到以下两方面困难:一方面是将会导致现有的互联网结构、路由器硬件和软件的根本性变化,带来十分昂贵的成本;另一方面是由于互联网业务具有“多供应商”的特点,因此若要改变当前互联网的架构就需要多个互联网服务提供商(Internet Service Providers,ISPs)的共同认可。由于上述因素的存在,构建全新的互联网架构就出现了很大的阻力,为了解决互联网的问题,网络虚拟化(Network Virtualization,NV)技术被提出,为解决网络僵化问题开辟了新的思路。网络虚拟化技术为解决网络僵化问题提供了一种有效的途径,而在网络虚拟化技术的研究中,虚拟网络映射是其中的一个重要研究内容,即使多个虚拟网络请求(Virtual Network Request,VNR)映射到同一个底层物理网络(Substrate Network,SN)上,共享SN上的资源。而如何有效地实现VNR映射到SN上成为网络虚拟化中面临的一个重要挑战。在当前,如何提高SN对VNR的接受率,是网络虚拟化领域中的一个研究重点。同时随着云计算的快速发展,网络数据中心的规模也变的越来越大,能量消耗就成为一个不可忽视的因素。尽管如此,但是能耗很少作为网络虚拟化的研究因素。本文的主要研究内容包括:(1)为解决网络能耗问题提出了虚拟网络映射的能耗感知模型(Energy-aware Virtual Network Embedding Model,EAVNE),该模型主要有三部分组成:节点能耗、链路能耗和路由能耗。以该模型作为虚拟网络映射的目标函数,进行虚拟网络映射,用算法进行优化求解,实验证明,提出的模型降低底层网络的能耗。(2)设计实现了基于群搜索的能耗感知虚拟网络映射算法(Energy-aware Virtual Network Embedding Algorithm based on Group Search Optimizer,GSOVNE)。运用GSO算法时,首先对算法中的算子重新定义,然后初始化算法种群,以能耗感知模型作为目标函数,通过GSO算法迭代搜寻最优节点映射方案,再次通过最短路径算法寻找最优链路映射方案,最后输出最优虚拟网络映射方案。该算法能够提高接受率,并降低能耗。(3)随着VNR的不断到来和离开,SN会由于资源分配不均衡而导致节点瓶颈或者链路瓶颈,造成VNR无法有效的映射。为此,在虚拟网络映射算法中引入虚拟网络重配机制,提出了一种面向能耗感知的虚拟网络重配置算法(A Virtual Network Reconfiguration Algorithm Oriented to Energy Consumption,VNREC)。当VNR被拒绝时,重配置算法对SN上满足迁移条件的已经映射的虚拟节点和虚拟链路进行迁移,然后在用GSO算法对被拒绝的VNR进行再次映射。通过对SN进行重配置,提高了虚拟网络的接受率,采用能耗模型进行优化,在一定程度上降低了SN的能耗。
[Abstract]:With the rise of cloud computing, the Internet business is exploding, which makes the traditional Internet structure face many challenges and makes its rigidity become more and more obvious. If a new Internet architecture is adopted to solve the rigid problem, the following two difficulties will be encountered: on the one hand, it will lead to fundamental changes in the existing Internet structure, router hardware and software, resulting in very expensive costs; On the other hand, because the Internet business has the characteristic of "multi-supplier", if we want to change the current Internet architecture, we need the common recognition of multiple Internet service providers (Internet Service Providers,ISPs). Due to the existence of the above factors, there is great resistance to the construction of a new Internet architecture. In order to solve the problem of the Internet, network virtualization (Network Virtualization,NV) technology has been proposed, which opens up a new way to solve the problem of network rigidity. Network virtualization technology provides an effective way to solve the problem of network rigidity, and virtual network mapping is one of the important research contents in the research of network virtualization technology, even if multiple virtual networks request (Virtual Network Request,. VNR) maps to the same underlying physical network (Substrate Network,SN) and shares resources on the SN. How to effectively map VNR to SN has become an important challenge in network virtualization. At present, how to improve the acceptance rate of SN to VNR is a research focus in the field of network virtualization. At the same time, with the rapid development of cloud computing, the scale of network data center is becoming larger and larger, and energy consumption has become a factor that can not be ignored. Nevertheless, energy consumption is rarely used as a research factor for network virtualization. The main research contents of this paper are as follows: (1) in order to solve the problem of network energy consumption, a virtual network mapping energy consumption perception model (Energy-aware Virtual Network Embedding Model,EAVNE) is proposed. The model consists of three parts: node energy consumption. Link energy consumption and routing energy consumption. The model is used as the objective function of virtual network mapping, the virtual network mapping is carried out, and the algorithm is used to optimize the solution. The experimental results show that, The proposed model reduces the energy consumption of the underlying network. (2) an energy consumption aware virtual network mapping algorithm based on group search (Energy-aware Virtual Network Embedding Algorithm based on Group Search Optimizer,GSOVNE) is designed and implemented. When using GSO algorithm, the operator in the algorithm is redefined, then the algorithm population is initialized, and the energy consumption perception model is used as the objective function to search for the optimal node mapping scheme iteratively through GSO algorithm. Thirdly, the optimal link mapping scheme is found by the shortest path algorithm, and the optimal virtual network mapping scheme is finally output. The algorithm can improve the acceptance rate and reduce the energy consumption. (3) with the continuous arrival and departure of VNR, SN will lead to node bottleneck or link bottleneck due to uneven resource allocation, resulting in VNR unable to map effectively. Therefore, a virtual network reconfiguration algorithm (A Virtual Network Reconfiguration Algorithm Oriented to Energy Consumption,VNREC for energy consumption awareness is proposed by introducing virtual network reallocation mechanism into virtual network mapping algorithm. When VNR is rejected, the reconfiguration algorithm is used to migrate the mapped virtual nodes and virtual links on the SN that meet the migration conditions, and then the rejected VNR is mapped again by the GSO algorithm. By reconfiguring SN, the acceptance rate of virtual network is improved, and the energy consumption model is used to optimize the energy consumption of SN to a certain extent.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.01
，
本文编号：2480878
[Abstract]:With the rise of cloud computing, the Internet business is exploding, which makes the traditional Internet structure face many challenges and makes its rigidity become more and more obvious. If a new Internet architecture is adopted to solve the rigid problem, the following two difficulties will be encountered: on the one hand, it will lead to fundamental changes in the existing Internet structure, router hardware and software, resulting in very expensive costs; On the other hand, because the Internet business has the characteristic of "multi-supplier", if we want to change the current Internet architecture, we need the common recognition of multiple Internet service providers (Internet Service Providers,ISPs). Due to the existence of the above factors, there is great resistance to the construction of a new Internet architecture. In order to solve the problem of the Internet, network virtualization (Network Virtualization,NV) technology has been proposed, which opens up a new way to solve the problem of network rigidity. Network virtualization technology provides an effective way to solve the problem of network rigidity, and virtual network mapping is one of the important research contents in the research of network virtualization technology, even if multiple virtual networks request (Virtual Network Request,. VNR) maps to the same underlying physical network (Substrate Network,SN) and shares resources on the SN. How to effectively map VNR to SN has become an important challenge in network virtualization. At present, how to improve the acceptance rate of SN to VNR is a research focus in the field of network virtualization. At the same time, with the rapid development of cloud computing, the scale of network data center is becoming larger and larger, and energy consumption has become a factor that can not be ignored. Nevertheless, energy consumption is rarely used as a research factor for network virtualization. The main research contents of this paper are as follows: (1) in order to solve the problem of network energy consumption, a virtual network mapping energy consumption perception model (Energy-aware Virtual Network Embedding Model,EAVNE) is proposed. The model consists of three parts: node energy consumption. Link energy consumption and routing energy consumption. The model is used as the objective function of virtual network mapping, the virtual network mapping is carried out, and the algorithm is used to optimize the solution. The experimental results show that, The proposed model reduces the energy consumption of the underlying network. (2) an energy consumption aware virtual network mapping algorithm based on group search (Energy-aware Virtual Network Embedding Algorithm based on Group Search Optimizer,GSOVNE) is designed and implemented. When using GSO algorithm, the operator in the algorithm is redefined, then the algorithm population is initialized, and the energy consumption perception model is used as the objective function to search for the optimal node mapping scheme iteratively through GSO algorithm. Thirdly, the optimal link mapping scheme is found by the shortest path algorithm, and the optimal virtual network mapping scheme is finally output. The algorithm can improve the acceptance rate and reduce the energy consumption. (3) with the continuous arrival and departure of VNR, SN will lead to node bottleneck or link bottleneck due to uneven resource allocation, resulting in VNR unable to map effectively. Therefore, a virtual network reconfiguration algorithm (A Virtual Network Reconfiguration Algorithm Oriented to Energy Consumption,VNREC for energy consumption awareness is proposed by introducing virtual network reallocation mechanism into virtual network mapping algorithm. When VNR is rejected, the reconfiguration algorithm is used to migrate the mapped virtual nodes and virtual links on the SN that meet the migration conditions, and then the rejected VNR is mapped again by the GSO algorithm. By reconfiguring SN, the acceptance rate of virtual network is improved, and the energy consumption model is used to optimize the energy consumption of SN to a certain extent.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.01
，
本文编号：2480878
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2480878.html
