CN103825963B - Virtual Service moving method - Google Patents

Abstract

The invention relates to a virtual service migration method, comprising: selecting migration evaluation parameters according to service types and user needs, a monitor monitors virtual nodes in the network, and when a new request arrives or a user request changes, the evaluation parameters of the virtual node are obtained State information of parameters; calculate the service migration cost of each virtual node according to the state information; obtain the virtual node with the smallest virtual service migration cost from the virtual nodes as a service node through the migration cost calculation method; select the service node pair The virtual service is migrated. The virtual service migration method provided by the present invention can effectively integrate the influence of various factors on the migration, judge and execute the migration, realize network resource management and improve user service experience quality by using the service migration.

Description

Virtual Service Migration Methods

technical field

The invention relates to the field of mobile Internet, in particular to a virtual service migration method.

Background technique

With the increasing development of science and technology, the mobile Internet has become one of the main trends in the development of the Internet. According to statistics, as of March 2013, the number of fixed Internet broadband users in my country was 181 million, while the number of mobile Internet users had reached 817 million. With the increasing demand for network mobility, changes in user behavior, continuous enrichment of information interaction types, and increasing data traffic, ensuring and improving user service experience quality has become a new challenge for the Internet field. The development of virtualization technology provides the possibility to solve the problem of mobility. The goal of virtualization is to realize the smooth movement of virtual services without considering the properties of the underlying physical network, realize the on-demand allocation of network resources, and improve the quality of user experience. FIG. 1 is a schematic diagram of network virtualization in the prior art. As shown in Figure 1, virtualization technology supports multiple virtual networks (Virtual Networks, VNs) on a common underlying physical network (Substrate Network, SN) through abstraction, separation, and isolation mechanisms. Each virtual network can be used independently of each other. According to the dynamically changing user needs, the node resources and link resources in the entire network are rationally allocated to maximize the advantages of resource sharing, maximize the utilization of network resources, and obtain maximum benefits.

FIG. 2 is a schematic diagram of Flowvisor implementing virtualization in the prior art. As shown in Figure 2, the network virtualization platform FlowVisor generates independent network slices by dividing the flow table space, effectively realizing network virtualization and dividing the physical network into multiple network slices. The network traffic on each network slice is isolated from each other, and the bandwidth, CPU usage, flow table configuration, etc. are managed. Users can conduct experimental research on various traffic models and protocol innovations that do not interfere with each other on each shard. At present, FlowVisor has been deployed in some large campuses such as Stanford University in the United States, and the famous future network test bed GENI and Internet2 projects are also using FlowVisor for virtualization management.

Fig. 3 is a schematic diagram of a network virtualization layered service provisioning model in the prior art. As shown in Figure 3, in a virtualized environment, the service provider (Service Provider, SP) describes the required resources (network resources, computing resources, storage resources, etc.) to the underlying layer in a certain form according to the needs of the user (User) Infrastructure Provider (InfrastructureProvider, InP). The infrastructure provider deploys and manages the underlying physical resources, and selects corresponding resources from the virtual resource pool to complete the creation of the virtual network. Service providers lease network resources from infrastructure providers to provide users with the required services.

Due to the addition of some new users, the removal of some old users, and the changes of some users' positions in the network, the number of users in the network, changes in user behavior, user preferences, or changes in some underlying networks, virtual services need to be adjusted according to the changes. Adjust and migrate. How to adjust the scale and resource distribution of the virtual network in a timely manner according to these changes, ensure the quality of service (QoS) and quality of experience (QoE) of users, and obtain reliable and stable network services with the minimum delay , is one of the challenges in implementing virtual services. The virtualization in the present invention mainly refers to the virtualization of the server/host, providing virtualized shareable and reusable software and hardware resources and information to computers and other devices on demand.

Fig. 4 is a schematic diagram of user migration in the prior art. As shown in Figure 4, service migration needs to consider the balance of various costs. When the virtual node closer to the user is used as the service provider, the delay of the service may be smaller, and the quality of service and quality of experience are higher. But at the same time, the migration will also bring other overheads, which will have negative impacts, such as the pressure on the network during the migration of a large amount of data transmission, and may even cause service interruption. The problem of service migration is how to reasonably adjust the location of network services, more effectively save network resources, reduce service response time and cost, and improve user experience.

In the existing technology, the research and application of using service migration to realize network resource management and reduce energy consumption in a virtualized environment is still relatively preliminary, such as J Grassler, S Schmid and others in "at 32nd IEEEConference on Computer Communications (INFOCOM Demo), Turin, Italy, April2013." "Move-with-the-Sun or Move-with-the-Moon? Wide-AreaCloudNet Migrations UnderLatency and Resource Constraints.Demo" research. Service migration in a virtualized environment is mainly divided into two application scenarios: single-domain and multi-domain.

In a single-domain environment, there are less differences between nodes in terms of resource type and quality, and less cost is required for migration. However, when the nodes in the domain cannot meet the needs of users, they need to be migrated to nodes in other domains to provide services. There are not only differences between nodes in multiple domains, but also additional roaming costs due to cross-domain.

The following is related to the virtual service balance migration algorithm (MIG) proposed by M Bienkowski, A Feldmann et al. in "In Proc.ACM SIGCOMMVISA, 2010.""Competitive analysis for service migration in vnets" and D Arora, M Bienkowski and others briefly introduced the cross-domain balancing algorithm (MIX _k ) proposed in "Proceedings of the 5th International Conference on Principles, Systems and Applications of IPTelecommunication, 2011.""Online strategies for intra and inter provider service migration in virtual networks".

(1) Virtual service balance migration algorithm (MIG)

The virtual service balance migration algorithm (MIG) quantifies the parameters of migration costs and benefits, and judges the timing of migration through dynamic comparison. The basic idea of the MIG algorithm is to achieve a balance between the migration cost _Mig and the income Cost _acc when the migration occurs, and select a feasible and better service provider.

When the user's location changes, the delay from the virtual server side to the client side will increase, which will affect the service quality of some services. Through migration, the optimization of service delay can be realized, that is, service delay is part of the migration benefit Cost _acc . In addition, whether the migration can be realized is also related to the available load of the server. If the potential node does not meet the user's needs, the migration cannot proceed. When migrating, under the same conditions, a node with a larger available server load should be selected for migration. Therefore, at time t, the revenue Cost _acc generated by migrating the request sequence R _t can be expressed as

In this paper, it is assumed that the available load of all servers can meet the needs of users, and the Cost _acc is simplified to obtain

Due to the influence of the bandwidth on the migration path and the size of the service itself, it is necessary to carefully consider whether to perform service migration. The size(s) of the service itself and the bandwidth w(p) of the migration path jointly determine the time required for migration. In this paper, the migration cost of _migrating service s is expressed as

In the virtual service balance migration algorithm, only the service migration problem in one infrastructure provision domain is considered, so the cost of service migration Cost _mig can be simplified as

Cost _mig (u,v)＝max _e size(s)/w(e) (4)

The virtual service balance migration algorithm divides the time into multiple time slots, assuming that the initial service provider is node v. When the request arrives, the time slot starts, and the cost _acc (v) due to the change of the request is calculated. If other virtual network nodes, such as node u, are used as service providers, that is, service migration occurs, and the migration cost Cost _mig (v, u) will be generated. Let β = max _u {Cost _mig (u, v)}. If Cost _acc (v) > β is satisfied, migration occurs, and a node is randomly selected from the solution set satisfying the inequality Cost _acc (u) < β as the service provider of the request. If there is no such node u, no migration is required and the slot ends. When the next request arrives, a new slot starts and the Cost _acc (v) is recalculated.

(2) Cross-domain balance algorithm (MIX _k )

The cross-domain balance algorithm (MIX _k ) is an improvement and optimization of the virtual service balance migration algorithm, considering the situation that the service provider nodes are located in multiple virtual networks. The request delay is Cost _acc (v), and the migration cost β=max _u {Cost _mig (u, v)}. Assuming that the additional cost of crossing a domain is π (where π≥β ² ), the roaming cost of crossing k domains is k*π.

Assuming that the initial service provider is node v, when the service request sequence arrives, the migration situation in a domain is first considered. Calculate the revenue Cost _acc (v) generated by the request, and compare it with the migration cost β=max _u {Cost _mig (u,v)}. If Cost _acc (v) > β is satisfied, a node is randomly selected from the solution set satisfying the inequality Cost _acc (u) < β as the destination node for migration. If there is no such node u in the domain, the service migration in the case of cross-domain is considered. Migrate if the benefit Cost _acc (v) is better than the additional cost caused by cross-domain.

To sum up, the migration methods in the prior art have the following problems: the MIG migration algorithm only simply considers two influencing factors of service delay and link bandwidth, but there are many factors affecting the selection of migration nodes. The service delay is simply quantified as the number of hops between the service request access node and the service provider. The quantification of the two factors is too simple and rigid, which is not conducive to the dynamic adjustment of migration timing. Moreover, the factors that determine the migration are interacted with each other, and the simple comparison of the size cannot reflect the relationship between the various factors. Although the roaming cost due to cross-domain is added to the MIX _k algorithm, there are still problems of simple quantification and interrelationship described above.

Contents of the invention

The purpose of the present invention is to provide a virtual service migration method based on a fair, just and dynamic QoS calculation model for the above problems.

To achieve the above object, the present invention provides a virtual service migration method, the method comprising:

Select the migration evaluation parameters according to the service type and user needs, monitor the virtual nodes in the network, and obtain the status information about the evaluation parameters of the virtual nodes when a new request arrives or a user request changes;

Calculate the service migration cost of each virtual node according to the state information;

Obtaining the virtual node with the smallest virtual service migration cost from the virtual nodes as a service node through the calculation of the service migration cost of each virtual node;

Selecting the service node to migrate the virtual service.

Preferably, the state information includes service delay, available load of the server, bandwidth on the migration path, size of the service itself, cost of service interruption and recovery, service provider credit and execution price.

Preferably, the calculating the service migration cost of each virtual node according to the state information specifically includes:

A calculation model based on QoS is established and generated according to the state information.

Preferably, said establishing and generating a QoS-based calculation model according to said state information specifically includes:

According to the state information, the matrix Q is obtained:

Among them, m and n are integers, q _{n and m} are elements in the matrix Q, C(v _i ) represents the migration cost of the virtual node v _i , each row in the matrix Q represents the state information of the virtual node, and the matrix Each column in Q represents a state information that affects the migration cost.

Preferably, the establishment and generation of the QoS-based calculation model according to the state information specifically further includes:

normalizing the matrix Q, and standardizing the state information of different dimensions into dimensionless standardized parameters to form a unified measurement standard;

The standardized parameters are grouped, and each group includes a variety of normalized parameters shown, and are operated on a group basis to obtain migration cost information of each virtual node after grouping.

Preferably, the normalization of the matrix Q, the normalization of the state information of different dimensions into dimensionless standardized parameters, and the formation of a unified measurement standard specifically include:

The first matrix is represented by N={n ₁ ,n ₂ ,...,n _m }, where n _j =0 or 1, 1≤j≤m;

The second matrix is represented by C={c ₁ ,c ₂ ,...,c _m }, c _j is a constant, 1≤j≤m;

Normalize each element in the matrix Q with the following formula:

in, is the average value of the jth cost standard in the matrix Q, and the matrix Q' is obtained:

Wherein, q _{i, j} are elements in the matrix Q, and q _{n, m} ' are elements in the matrix Q'.

Preferably, the method further includes: not migrating the virtual node when no new request arrives or the transmission of the user request does not change.

The beneficial effects brought by the present invention are:

1. Use the method based on the QoS calculation model to calculate the cost of service migration, avoiding the problem that the quantification of influencing factors is too simple and rigid, which is not conducive to the dynamic adjustment of migration timing;

2. Effectively integrate the impact of various factors on the migration cost, allowing the addition of custom input parameters, which can be dynamically adjusted according to the virtual network environment and user needs;

3. Considering the influencing parameters related to the nature of the virtual node itself, the credit and price of the virtual node are increased, and the service quality of the virtual service node is evaluated more comprehensively and realistically;

4. Select the node with the least migration cost for migration, which is beneficial to minimize the cost of migration, improve the user's service experience quality, and realize green energy-saving and efficient virtual resource migration and scheduling.

Description of drawings

FIG. 1 is a schematic diagram of network virtualization in the prior art of the present invention;

Fig. 2 is the schematic diagram that Flowvisor realizes virtualization in the prior art of the present invention;

3 is a schematic diagram of a network virtualization layered service provision model in the prior art of the present invention;

FIG. 4 is a schematic diagram of user migration in the prior art of the present invention;

FIG. 5 is a flowchart of a virtual service migration method in an embodiment of the present invention;

Fig. 6 is a schematic diagram of monitoring each virtual node in the network by a local monitor in an embodiment of the present invention;

FIG. 7 is a flow chart of calculating the service migration cost of each virtual service node in an embodiment of the present invention.

detailed description

The technical solutions of the present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

FIG. 5 is a flowchart of a virtual service migration method in an embodiment of the present invention.

As shown in FIG. 5 , first in step 501 , migration evaluation parameters are selected according to service types and user requirements. The monitor monitors the virtual nodes in the network, and when a new request arrives or a user request changes, the status information about the evaluation parameters of the virtual node is obtained.

Specifically, the cost of a virtual node/virtual server is jointly affected by multiple factors. In order to simplify the problem, an embodiment of the present invention only considers the situation that there is only one virtual network on the physical network, and the situation that there is only one virtual node/server in the virtual network.

Table 1 shows the main factors that affect the cost of migration.

Table 1

Influencing parameters Parameter meaning C _delay service delay C _load Server's available load C _bandwidth Bandwidth on Migration Path C _size the size of the service itself C _interrupt Cost of Service Interruption and Restoration C _reputation service provider credit C _price execution price

Among them, in order to select a virtual node with better service quality as a service provider, the present invention adds two attributes related to the nature of the service provider itself, namely C _reputation and C _price .

C _reputation is the credit of the service provider, which is the main evaluation index of the trustworthiness of virtual nodes. It depends on the end user's historical service experience. For the same service provider, different end users will have different evaluations. The credit of a virtual node is defined as the average value of multiple end-user feedback evaluations, namely Among them, n is the total number of times the node is evaluated, R _i is the feedback value of the end user to the node credit, and R _i is an integer belonging to [0,5].

C _price is the execution price. When the virtual node provides the service, the user needs to pay the required fee. Under the same circumstances, requesters are more inclined to choose nodes with lower prices to provide services.

Fig. 6 is a schematic diagram of monitoring each virtual node in the network by a local monitor in an embodiment of the present invention.

As shown in Figure 6, the monitor monitors each virtual node in the network and collects state information on the virtual node. Wherein, the state information includes the state information shown in Table 1, such as delay time, link bandwidth, and available load of nodes.

Specifically, in an embodiment of the present invention, a set of all virtual service providers is represented by V={v ₁ , v ₂ ,...,v _n }, and at time t _k , a request δ _k arrives. The virtual service provider in the previous stage is denoted by _sk-1 . The goal of this embodiment is to determine whether the virtual service provider needs to be migrated to a new virtual node and which node to migrate to when the request δ _k arrives at time t _k . When migration is required, we choose the node with the least migration cost for migration. The migration cost of node v _i is represented by C(v _i ). Using m criteria to evaluate the cost of nodes, the following matrix Q can be obtained:

Matrix Q is a matrix with n rows and m columns, and both n and m are integers. Among them, the i-th row in the matrix Q represents various indicators of the virtual service node v _i , and each column represents a factor that affects the cost.

As shown in FIG. 5 , in the next step 502 , the service migration cost of each virtual node is calculated according to the state information. The specific steps of the calculation method are as follows:

Fig. 7 is a flow chart of calculating the service cost of each virtual service node in an embodiment of the present invention.

As shown in FIG. 7 , firstly, the matrix Q is normalized in step 701 , and state information of different dimensions is standardized into dimensionless standardized parameters to form a unified measurement standard.

Specifically, the purpose of normalizing the matrix Q is: (1) to allow a unit-independent, standard migration cost measurement method; (2) to provide each virtual service provider with a unified indicator for evaluating the migration cost; (3) Set the threshold.

First, define two matrices. The first matrix N={n ₁ ,n ₂ ,...,n _m } takes 1≤j≤m, where n _j =0 or 1. If the cost of migration increases as q _ij increases, set n _j = 1 (for example, the larger the parameter C _size , the greater the migration cost). On the contrary, n _j =0 (for example, the node with larger C _aload is more inclined to migrate). The second matrix C={c ₁ ,c ₂ ,...,c _m }, where c _j is a constant, is set to the normalized maximum value of each cost evaluation index. Each element in Q is normalized by formula (6) (7).

In the above formula, is the average value of the jth cost standard in the matrix Q. Operate Q through the above formula to get a new matrix Q', as shown in formula (8):

In the next step 702, the standardized parameters are grouped, and each group includes a variety of normalized parameters shown, and are operated on a group basis to obtain the migration cost information of each virtual node after grouping.

Specifically, the cost parameters of the virtual nodes are grouped, each group can contain multiple parameters, and are operated in groups. In an embodiment of the present invention, it can be divided into three parts: access cost Cost _acc , migration cost Cost _mig , and server cost Cost _{pri .} Delay and available load are part of the access cost, Cost _mig , and server prices and credits are part of the server cost, Cost _pri . In this embodiment, a matrix D is introduced, and the matrix D is used to define the relationship between cost indicators and cost groups.

Wherein, each row in the matrix D represents a cost index factor, and each column represents a cost grouping, where l is the total number of groups. In the matrix D, if the i-th cost index in Q' belongs to the j-th group, then d _i,j =1, otherwise d _i,j =0.

Matrix G is the cost information of each virtual service node after grouping.

Among them, each row in G represents a network service provider, and each column represents the value of a cost group.

The matrix G can be calculated by the following formula:

G=Q, *D (11)

To normalize the matrix G, first define two new matrices: matrix T and matrix F. Matrix T={t ₁ ,t ₂ ,...,t _l }, where the constant t _j represents the normalized threshold of each group. G is normalized according to formula (12), and the obtained result is G'.

As shown in FIG. 7 , in the last step 703 , the migration costs of all virtual nodes are comprehensively calculated according to group weights. Define the matrix F={f ₁ ,f ₂ ,...,f _l }, where f _j represents the weight of the cost evaluation group j in the total evaluation, which can be used to express the user's preference for the evaluation criteria of the j group. The calculation of the cost of node s _i providing virtual services can be obtained by formula (14):

In the following step 503, the virtual node with the smallest virtual service migration cost is obtained from the virtual nodes as the service node through the migration cost calculation method.

In an embodiment of the present invention, a virtual network preferably has 6 nodes, and the set of nodes is V={v ₁ , v ₂ , . . . , v ₆ }. The initial service node S ₀ is node v ₁ , and at time t ₁ , the request sequence δ ₁ arrives. Five migration cost evaluation parameters are optimized: service delay C _delay , server available load C _aload , migration path bandwidth C _bandwidth , price C _price and credit C _reputation . The matrix Q of evaluation parameters of each node is obtained through the local monitor:

Define the threshold C=(5,5,5,5,5), and get N=(1,0,0,1,0) according to the five evaluation indicators. Through the above formulas (6) and (7), the normalized matrix Q' is calculated:

In the embodiment, the five evaluation indicators are divided into three categories, namely, the access cost Costacc, the migration cost Costmig, and the server cost Costpri. Then the matrix D is:

The matrix G calculated by formula (11) is:

In the embodiment, the threshold matrix T=(3,3,3) is defined, and G' is obtained through normalization by formula (12):

The weights of the three types of evaluation criteria in the overall migration evaluation are defined as 0.4, 0.4, and 0.2, respectively. Then the migration costs of the six service nodes calculated by the above formula (14) are 1.1941, 1.0616, 1.2203, 1.0025, 0.6944 and 0.8271 respectively.

Specifically, the migration costs of the six service nodes obtained from the above calculations are 1.1941, 1.0616, 1.2203, 1.0025, 0.6944, and 0.8271, respectively, among which the migration cost of 0.6944 is the smallest, and the virtual node v ₅ corresponding to 0.694 is selected as the service provider node.

Returning to FIG. 5 , in the last step 504 , the service node is selected to migrate the virtual service, that is, the service node at the previous moment is migrated to the virtual node with the least migration cost.

Specifically, among the above six service nodes, the service node v ₁ at the previous moment is migrated to v ₅ , that is, the virtual service node corresponding to the minimum migration cost of 0.6944. To sum up, the present invention provides a virtual service migration method based on an open, fair and dynamic QoS calculation model. Among them, QoS makes the needs of the service requester and the service provider agree according to the available network resources. QoS includes the possibility of a service responding to a request at a given time, how well the service performs tasks, how fast the service runs, and the reliability and security of the service. Therefore, the QoS parameters that affect services include throughput, delay time, execution time, reliability, cost, security, credit, etc. The units of various parameters are different, and the values are very different. There is no comparability between the indicators, and the result of direct comparison is inaccurate. The parameter information needs to be normalized and comprehensively calculated, so as to allow the service requester to choose a service with higher service quality among multiple services that meet the requirements.

The virtual service migration method provided by the present invention applies the QoS calculation idea to the calculation of the service migration cost. In the process of selecting the destination migration node, not only the two influencing parameters of service delay and link bandwidth are considered, but also the availability of the server is considered. Load, price and other parameters that affect the cost of service migration are calculated comprehensively, and the node with the smallest migration cost is selected as the service provider of the request sequence. It can effectively integrate the impact of various factors on migration, judge and execute migration, and use service migration to realize network resource management and improve user service experience quality.

Professionals should further realize that the units and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the relationship between hardware and software Interchangeability. In the above description, the composition and steps of each example have been generally described according to their functions. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.

The steps of the methods or algorithms described in connection with the embodiments disclosed herein may be implemented by hardware, software modules executed by a processor, or a combination of both. Software modules can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other Any other known storage medium.

The specific embodiments described above have further described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above descriptions are only specific embodiments of the present invention and are not intended to limit the scope of the present invention. Protection scope, within the spirit and principles of the present invention, any modification, equivalent replacement, improvement, etc., shall be included in the protection scope of the present invention.

Claims (5)

1. A virtual service migration method, characterized in that the method comprises: Select the migration evaluation parameters according to the service type and user needs, monitor the virtual nodes in the network, and obtain the status information of the virtual nodes about the migration evaluation parameters when a new request arrives or a user request changes; Calculate the service migration cost of each virtual node according to the state information; By calculating the service migration cost of each virtual node, obtain the virtual node with the smallest service migration cost from the virtual nodes as a service node; Selecting the service node to migrate the virtual service; The status information includes the service delay, the available load of the server, the bandwidth on the migration path, the size of the service itself, the cost of service interruption and recovery, service provider credit and execution price; The calculation of the service migration cost of each virtual node according to the state information specifically includes: A QoS-based calculation model is established according to the state information. 2. The virtual service migration method according to claim 1, wherein said establishing a QoS-based computing model according to said state information specifically comprises: According to the state information, the matrix Q is obtained: <mrow><mi>Q</mi><mo>=</mo><mfenced open = "(" close = ")"><mtable><mtr><mtd><msub><mi>q</mi><mrow><mn>1</mn><mo>,</mo><mn>1</mn></mrow></msub></mtd><mtd><msub><mi>q</mi><mrow><mn>1</mn><mo>,</mo><mn>2</mn></mrow></msub></mtd><mtd><mo>...</mo></mtd><mtd><msub><mi>q</mi><mrow><mn>1</mn><mo>,</mo><mi>m</mi></mrow></msub></mtd></mtr><mtr><mtd><msub><mi>q</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn></mrow></msub></mtd><mtd><msub><mi>q</mi><mrow><mn>2</mn><mo>,</mo><mn>2</mn></mrow></msub></mtd><mtd><mo>...</mo></mtd><mtd><msub><mi>q</mi><mrow><mn>2</mn><mo>,</mo><mi>m</mi></mrow></msub></mtd></mtr><mtr><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr>mtr></mtable></mtd><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd></mtr><mtr><mtd><msub><mi>q</mi><mrow><mi>n</mi><mo>,</mo><mn>1</mn></mrow></msub></mtd><mtd><msub><mi>q</mi><mrow><mi>n</mi><mo>,</mo><mn>2</mn></mrow></msub></mtd><mtd><mo>...</mo></mtd><mtd><msub><mi>q</mi><mrow><mi>n</mi><mo>,</mo><mi>m</mi></mrow></msub></mtd></mtr></mtable></mfenced></mrow> Among them, m and n are integers, m is the number of types of state information, n is the number of virtual nodes, q _{i, j} are elements in the matrix Q, i is an integer not greater than n, and j is an integer not greater than m , use C(v _i ) to represent the service migration cost of the virtual node v _i , the i-th row in the matrix Q represents the state information of the virtual node v _i , and each column in the matrix Q represents the factors affecting the service migration cost A status message. 3. The virtual service migration method according to claim 2, wherein said establishing a QoS-based computing model according to said state information specifically further comprises: normalizing the matrix Q, and standardizing the state information of different dimensions into dimensionless standardized parameters to form a unified measurement standard; The standardized parameters are grouped, and each group includes a variety of standardized parameters, and are operated according to the group to obtain the service migration cost information of each virtual node after grouping. 4. The virtual service migration method according to claim 3, wherein the matrix Q is normalized, and the state information of different dimensions is standardized into dimensionless standardized parameters to form a unified The measurement criteria specifically include: The first matrix is represented by N={n ₁ ,n ₂ ,...,n _m }, where m is the number of types of state information, 1≤j≤m, if the q _i,j increases, If the service migration cost becomes larger, then n _j =1, otherwise, n _j =0; The second matrix is represented by C={c ₁ ,c ₂ ,...,c _m }, c _j is a constant, set to the maximum value of normalization of each state information, m is the number of types of state information, 1 ≤j≤m; Normalize each element in the matrix Q with the following formula: <mrow><msubsup><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow><mo>&amp;prime;</mo></msubsup><mo>=</mo><mfenced open = "{" close = ""><mtable><mtr><mtd><mrow><mfrac><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mrow><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mrow></mfrac><mo>,</mo><mi>i</mi><mi>f</mi><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mo>&amp;NotEqual;</mo><mn>0</mn><mo>,</mo><mi>a</mi><mi>n</mi><mi>d</mi><mo>,</mo><mfrac><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mrow><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mrow></mfrac><mo>&lt;</mo><msub><mi>c</mi><mi>j</mi></msub><mo>,</mo><mi>a</mi><mi>n</mi><mi>d</mi><mo>,</mo><msub><mi>n</mi><mi>j</mi></msub><mo>=</mo><mn>1</mn></mrow></mtd></mtr><mtr><mtd><mrow><msub><mi>c</mi><mi>j</mi></msub><mo>,</mo><mi>i</mi><mi>f</mi><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mo>=</mo><mn>0</mn><mo>,</mo><mi>a</mi><mi>n</mi><mi>d</mi><mo>,</mo><msub><mi>n</mi><mi>j</mi></msub><mo>=</mo><mn>1</mn><mo>,</mo><mi>o</mi><mi>r</mi><mo>,</mo><mfrac><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mrow><mfrac><mn>1</mn><mi>n</mi></msub>mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mrow></mfrac><mo>&amp;GreaterEqual;</mo><msub><mi>c</mi><mi>j</mi></msub></mrow></mtd></mtr></mtable></mfenced></mrow> <mrow><msubsup><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow><mo>&amp;prime;</mo></msubsup><mo>=</mo><mfenced open = "{" close = ""><mtable><mtr><mtd><mrow><mfrac><mrow><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mrow><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mfrac><mo>,</mo><mi>i</mi><mi>f</mi><mo>,</mo><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mo>&amp;NotEqual;</mo><mn>0</mn><mo>,</mo><mi>a</mi><mi>n</mi><mi>d</mi><mo>,</mo><msub><mi>n</mi><mi>j</mi></msub><mo>=</mo><mn>0</mn><mo>,</mo><mi>a</mi><mi>n</mi><mi>d</mi><mo>,</mo><mfrac><mrow><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mrow><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mfrac><mo>&lt;</mo><msub><mi>c</mi><mi>j</mi></msub></mrow></mtd></mtr><mtr><mtd><mrow><msub><mi>c</mi><mi>j</mi></msub><mo>,</mo><mi>i</mi><mi>f</mi><mo>,</mo><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub><mo>=</mo><mn>0</mn><mo>,</mo><mi>a</mi><mi>n</mi><mi>d</mi><mo>,</mo><msub><mi>n</mi><mi>j</mi></msub><mo>=</mo><mn>0</mn><mo>,</mo><mi>o</mi><mi>r</mi><mo>,</mo><mfrac><mrow><mfrac><mn>1</mn><mi>n</mi></mfrac><msubsup><mi>&amp;Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>n</mi></msubsup><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mrow><msub><mi>q</mi><mrow><mi>i</mi><mo>,</mo><mi>j</mi></mrow></msub></mfrac><mo>&amp;GreaterEqual;</mo><msub><mi>c</mi><mi>j</mi></msub></mrow></mtd></mtr></mtable></mfenced></mrow> in, is the average value of the jth state information in the matrix Q, and the matrix Q' is obtained: <mrow><msup><mi>Q</mi><mo>&amp;prime;</mo></msup><mo>=</mo><mfenced open = "(" close = ")"><mtable><mtr><mtd><mrow><msup><msub><mi>q</mi><mrow><mn>1</mn><mo>,</mo><mn>1</mn></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd><mtd><mrow><msup><msub><mi>q</mi><mrow><mn>1</mn><mo>,</mo><mn>2</mn></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd><mtd><mo>...</mo></mtd><mtd><mrow><msup><msub><mi>q</mi><mrow><mn>1</mn><mo>,</mo><mi>m</mi></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd></mtr><mtr><mtd><mrow><msup><msub><mi>q</mi><mrow><mn>2</mn><mo>,</mo><mn>1</mn></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd><mtd><mrow><msup><msub><mi>q</mi><mrow><mn>2</mn><mo>,</mo><mn>2</mn></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd><mtd><mo>...</mo></mtd><mtd><mrow><msup><msub><mi>q</mi><mrow><mn>2</mn><mo>,</mo><mi>m</mi></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd></mtr><mtr><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd><mtd><mtable><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr><mtr><mtd><mo>.</mo></mtd></mtr></mtable></mtd></mtr><mtr><mtd><mrow><msup><msub><mi>q</mi><mrow><mi>n</mi><mo>,</mo><mn>1</mn></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd><mtd><mrow><msup><msub><mi>q</mi><mrow><mi>n</mi><mo>,</mo><mn>2</mn></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd><mtd><mo>...</mo></mtd><mtd><mrow><msup><msub><mi>q</mi><mrow><mi>n</mi><mo>,</mo><mi>m</mi></mrow></msub><mo>&amp;prime;</mo></msup></mrow></mtd></mtr></mtable></mfenced></mrow> Wherein, q _{i, j} is an element in the matrix Q, and q _{i, j} ' is an element in the matrix Q'. 5 . The method for virtual service migration according to claim 1 , further comprising: not migrating the virtual service when no new request arrives or the transmission of the user request does not change. 6 .