[go: up one dir, main page]

CN104754033B - The method that virtual machine migrates online is realized based on two-way pipe method under NDN - Google Patents

The method that virtual machine migrates online is realized based on two-way pipe method under NDN Download PDF

Info

Publication number
CN104754033B
CN104754033B CN201510093108.4A CN201510093108A CN104754033B CN 104754033 B CN104754033 B CN 104754033B CN 201510093108 A CN201510093108 A CN 201510093108A CN 104754033 B CN104754033 B CN 104754033B
Authority
CN
China
Prior art keywords
data
ndn
virtual machine
migration
name
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201510093108.4A
Other languages
Chinese (zh)
Other versions
CN104754033A (en
Inventor
张大陆
周德江
金翔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to CN201510093108.4A priority Critical patent/CN104754033B/en
Publication of CN104754033A publication Critical patent/CN104754033A/en
Application granted granted Critical
Publication of CN104754033B publication Critical patent/CN104754033B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

本发明涉及一种在NDN下基于双向管道方式实现虚拟机在线迁移的方法,包括以下步骤:(1)NDN中名字空间设计步骤:定义用于命名数据包的三个名字空间;(2)双向管道通信步骤:数据请求方和数据发送方之间通过双向管道通信数据包;双向管道通信步骤包括逆序和丢包处理:设置用于保存因逆序提前到达或者提前产生的Data的缓存区,同时采用超时重传机制来重新获取Data,当超时多次,则采用分包的方式重新请求;(3)性能测试步骤。与现有技术相比,本发明实现了在NDN网络下的局域网和广域网的虚拟机在线迁移,使虚拟机上的应用在虚拟机迁移的过程中保持不断,实现虚拟机的无缝迁移,并通过动态自适应机制保证迁移完成之后网络畅通。

The present invention relates to a method for realizing online migration of a virtual machine based on a two-way pipeline under NDN, comprising the following steps: (1) Name space design step in NDN: defining three name spaces for naming data packets; (2) two-way Pipeline communication steps: data packets are communicated between the data requester and the data sender through a two-way pipeline; the two-way pipeline communication steps include reverse order and packet loss processing: set a buffer area for storing data that arrives in advance or is generated in advance due to reverse order, and uses The timeout retransmission mechanism is used to re-acquire the Data. When the timeout is repeated multiple times, the request is re-requested in the form of sub-packages; (3) Performance test steps. Compared with the prior art, the present invention realizes the online migration of the virtual machine in the local area network and the wide area network under the NDN network, keeps the application on the virtual machine in the process of virtual machine migration, realizes the seamless migration of the virtual machine, and The dynamic self-adaptation mechanism is used to ensure that the network is smooth after the migration is completed.

Description

在NDN下基于双向管道方式实现虚拟机在线迁移的方法Method of Realizing Online Migration of Virtual Machine Based on Two-way Pipeline under NDN

技术领域technical field

本发明涉及网络通信技术领域和云计算技术领域,尤其是涉及一种在NDN(NamedData Networking,命名数据网络)下基于双向管道(pipeline)方式实现虚拟机在线迁移的方法。The present invention relates to the field of network communication technology and cloud computing technology, and in particular to a method for realizing online migration of a virtual machine based on a bidirectional pipeline under NDN (Named Data Networking).

背景技术Background technique

随着网络的发展,TCP/IP结构出现了越来越多的限制,包括可扩展性、稳定性、安全性和移动性。这些缺点导致了学术界开始了对未来网络的研究。FIA-NP目前主要有三个项目MobilityFirst-NP、Named Data Networking-NP、eXpressive InternetArchitecture(XIA)。其中,NDN的核心理念在于网络关注的不是位置(Where),而是数据(What),这种转变不仅极大地提高了网络的灵活性、可扩展性和移动性,而且内置的安全性策略也提高了网络及其应用的安全性。With the development of the network, more and more limitations have appeared in the TCP/IP structure, including scalability, stability, security and mobility. These shortcomings have led the academic community to start research on future networks. FIA-NP currently has three main projects: MobilityFirst-NP, Named Data Networking-NP, and eXpressive Internet Architecture (XIA). Among them, the core concept of NDN is that the network does not focus on location (Where), but data (What). This transformation not only greatly improves the flexibility, scalability and mobility of the network, but also the built-in security strategy Improve the security of the network and its applications.

作为面向数据网络的代表,NDN现已在全球范围内展开了研究并进行了测试平台的部署。目前NDN的研究已进入下一阶段,主要致力于NDN应用的设计、安全性、路由和转发策略、可伸缩的转发、库和工具的开发、社会和经济影响等方面。NDN只包含两种数据包Interest和Data,是以接收方驱动的网络,称NDN中数据请求方为consumer,称数据发送方为producer,因此,NDN是一种通过consumer发送Interest向producer请求Data进行通信的网络架构。为了接收数据时,一个consumer发出一个Interest报文,该Interest报文携带一个名字,由名字标识期望的数据。NDN节点接受到该Interest报文后,会记录Interest报文进入节点的接口,然后通过基于命名的路由转发协议将其转发。一旦Interest报文达到拥有被请求数据的一个节点,则发回一条Data报文,它携带数据的名字和内容,还有生产者密钥的一个签名。这条Data报文经Interest报文所产生的反向路径到达consumer。As a representative of data-oriented networks, NDN has launched research and deployed test platforms around the world. At present, NDN research has entered the next stage, mainly focusing on the design of NDN applications, security, routing and forwarding strategies, scalable forwarding, library and tool development, social and economic impact, etc. NDN only contains two kinds of data packets, Interest and Data. It is a network driven by the receiver. The data requester in NDN is called the consumer, and the data sender is called the producer. Communication network architecture. In order to receive data, a consumer sends an Interest message, which carries a name, and the name identifies the desired data. After the NDN node receives the Interest message, it will record the interface where the Interest message enters the node, and then forward it through the naming-based routing and forwarding protocol. Once the Interest message reaches a node that owns the requested data, a Data message is sent back, which carries the name and content of the data, and a signature of the producer key. This Data message reaches the consumer through the reverse path generated by the Interest message.

随着计算机技术的飞速发展,云计算顺应时代要求登上了历史舞台。虚拟机迁移是云计算技术中虚拟化和高效分配资源的重要技术之一。虚拟机迁移分为在线迁移和离线迁移,在线迁移因为其能够保证迁移过程中不中断而成为虚拟机迁移的主要研究点。未来网络中虚拟机也是必须的,因为虚拟机不仅有利于主机资源的高效利用,而且对于用户的隔离也有着重要的作用。而虚拟机迁移有利于负载均衡、灾难备份以及用户就近等方面的网络性能优化策略。因此,虚拟机在线迁移也是必需的。With the rapid development of computer technology, cloud computing has entered the stage of history in line with the requirements of the times. Virtual machine migration is one of the important technologies of virtualization and efficient allocation of resources in cloud computing technology. Virtual machine migration is divided into online migration and offline migration. Online migration has become the main research point of virtual machine migration because it can ensure uninterrupted migration process. Virtual machines are also necessary in the future network, because virtual machines are not only conducive to the efficient use of host resources, but also play an important role in the isolation of users. Virtual machine migration is beneficial to network performance optimization strategies in terms of load balancing, disaster backup, and user proximity. Therefore, online migration of virtual machines is also required.

在未来网络方面,虚拟机在线迁移也已经有了一些研究,但是仍然存在一些问题,比如迁移总时间和停机时间比较长。In terms of future networks, there have been some studies on online migration of virtual machines, but there are still some problems, such as relatively long total migration time and downtime.

发明内容Contents of the invention

本发明的目的就是为了克服上述现有技术存在的缺陷而提供一种在NDN下基于双向管道方式实现虚拟机在线迁移的方法,对于NDN通信模型的研究以及对于虚拟机在线迁移在未来网络中的研究都有着重要的推动作用。The purpose of the present invention is to provide a method for realizing online migration of virtual machines under NDN based on a two-way pipeline in order to overcome the above-mentioned defects in the prior art, research on the communication model of NDN and the implementation of online migration of virtual machines in the future network Research plays an important role in promoting.

本发明的目的可以通过以下技术方案来实现:The purpose of the present invention can be achieved through the following technical solutions:

一种在NDN下基于双向管道方式实现虚拟机在线迁移的方法,包括以下步骤:A method for realizing online migration of a virtual machine based on a two-way pipeline under NDN, comprising the following steps:

(1)NDN中名字空间设计步骤:构建NDN进行数据迁移,NDN中数据包分为Interest和Data,定义用于命名数据包的三个名字空间:(1) Namespace design steps in NDN: build NDN for data migration, data packets in NDN are divided into Interest and Data, define three namespaces for naming data packets:

第一名字空间,用于标识迁移过程中的数据包;The first name space is used to identify the data package in the migration process;

第二名字空间,用于标识迁移结束的最后一个数据包;The second name space is used to identify the last data packet at the end of the migration;

第三名字空间,数据迁移过程中因数据包过大而丢包时,将该数据包分成若干个后,用第三名字空间进行标识;The third name space, when the data packet is lost due to the large data packet during the data migration process, the data packet is divided into several, and the third name space is used for identification;

(2)双向管道通信步骤:数据请求方和数据发送方之间通过双向管道通信数据包,具体为:数据请求方以管道的方式向数据发送方发送Interest,数据发送方接收后产生对应的Data,并以管道的方式向数据请求方发送Data,数据请求方接收到返回的Data后产生新的Interest,数据迁移过程中的数据包以第一名字空间进行命名,当以第二名字空间命名的数据包传输结束后,数据迁移结束;(2) Two-way pipeline communication step: The data requester and the data sender communicate data packets through a two-way pipeline, specifically: the data requester sends Interest to the data sender in the form of a pipeline, and the data sender generates the corresponding Data after receiving , and send Data to the data requester in the form of a pipeline, and the data requester generates a new Interest after receiving the returned Data. After the data packet transmission ends, the data migration ends;

双向管道通信步骤包括逆序和丢包处理,逆序的处理为在数据请求方和数据发送方分别设置一个用于保存因逆序提前到达或者提前产生的Data的缓存区,丢包的处理为采用超时重传机制来重新获取Data,当超时Q次后,采用分包的方式将Interest分成若干个小数据块的Interest来请求该Data,分包后的数据包以第三名字空间命名,Q为超时次数阈值;The two-way pipeline communication steps include reverse order and packet loss processing. The reverse order processing is to set a buffer area for saving the data that arrives in advance or is generated in advance due to the reverse order respectively on the data requester and the data sender. Transfer mechanism to re-acquire Data. When the timeout is Q times, the Interest is divided into several small data blocks in the form of subcontracting to request the Data. The subcontracted data packets are named after the third namespace, and Q is the number of timeouts threshold;

(3)性能测试步骤:在虚拟机中运行以与位置无关的名字为前缀的应用,通过访问该应用验证虚拟机迁移的在线性,同时使用NDN的ping程序测量虚拟机的停机时间。(3) Performance testing steps: run an application prefixed with a location-independent name in a virtual machine, verify the onlineness of virtual machine migration by accessing the application, and use the NDN ping program to measure the downtime of the virtual machine.

所述第一名字空间的格式为/ndn/location/vm/migration/live/version/data/sequence,其中,/ndn为NDN中名字空间所共有的root组件,/location表示数据发送方所在的设备或者网关的名字,/vm表明该名字属于与虚拟机相关的名字,/migration表明该名字是迁移应用所使用的名字,/live表明该名字是在线迁移,/version表明该在线迁移名字空间的版本号,/data表明该名字是标识迁移数据的名字,/sequence表明迁移数据的序号,用于唯一标识不同的虚拟机在线迁移的数据块。The format of the first name space is /ndn/location/vm/migration/live/version/data/sequence, where /ndn is the root component shared by the name space in NDN, and /location indicates the device where the data sender is located Or the name of the gateway, /vm indicates that the name belongs to the name related to the virtual machine, /migration indicates that the name is the name used by the migration application, /live indicates that the name is online migration, /version indicates the version of the online migration namespace number, /data indicates that the name is the name that identifies the migrated data, and /sequence indicates the sequence number of the migrated data, which is used to uniquely identify the data blocks that are migrated online by different virtual machines.

所述第二名字空间的格式为/ndn/location/vm/migration/live/version/data/end/last,其中,/end表明迁移数据已经结束,/last标识迁移数据块的最后一个序号,且sequence的数值不大于last的数值。The format of the second namespace is /ndn/location/vm/migration/live/version/data/end/last, where /end indicates that the migration data has ended, and /last indicates the last serial number of the migration data block, and The value of sequence is not greater than the value of last.

所述第三名字空间的格式为/ndn/location/vm/migration/live/version/data/sequence/frag/nums/seq,其中,/frag表明该名字是一个分包的名字,/nums的组件表明一个数据包被分成了几个,/seq表明该名字分包的序号。The format of the third namespace is /ndn/location/vm/migration/live/version/data/sequence/frag/nums/seq, where /frag indicates that the name is a subcontracted name, and the component of /nums Indicates that a data packet is divided into several, /seq indicates the sequence number of the sub-packet with the name.

所述双向管道通信数据包包括两个阶段:启动阶段和稳定阶段;The two-way pipeline communication data packet includes two phases: a start-up phase and a stable phase;

启动阶段:数据请求方初始发送2N个Interest,数据发送方产生并返回N个data;Start-up phase: the data requester initially sends 2N Interests, and the data sender generates and returns N data;

稳定阶段:数据请求方接收一个Data同时产生一个新的Interest;Stable stage: the data requester receives a Data and generates a new Interest at the same time;

启动阶段后进入稳定阶段,稳定阶段时,Data和Interest的传输构成两条相反的管道,形成一个数据流环,其中,N=BD/L,N为网络中正在传输的Data的最大个数,B是网络带宽,D是网络延迟,L是Data的长度。After the startup phase, it enters the stable phase. During the stable phase, the transmission of Data and Interest constitutes two opposite pipelines, forming a data flow ring, where N=BD/L, and N is the maximum number of Data being transmitted in the network. B is the network bandwidth, D is the network delay, and L is the length of Data.

所述缓存区中缓存数组的长度Len由网络中正在传输的Data的最大个数N、超时次数阈值Q、超时时间T以及网络延迟D决定。The length Len of the cache array in the cache area is determined by the maximum number N of data being transmitted in the network, the timeout threshold Q, the timeout time T and the network delay D.

所述逆序的处理具体为:数据请求方中设置第一Data缓冲区,用于保存提前到达的Data,数据发送方中设置第二Data缓冲区,用于保存提前产生的Data,数据请求方将收到的正确序号的Data连同其后缓存的提前到达的Data同时写入到虚拟机。The processing of the reverse order is specifically as follows: the first Data buffer is set in the data requester to save the Data that arrives in advance, and the second Data buffer is set in the data sender to save the Data that is generated in advance, and the data requester will The received Data with the correct sequence number is written to the virtual machine at the same time as the cached and early-arriving Data.

性能测试步骤中,使用NDN的ping程序每隔Pms发送Interest向虚拟机请求Data响应,根据停机过程中ping程序中丢包数M,获得虚拟机在线迁移的停机时间,停机时间的数值范围为[(M-1)×P,(M+1)×P]ms,误差范围为±Pms,其中,P≥300。In the performance test step, the ping program using NDN sends Interest to the virtual machine to request a Data response every Pms. According to the number M of packet loss in the ping program during the shutdown process, the downtime of the online migration of the virtual machine is obtained. The value range of the downtime is [ (M-1)×P, (M+1)×P]ms, the error range is ±Pms, where P≥300.

与现有技术相比,本发明提出了NDN名字的设计、双向管道的通信模型、以及对逆序和丢包的处理方法,具有以下优点:Compared with the prior art, the present invention proposes the design of the NDN name, the communication model of the two-way pipeline, and the processing method for reverse sequence and packet loss, which has the following advantages:

1)NDN中名字的设计是一个关键的问题,本发明基于NDN的构架,针对名字空间进行设计,提出了三种情况下的名字空间,分别用于标识迁移数据过程中、迁移的结束时和丢包过多时的数据包,保证虚拟机在线迁移在NDN下的可行性,具有NDN网络通信的安全性、方便性。1) The design of the name in NDN is a key issue. The present invention is based on the framework of NDN, designs the name space, and proposes name spaces in three cases, which are used to identify the process of migrating data, when the end of the migration, and Data packets when there are too many packet loss ensure the feasibility of virtual machine online migration under NDN, and have the security and convenience of NDN network communication.

2)建立双向管道的通信模型,将网络看成一个缓存区,Interest和Data构成一个环,保证在consumer和producer中一直有数据包处理。与现有的NDN下虚拟机在线迁移研究相比,提高了在线迁移的速率,减少在线迁移的迁移总时间。2) Establish a two-way pipeline communication model, regard the network as a cache area, and Interest and Data form a ring to ensure that there is always data packet processing in the consumer and producer. Compared with the existing research on online migration of virtual machines under NDN, the speed of online migration is improved and the total migration time of online migration is reduced.

3)针对实际在线迁移过程中存在的逆序和丢包问题,设置用于保存因逆序提前到达或者提前产生的Data的缓存区,从而解决逆序,同时丢包次数较少时采用超时重传机制,但是若多次超时重传仍然丢包时,则将一个请求大数据块的Interest分成若干个请求小数据块的Interest,从减小数据块大小的角度,解决丢包过多时的问题,保证在线迁移的稳定性、数据传输的可靠性。3) Aiming at the problems of reverse order and packet loss in the actual online migration process, set up a buffer area for saving the data that arrives in advance or is generated in advance due to the reverse order, so as to solve the reverse order. At the same time, the timeout retransmission mechanism is adopted when the number of packet loss is small. However, if packet loss still occurs after repeated overtime retransmissions, an Interest requesting a large data block is divided into several Interests requesting a small data block. From the perspective of reducing the size of the data block, the problem of excessive packet loss is solved and online The stability of migration and the reliability of data transmission.

4)提出评价本发明虚拟机在线迁移方法性能的方法,在虚拟机中运行以与位置无关的名字为前缀的应用,其中与位置无关的名字具有很好的通用性,从不同的测试端均可以访问到虚拟机的该应用,便于通过访问该应用验证虚拟机迁移的在线性,同时使用NDN的ping程序测量虚拟机的停机时间,可以用于在实验中进行对比实验验证等,在实验室理想环境和真实网络环境下对NDN与TCP/IP下的虚拟机在线迁移的性能比较,可以证实TCP/IP下无法内在地支持虚拟机的在线迁移,但是NDN可以内在的支持虚拟机的在线迁移。同时,NDN的total time小于TCP/IP的total time,表明NDN下的迁移程序就丢包、延迟、逆序等方面的处理上性能要好于TCP/IP。4) Propose a method for evaluating the performance of the virtual machine online migration method of the present invention, run the application prefixed with a location-independent name in the virtual machine, wherein the location-independent name has good versatility, and can be obtained from different test terminals You can access the application of the virtual machine, which is convenient to verify the online migration of the virtual machine by accessing the application. At the same time, use the ping program of NDN to measure the downtime of the virtual machine, which can be used for comparative experiment verification in experiments, etc., in the laboratory The performance comparison between NDN and TCP/IP virtual machine online migration under ideal environment and real network environment can prove that TCP/IP cannot inherently support virtual machine online migration, but NDN can inherently support virtual machine online migration . At the same time, the total time of NDN is less than the total time of TCP/IP, indicating that the migration program under NDN is better than TCP/IP in terms of packet loss, delay, and reverse sequence processing.

5)采用本发明方法在局域网和广域网中都可以实现虚拟机的在线迁移,适用范围广,使虚拟机上的应用在虚拟机迁移的过程中保持不断,实现虚拟机的无缝迁移,并通过动态自适应机制保证迁移完成之后网络畅通。5) The method of the present invention can realize the online migration of the virtual machine in both the local area network and the wide area network. The dynamic self-adaptation mechanism ensures that the network is smooth after the migration is completed.

附图说明Description of drawings

图1为本发明双向管道的通信模型示意图;Fig. 1 is the schematic diagram of the communication model of two-way pipeline of the present invention;

图2为理想环境下的虚拟机在线迁移实验测试床示意图;Fig. 2 is a schematic diagram of a virtual machine online migration experiment test bed in an ideal environment;

图3为真实环境下的虚拟机在线迁移实验测试床示意图。FIG. 3 is a schematic diagram of a virtual machine online migration experiment test bed in a real environment.

图中:HOST_A为源宿主主机,HOST_B为目的宿主主机,HOST_VM为待迁移的虚拟机。In the figure: HOST_A is the source host, HOST_B is the destination host, and HOST_VM is the virtual machine to be migrated.

具体实施方式Detailed ways

下面结合附图和具体实施例对本发明进行详细说明。本实施例以本发明技术方案为前提进行实施,给出了详细的实施方式和具体的操作过程,但本发明的保护范围不限于下述的实施例。The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is carried out on the premise of the technical solution of the present invention, and detailed implementation and specific operation process are given, but the protection scope of the present invention is not limited to the following embodiments.

一种在NDN下基于双向管道方式实现虚拟机在线迁移的方法包括以下步骤:A method for realizing online migration of a virtual machine based on a two-way pipeline under NDN includes the following steps:

(1)NDN中名字空间设计步骤:构建NDN进行数据迁移,NDN中数据包分为Interest和Data,而NDN中名字是独立于NDN转发程序(NFD)但却相当重要的组成部分,因此虚拟机在线迁移过程中,定义用于命名数据包的三个名字空间:(1) Name space design steps in NDN: build NDN for data migration, data packets in NDN are divided into Interest and Data, and names in NDN are independent of the NDN forwarding program (NFD) but a very important part, so the virtual machine During live migration, three namespaces are defined for naming packages:

第一名字空间,用于标识迁移过程中的数据包;The first name space is used to identify the data package in the migration process;

第二名字空间,用于标识迁移结束的最后一个数据包;The second name space is used to identify the last data packet at the end of the migration;

第三名字空间,数据迁移过程中因数据包过大而丢包时,将该数据包分成若干个后,用第三名字空间进行标识;具体每个名字空间中组件的设置如下:The third name space, when the data packet is lost due to the large data packet during the data migration process, the data packet is divided into several, and then identified with the third name space; the specific settings of the components in each name space are as follows:

第一名字空间的格式为/ndn/location/vm/migration/live/version/data/sequence,其中,/ndn为NDN中名字空间所共有的root组件,/location表示数据发送方所在的设备或者网关的名字,/vm表明该名字属于与虚拟机相关的名字,/migration表明该名字是迁移应用所使用的名字,/live表明该名字是在线迁移,/version表明该在线迁移名字空间的版本号,/data表明该名字是标识迁移数据的名字,/sequence表明迁移数据的序号,用于唯一标识不同的虚拟机在线迁移的数据块。The format of the first namespace is /ndn/location/vm/migration/live/version/data/sequence, where /ndn is the root component shared by namespaces in NDN, and /location indicates the device or gateway where the data sender is located /vm indicates that the name belongs to the name related to the virtual machine, /migration indicates that the name is the name used by the migration application, /live indicates that the name is online migration, /version indicates the version number of the online migration name space, /data indicates that the name is the name that identifies the migration data, and /sequence indicates the sequence number of the migration data, which is used to uniquely identify the data blocks that are migrated online by different virtual machines.

第二名字空间的格式为/ndn/location/vm/migration/live/version/data/end/last,其中,/end表明迁移数据已经结束,/last标识迁移数据块的最后一个序号,且sequence的数值不大于last的数值。The format of the second namespace is /ndn/location/vm/migration/live/version/data/end/last, where /end indicates that the migration data has ended, /last indicates the last serial number of the migration data block, and the sequence The value is not greater than the value of last.

第三名字空间的格式为/ndn/location/vm/migration/live/version/data/sequence/frag/nums/seq,其中,/frag表明该名字是一个分包的名字,/nums的组件表明一个数据包被分成了几个,/seq表明该名字分包的序号。The format of the third namespace is /ndn/location/vm/migration/live/version/data/sequence/frag/nums/seq, where /frag indicates that the name is a subcontracted name, and the component of /nums indicates a The data packet is divided into several, and /seq indicates the sequence number of the sub-packet with the name.

(2)双向管道通信步骤:数据请求方(consumer)和数据发送方(producer)之间通过双向管道通信数据包,双向管道通信是指:数据请求方以管道的方式连续向数据发送方发送Interest,数据发送方接收后产生对应的Data,并以管道的方式连续向数据请求方发送Data,数据请求方接收到返回的Data后产生新的Interest,在双向管道通信时,数据迁移过程中的数据包以第一名字空间进行命名,当以第二名字空间命名的数据包传输结束后,数据迁移结束。(2) Two-way pipeline communication steps: The data requester (consumer) and the data sender (producer) communicate data packets through a two-way pipeline. The two-way pipeline communication means: the data requester continuously sends Interest to the data sender in the form of a pipeline , the data sender generates the corresponding Data after receiving it, and continuously sends Data to the data requester in the form of a pipeline, and the data requester generates a new Interest after receiving the returned Data. In the two-way pipeline communication, the data in the process of data migration The package is named in the first name space, and the data migration ends when the transmission of the data package named in the second name space ends.

双向管道通信数据包包括两个阶段:启动阶段和稳定阶段;The two-way pipeline communication packet includes two phases: startup phase and stabilization phase;

启动阶段:数据请求方初始发送2N个Interest,数据发送方产生并返回N个data;Start-up phase: the data requester initially sends 2N Interests, and the data sender generates and returns N data;

稳定阶段:数据请求方接收一个Data同时产生一个新的Interest,当且仅当Consumer不断地发送Interest,producer不断地产生Data,虚拟机在线迁移才能以一个较快的速度进行传输。Stable stage: The data requester receives a piece of data and generates a new interest at the same time. If and only when the consumer continuously sends interest and the producer continuously generates data, the online migration of the virtual machine can be transmitted at a faster speed.

迁移开始后由启动阶段后进入稳定阶段,稳定阶段时,Data和Interest的传输构成两条相反的管道,形成一个数据流环,则这种双向的管道模型使得迁移数据能够不断地从producer发送到consumer。其中,N=BD/L,N为网络中正在传输的Data的最大个数,B是网络带宽,D是网络延迟,L是Data的长度。双向管道的通信模型如图1所示,虚线表示当前正在进行的步骤,实线表示已完成的步骤,可分为以下步骤:After the migration starts, it enters the stable stage from the startup stage. In the stable stage, the transmission of Data and Interest constitutes two opposite pipelines, forming a data flow ring. This bidirectional pipeline model enables the migration data to be continuously sent from the producer to the consumer. Among them, N=BD/L, N is the maximum number of Data being transmitted in the network, B is the network bandwidth, D is the network delay, and L is the length of Data. The communication model of the two-way pipeline is shown in Figure 1. The dotted line indicates the steps currently in progress, and the solid line indicates the completed steps, which can be divided into the following steps:

步骤S1:consumer初始发送2N个Interest;Step S1: The consumer initially sends 2N Interests;

步骤S2:producer接收Interest后连续产生x个Data,x<N;Step S2: Producer generates x data continuously after receiving Interest, x<N;

步骤S3:producer产生并返回N个Data,consumer接收Data并产生新的Interest;Step S3: Producer generates and returns N data, consumer receives Data and generates new Interest;

步骤S4:重复步骤S2、S3,由启动阶段过渡到稳定阶段,consumer接收一个Data同时产生一个新的Interest,X表示producer当前处理结束的Data的序号。Step S4: Repeat steps S2 and S3, transition from the start-up phase to the stable phase, the consumer receives a Data and generates a new Interest at the same time, X represents the sequence number of the Data currently processed by the producer.

双向管道通信步骤包括逆序和丢包处理,逆序的处理为在数据请求方和数据发送方分别设置一个用于保存因逆序提前到达或者提前产生的Data的缓存区。具体为:数据请求方中设置第一Data缓冲区,用于保存因逆序而提前到达的Data,数据发送方中设置第二Data缓冲区,用于保存因逆序而提前产生的Data,数据请求方将收到的正确序号的Data连同其后缓存的提前到达的Data同时写入到虚拟机。The two-way pipeline communication steps include reverse sequence and packet loss processing. The reverse sequence processing is to set a buffer area for saving the data that arrives in advance or is generated in advance due to the reverse sequence, respectively, on the data requester and the data sender. Specifically: the first Data buffer is set in the data requester to save the Data that arrives in advance due to the reverse order, and the second Data buffer is set in the data sender to save the Data that is generated in advance due to the reverse order, and the data requester Write the received Data with the correct serial number together with the cached early-arriving Data to the virtual machine at the same time.

丢包的处理为采用超时重传机制来重新获取Data,当超时Q次后,采用分包的方式将Interest分成若干个小数据块的Interest来请求该Data,则对应Data会相应的被分成若干个小数据块的Data进行返回,分包后的数据包以第三名字空间命名,Q为超时次数阈值。The processing of packet loss is to use the timeout retransmission mechanism to reacquire the Data. When the timeout is Q times, the Interest is divided into several small data blocks to request the Data, and the corresponding Data will be correspondingly divided into several Data of small data blocks are returned, and the subpackaged data packets are named after the third name space, and Q is the timeout threshold.

缓存区中缓存数组的长度Len由网络中正在传输的Data的最大个数N、超时次数阈值Q、超时时间T(超时重传机制的预设值)以及网络延迟D决定。极限情况下,当序号为X的Data一直没有收到,但是序号X之后的Data总是能够收到,此时必须等待超时Q次之后,X对应的Data被分成若干个小的Data进行传输,consumer才能够取回Data。因此,Len必须大于 The length Len of the cache array in the cache area is determined by the maximum number N of data being transmitted in the network, the timeout threshold Q, the timeout time T (the default value of the timeout retransmission mechanism), and the network delay D. In the extreme case, when the data with the serial number X has not been received, but the data after the serial number X can always be received, at this time, it must wait for the timeout Q times, and the data corresponding to X is divided into several small data for transmission. The consumer can retrieve the Data. Therefore, Len must be greater than which is

(3)性能测试步骤:NDN使用名字作为标识符,与位置无关,因此,当虚拟机迁移到新的位置的时候,使用以与位置无关的名字为前缀可以保证迁移结束后通信可以自然地恢复。在虚拟机中运行以与位置无关的名字为前缀(例如采用广播名字前缀作为一种通用的名字前缀)的应用,通过访问该应用验证虚拟机迁移的在线性,同时使用NDN的ping程序每隔Pms发送Interest向虚拟机请求Data响应,根据停机过程中ping程序中丢包数M,获得虚拟机在线迁移的停机时间,停机时间的数值范围为[(M-1)×P,(M+1)×P]ms,误差范围为±Pms,其中,P≥300,实验测试中,分别对P=200ms、250ms、300ms的三种情况进行测试,发现P=200ms、250ms时,因发包速率太快而不能测出M,而P=300ms时,测出的M能够较准确地反应丢包情况,一般情况下P越小越好,但是考虑到实际测试结果,所以P可选取大于等于300,同时P=300ms时为最优选择。(3) Performance testing steps: NDN uses names as identifiers, which have nothing to do with location. Therefore, when a virtual machine is migrated to a new location, using a location-independent name as a prefix can ensure that communication can be naturally restored after the migration . Run an application prefixed with a location-independent name (for example, using a broadcast name prefix as a common name prefix) in the virtual machine, verify the onlineness of the virtual machine migration by accessing the application, and use the NDN ping program every Pms sends Interest to the virtual machine to request a Data response, and according to the number M of packet loss in the ping program during the shutdown process, the downtime of the online migration of the virtual machine is obtained. The value range of the downtime is [(M-1)×P, (M+1 )×P]ms, the error range is ±Pms, wherein, P≥300, in the experimental test, the three cases of P=200ms, 250ms, and 300ms were tested respectively, and it was found that when P=200ms and 250ms, the packet sending rate was too high It is too fast to measure M, and when P=300ms, the measured M can more accurately reflect the packet loss situation. Generally, the smaller the P, the better, but considering the actual test results, so P can be selected to be greater than or equal to 300. At the same time, it is the best choice when P=300ms.

本发明实验分成两个部分:理想环境(局域网)下的虚拟机在线迁移和真实网络环境(广域网)下的虚拟机在线迁移,图2为理想环境下的虚拟机在线迁移实验测试床,图3为真实环境下的虚拟机在线迁移实验测试床。Experiment of the present invention is divided into two parts: the online migration of virtual machine under ideal environment (local area network) and the online migration of virtual machine under real network environment (wide area network), Fig. 2 is the experimental test bed of online migration of virtual machine under ideal environment, Fig. 3 Experimental test bed for online migration of virtual machines in a real environment.

第一部分理想环境的具体实施过程如下所示:The specific implementation process of the first part of the ideal environment is as follows:

步骤一:运行源宿主主机HOST_A和目的宿主主机HOST_B的NDN转发例程NFD。具体命令为:nfd-start。在HOST_B注册FIB表项,具体命令为:nfdc register/ndn/cn/edu/tongji/vm udp4://10.0.2.1;Step 1: Run the NDN forwarding routine NFD of the source host host HOST_A and the destination host host HOST_B. The specific command is: nfd-start. Register FIB entries on HOST_B, the specific command is: nfdc register/ndn/cn/edu/tongji/vm udp4://10.0.2.1;

步骤二:在源宿主主机HOST_A上启动虚拟机HOST_VM,启动参数为:-m 640-smp 1-drive file=/var/lib/libvirt/images/ubuntu.img,if=virtio,format=raw-netnic,vlan=0,macaddr=52:54:00:6d:82:f9-net tap,vlan=0,ifname=tap0,script=no,其中,/var/lib/libvirt/images/ubuntu.img的大小是4G;Step 2: Start the virtual machine HOST_VM on the source host host HOST_A, the startup parameters are: -m 640-smp 1-drive file=/var/lib/libvirt/images/ubuntu.img,if=virtio,format=raw-netnic ,vlan=0,macaddr=52:54:00:6d:82:f9-net tap,vlan=0,ifname=tap0,script=no, where, the size of /var/lib/libvirt/images/ubuntu.img is 4G;

步骤三:在HOST_VM中启动虚拟机中的响应回显请求的服务ndnpingserver,具体命令为:ndnpingserver/ndn/broadcast/tongji/hostVM。在HOST_A上启动NDN的ping客户端,具体命令为:ndnping–i 300/ndn/broadcast/tongji/hostVM。Step 3: Start the service ndnpingserver in the virtual machine that responds to the echo request in HOST_VM. The specific command is: ndnpingserver/ndn/broadcast/tongji/hostVM. Start the NDN ping client on HOST_A, the specific command is: ndnping–i 300/ndn/broadcast/tongji/hostVM.

步骤四:在目的主机HOST_B上启动虚拟机HOST_VM在线迁移的接收进程,启动方式为在步骤二中源宿主主机HOST_A启动虚拟机的启动参数后增加QEMU-KVM的exec方式接收迁移数据的进程:-incoming testconsumer,其中testconsumer为迁移数据接收进程。Step 4: Start the receiving process of the virtual machine HOST_VM online migration on the destination host HOST_B. The starting method is to add the exec method of QEMU-KVM to receive the migration data after the source host host HOST_A starts the virtual machine startup parameters in step 2:- Incoming testconsumer, where testconsumer is the migration data receiving process.

步骤五:在源宿主主机HOST_A的QEMU的控制窗口中输入迁移的启动命令,命令为:migrate–b exec:testproducer,其中参数表明迁移方式为块迁移,testproducer为迁移数据发送进程名。Step 5: Enter the migration startup command in the QEMU control window of the source host host HOST_A, the command is: migrate–b exec:testproducer, where the parameter indicates that the migration method is block migration, and testproducer is the process name for sending the migration data.

步骤六:迁移结束之后,查看源宿主主机HOST_A的用于性能测试的Ping客户端是否仍然可以继续运行,并且查看在停机过程中接收到的响应包和发送的请求包的时间差。Step 6: After the migration is complete, check whether the Ping client used for performance testing of the source host host HOST_A can still continue to run, and check the time difference between the response packet received and the request packet sent during the shutdown process.

第二部分真实环境的具体实施过程除了步骤一与上述理想环境不同,其余步骤都相同,真实环境中以同济大学(Tongji University)与美国韦恩州立大学(Wayne StateUniversity,WSU)之间虚拟机在线迁移为例,真实环境的具体实施过程的步骤一如下:The specific implementation process of the second part of the real environment is the same except that step 1 is different from the ideal environment above, and the rest of the steps are the same. In the real environment, the virtual machine between Tongji University and Wayne State University (WSU) is online. Taking migration as an example, step 1 of the specific implementation process in the real environment is as follows:

步骤一:运行源宿主主机HOST_A和目的宿主主机HOST_B的NDN转发例程NFD。具体命令为:nfd-start。在HOST_A注册FIB表项,具体命令为:nfdc register/ndn udp4://10.0.2.2。在HOST_B注册FIB表项,具体命令为:nfdc register/ndn udp4://192.168.0.1。在Tongji testbed node上注册FIB表项,具体命令为:nfdc register/ndn/cn/edu/tongji/vm/migration udp4://10.0.2.1。在WSU gateway上注册FIB表项,具体命令为:nfdc register/ndn/cn/edu/tongji udp4://202.120.188.176。Step 1: Run the NDN forwarding routine NFD of the source host host HOST_A and the destination host host HOST_B. The specific command is: nfd-start. Register the FIB entry on HOST_A, the specific command is: nfdc register/ndn udp4://10.0.2.2. Register the FIB entry on HOST_B, the specific command is: nfdc register/ndn udp4://192.168.0.1. Register the FIB entry on the Tongji testbed node. The specific command is: nfdc register/ndn/cn/edu/tongji/vm/migration udp4://10.0.2.1. To register the FIB entry on the WSU gateway, the specific command is: nfdc register/ndn/cn/edu/tongji udp4://202.120.188.176.

同时对在实验室理想环境和真实网络环境下(与上述本发明方法的实验环境相同)TCP/IP下的虚拟机在线迁移的性能进行测试,其中实验室理想环境TCP/IP下的虚拟机在线迁移采用UDP通信技术,真实网络环境TCP/IP下的虚拟机在线迁移采用TCP通信技术,最后得到理想环境下和真实环境下NDN与TCP/IP下的虚拟机在线迁移的总时间Total time和停机时间Downtime数值对比表格,如表1、表2。Simultaneously test the performance of the virtual machine online migration under the laboratory ideal environment and the real network environment (identical with the experimental environment of the above-mentioned inventive method) TCP/IP, wherein the virtual machine under the laboratory ideal environment TCP/IP is online UDP communication technology is used for migration, and TCP communication technology is used for online migration of virtual machines under TCP/IP in the real network environment. Finally, the total time and downtime of online migration of virtual machines under NDN and TCP/IP under ideal environment and real environment are obtained Time Downtime value comparison table, as shown in Table 1 and Table 2.

表1理想环境下NDN与TCP/IP下的虚拟机在线迁移的Total time和DowntimeTable 1 Total time and Downtime of online migration of virtual machines under NDN and TCP/IP in an ideal environment

表2真实环境下NDN与TCP/IP下的虚拟机在线迁移的Total time和DowntimeTable 2 Total time and Downtime of virtual machine online migration under NDN and TCP/IP in real environment

网络架构Network Architecture Total timetotal time DowntimeDowntime NDNNDN 7min21sec7min21sec 1.5sec1.5sec TCP/IPTCP/IP 10min10sec10min10sec NPNP

注:NP表示无测量结果。Note: NP means no measurement result.

表1可以发现,NDN下的虚拟机在线迁移的总时间接近于TCP/IP下的迁移的总时间,两者停机时间相同。表2可以发现,在TCP/IP下无法内在地支持虚拟机的在线迁移,但是NDN可以内在地支持虚拟机的在线迁移。同时,NDN的total time小于TCP/IP的total time。通过实验表明NDN下的迁移程序就丢包、延迟、逆序等方面的处理上性能要好于TCP/IP。综上,本发明完成了在未来网络架构NDN下的虚拟机在线迁移及其性能的测量,并通过对比实验验证了本发明的可行性和优点,是面向数据网络中在线迁移的一次重要探索。It can be found from Table 1 that the total time of virtual machine online migration under NDN is close to the total time of migration under TCP/IP, and the downtime of both is the same. Table 2 shows that the online migration of virtual machines cannot be inherently supported under TCP/IP, but NDN can inherently support online migration of virtual machines. At the same time, the total time of NDN is less than the total time of TCP/IP. Experiments show that the migration program under NDN is better than TCP/IP in terms of packet loss, delay, and reverse order processing. In summary, the present invention completes the virtual machine online migration and its performance measurement under the future network architecture NDN, and verifies the feasibility and advantages of the present invention through comparative experiments. It is an important exploration for online migration in data networks.

Claims (8)

1. a kind of realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is characterised in that including following Step:
(1) name space design procedure in NDN:Build NDN and carry out Data Migration, in NDN data packet be divided into Interest and Data, definition are used for three name spaces for naming data packet:
First name space, for identifying the data packet in transition process;
Second name space, last data packet terminated for identifying migration;
Third word space, in data migration process because data packet is excessive and during packet loss, after which is divided into several, use Third word space is identified;
(2) two-way pipeline communication step:By two-way pipeline communication data packet between request of data side and data sender, specifically For:Request of data side sends Interest in a manner of pipeline to data sender, and data sender produces corresponding after receiving Data, and Data is sent to request of data side in a manner of pipeline, request of data side produces newly after receiving the Data of return Interest, the data packet in data migration process is named with the first name space, when what is named with the second name space After the data packet end of transmission, Data Migration terminates;
Two-way pipeline communication step includes backward and packet loss is handled, and the processing of backward is in request of data side and data sender point Not She Zhi one be used to preserve because backward advances to the buffer area of the Data for reaching or producing in advance, the processing of packet loss is using super When retransmission mechanism reacquire Data, when Q times overtime after, by the way of subpackage Interest is divided into several small datas The Interest of block asks the Data, and the data packet after subpackage is named with third word space, and Q is overtime time threshold value;
(3) performance test step:Application of the operation using the name unrelated with position as prefix in virtual machine, should be answered by accessing With verifying virtual machines migration in downtime that is linear, while using the ping utility measurement virtual machine of NDN.
2. according to claim 1 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is that the form of first name space is /ndn/location/vm/migration/live/version/data/ Sequence, wherein ,/ndn is the root components common to name space in NDN, and/location is represented where data sender Equipment or gateway name ,/vm shows that the name belongs to and shows the name with the relevant name of virtual machine ,/migration It is that used name is applied in migration ,/live shows that the name is to migrate online, and/version shows that the online migration name is empty Between version number ,/data shows that the name is to identify the name of migrating data, and/sequence shows the sequence number of migrating data, uses In the data block that the different virtual machine of unique mark migrates online.
3. according to claim 2 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is that the form of second name space is /ndn/location/vm/migration/live/version/data/ End/last, wherein ,/end shows that migrating data is over, and/last identifies last sequence number of migrating data block, and The numerical value of sequence is not more than the numerical value of last.
4. according to claim 2 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is that the form of the third word space is /ndn/location/vm/migration/live/version/data/ Sequence/frag/nums/seq, wherein ,/frag shows that the name is the name of a subpackage, and the component of/nums shows one A data coating divide into several, and/seq shows the sequence number of the name subpackage.
5. according to claim 1 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is that the two-way pipeline communication data packet includes two stages:Startup stage and stabilization sub stage;
Startup stage:Request of data side sends 2N Interest, and data sender produces and returns to N number of data;
Stabilization sub stage:Request of data side receives a Data and produces a new Interest at the same time;
Entering the stabilization sub stage after startup stage, during the stabilization sub stage, the transmission of Data and Interest form two opposite pipelines, A data flow ring is formed, wherein, N=BD/L, N are the maximum number of Data transmitted in network, and B is network bandwidth, D It is network delay, L is the length of Data.
6. according to claim 1 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is, maximum number N by the Data that is being transmitted in network of the length Len of array, time-out time are cached in the buffer area Number threshold value Q, time-out time T and network delay D are determined.
7. according to claim 1 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is that the processing of the backward is specially:First Data buffering areas are set in request of data side, advance to what is reached for preserving Data, the 2nd Data buffering areas are set in data sender, and for preserving the Data produced in advance, request of data side will be received The Data of correct sequence number is written to virtual machine at the same time together with the Data that advanceing to of caching thereafter reaches.
8. according to claim 1 realize the method that virtual machine migrates online under NDN based on two-way pipe method, it is special Sign is, in performance test step, sends Interest every Pms using the ping utility of NDN and is rung to virtual machine request Data Should, according to number of dropped packets M in ping utility in stopping process, obtain the downtime that virtual machine migrates online, the number of downtime Value scope is [(M-1) × P, (M+1) × P] ms, and error range is ± Pms, wherein, P >=300.
CN201510093108.4A 2015-03-02 2015-03-02 The method that virtual machine migrates online is realized based on two-way pipe method under NDN Expired - Fee Related CN104754033B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510093108.4A CN104754033B (en) 2015-03-02 2015-03-02 The method that virtual machine migrates online is realized based on two-way pipe method under NDN

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510093108.4A CN104754033B (en) 2015-03-02 2015-03-02 The method that virtual machine migrates online is realized based on two-way pipe method under NDN

Publications (2)

Publication Number Publication Date
CN104754033A CN104754033A (en) 2015-07-01
CN104754033B true CN104754033B (en) 2018-04-17

Family

ID=53593121

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510093108.4A Expired - Fee Related CN104754033B (en) 2015-03-02 2015-03-02 The method that virtual machine migrates online is realized based on two-way pipe method under NDN

Country Status (1)

Country Link
CN (1) CN104754033B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104184803B (en) * 2014-08-08 2017-12-15 同济大学 The method that virtual machine migrates online in adaptive completion XIA networks based on DAG

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103546556A (en) * 2013-10-22 2014-01-29 同济大学 A method for virtual machine online migration in future network XIA
CN103577249A (en) * 2013-11-13 2014-02-12 中国科学院计算技术研究所 Method and system for virtual machine online migration
CN103716221A (en) * 2013-12-10 2014-04-09 同济大学 On-line migration method of virtual machine in wide area network under future network NDN (Named Data Network)

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8891541B2 (en) * 2012-07-20 2014-11-18 International Business Machines Corporation Systems, methods and algorithms for named data network routing with path labeling

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103546556A (en) * 2013-10-22 2014-01-29 同济大学 A method for virtual machine online migration in future network XIA
CN103577249A (en) * 2013-11-13 2014-02-12 中国科学院计算技术研究所 Method and system for virtual machine online migration
CN103716221A (en) * 2013-12-10 2014-04-09 同济大学 On-line migration method of virtual machine in wide area network under future network NDN (Named Data Network)

Also Published As

Publication number Publication date
CN104754033A (en) 2015-07-01

Similar Documents

Publication Publication Date Title
US10491523B2 (en) Load distribution in data networks
CN102801784B (en) A kind of distributed data storage method and equipment
CN102316043B (en) Port virtualization method, switch and communication system
CN104702592B (en) Stream media downloading method and device
CN110427270B (en) Dynamic load balancing method for distributed connection operator in RDMA (remote direct memory Access) network
CN107634984A (en) A file synchronization method based on one-way transmission channel
WO2019052058A1 (en) Domain name redirecting method and system
CN104866528B (en) Multi-platform data acquisition method and system
CN102209019A (en) Load balancing method and load balancing equipment based on message payload
CN104753814B (en) Packet distribution processing method based on network card
CN100414936C (en) Method for Balancing the Load among Multiple Network Cards of Network File System Server
EP4136809B1 (en) Distributed flow processing and flow cache
CN103179162B (en) A kind of method and system of output journal
CN103716221B (en) On-line migration method of virtual machine in wide area network under future network NDN (Named Data Network)
CN102075584B (en) Distributed file system and access method thereof
CN104754033B (en) The method that virtual machine migrates online is realized based on two-way pipe method under NDN
CN104660636B (en) Point-to-point application identifying processing method and apparatus
CN103002038B (en) The processing method of elastic load equilibrium and device
WO2015062390A1 (en) Virtual machine migration method, device and system
JP5716745B2 (en) Data transfer system
JP2011258014A (en) Relay processing method, program and relay device
CN103078968A (en) Domain name querying method, IP (Internet Protocol) grouping method, device and equipment
CN103414768B (en) Data processing method, system and gateway
CN103414656B (en) Message transmission control method and network interface card
CN105227659B (en) A kind of network communication method and equipment

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20180417

Termination date: 20210302

CF01 Termination of patent right due to non-payment of annual fee