CN110166304A - A kind of method of combination, device, electronic equipment and the storage medium of cross-domain SFC - Google Patents

Abstract

Embodiments of the present invention provide a cross-domain SFC arrangement method, device, electronic equipment, and storage medium. The method includes: for a plurality of SFCs to be arranged and a plurality of data center nodes, deploying them in different data centers based on different network function nodes According to the VNF overhead and bandwidth overhead generated by the central node, the initial probability set is constructed, and the transition probability matrix and output probability matrix are constructed. Then, based on the constructed initial probability set, transition probability matrix, and output probability matrix, the hidden Mark can be constructed Using the HMM model, using the initial probability, transition probability, and output probability in the hidden Markov model, calculate the hidden state probability corresponding to the deployment of the network function node in the SFC in the data center node, and based on the hidden state probability Each SFC is arranged to obtain the hidden state subsequence corresponding to each SFC. The embodiment of the present invention can reduce the cross-domain bandwidth overhead of the orchestrated SFC.

Description

A cross-domain SFC orchestration method, device, electronic equipment and storage medium

technical field

The present invention relates to the technical field of communication, and in particular to a cross-domain SFC arrangement method, device, electronic equipment and storage medium.

Background technique

With the rapid development of communication technology, the continuous growth of network scale and the continuous expansion of business, in order to complete the corresponding services in the network, the traffic in the network needs to be processed in order according to the network functions in a specific order. SFC (Service Function Chains, business function chain ) defines the order in which network functions are processed. In practical applications, network operators need to guide traffic to proprietary hardware network functions to complete the services corresponding to SFC, but the network functions of traditional proprietary devices perform well in terms of network flexibility, scalability, manageability and operational efficiency. Due to obvious limitations, NFV (Network Functions Virtualization, Network Functions Virtualization) came into being. NFV is a new network architecture that decouples network functions from proprietary hardware devices and runs them on general-purpose devices in the form of VNF (Virtualized Network Function).

With the widespread application of cloud computing and big data, multiple IDCNs (Inter-DCNetwork, data center networks) distributed in different geographical locations have been widely deployed. By deploying NFV-based SFC in multiple IDCNs, users can flexibly use the computing resources, network resources, and storage resources of the data center, and realize economical and fast user service deployment. Although deploying SFC in multiple IDCNs has great advantages, orchestrating SFC across data centers is a current SFC orchestration challenge.

The current method for SFC orchestration is as follows: for all SFCs, based on the correlation of VNF types in SFC, traverse and integrate the network function types of multiple SFCs to be arranged one by one until all SFCs to be arranged are integrated into one or more sheets In the service function diagram, the service function diagram contains all network function sequences to be arranged for SFC requests, then select the largest service function diagram, perform topology sorting on the service function diagram, and further select the network function sequence with the lowest bandwidth demand, Use the shortest path algorithm to deploy the network functions of the SFC to be arranged to the underlying network in sequence.

However, in practical applications, SFC usually needs to be deployed in data centers distributed in different geographical locations to meet the performance requirements or location constraints of SFC. For example, "proxy functions" and "caching functions" should be deployed in data centers near the enterprise network , "packet filter" should be deployed on the data center close to the traffic source, etc. Existing methods for SFC orchestration usually integrate the SFC to be orchestrated into one or more service function diagrams based on the correlation of VNF types in the SFC, so as to realize the orchestration of the SFC. However, when the number of SFCs to be arranged is large, in the integrated service function diagram, it may appear that the previous function nodes in the SFC to be arranged are not adjacent to the subsequent function nodes, that is, the previous function nodes and subsequent function nodes are not adjacent to each other. Functional nodes may cross domains, resulting in an increase in the cross-domain bandwidth overhead between the previous functional node and the subsequent functional node, resulting in an increase in the cross-domain bandwidth overhead of the SFC after orchestration.

Contents of the invention

The purpose of the embodiments of the present invention is to provide a cross-domain SFC orchestration method, device, electronic equipment, and storage medium, so as to reduce the cross-domain bandwidth overhead of the orchestrated SFC. The specific technical scheme is as follows:

In a first aspect, an embodiment of the present invention provides a cross-domain SFC orchestration method, the method including:

Obtain multiple service function chains SFCs to be programmed and multiple data center nodes, each of which includes multiple network function nodes, the data center nodes are used to deploy network function nodes in the SFC, and one The network function node is deployed in one of the data center nodes;

Based on the virtual network function VNF overhead and bandwidth overhead generated when different network function nodes are deployed in different data center nodes, determine the initial probability of deploying different network function nodes in different data center nodes, and based on the determined multiple Initial probability constructs an initial probability set;

Based on the transition probability determined by the transition from the first state to the second state, a transition probability matrix is constructed; the first state is the state corresponding to the process of deploying the first network function node in the SFC in the first data center node, so The second state is a state corresponding to the process of deploying the second network function node in the SFC in the second data center node, and the second network function node is an adjacent node of the first network function node;

Based on the output probability of the function type corresponding to the third network function node in the output SFC in the third state, an output probability matrix is constructed; the third state is to deploy the third network function node in the SFC in the third data center node The state corresponding to the process;

Constructing a Hidden Markov Model based on the constructed initial probability set, the transition probability matrix, and the output probability matrix;

Using the initial probability, the transition probability, and the output probability in the hidden Markov model, calculate the hidden state probability corresponding to deploying the network function node in the SFC in the data center node, and based on the The hidden state probability arranges each SFC in the plurality of SFCs, obtains a hidden state subsequence corresponding to each of the SFCs, and obtains a plurality of hidden state subsequences corresponding to the plurality of SFCs; the hidden state subsequence The elements contained in are: the data center nodes deployed by the network function nodes in the SFC.

Optionally, using a first preset expression to determine an initial probability of deploying the different network function nodes in the different data center nodes;

using a second preset expression to determine a transition probability from the first state to the second state;

Using a third preset expression to determine the output probability of the function type corresponding to the third network function node in the output SFC in the third state;

The first preset expression is:

In the formula, π _m represents the initial probability of deploying the network function node on the mth data center node, M represents the number of data center nodes, Indicates the sth VNF instance in the mth data center node, Indicates the overhead generated by deploying the first network function node to the sth VNF instance in the mth data center node, Indicates the transfer bandwidth overhead generated by deploying the initial network function node on the initial data center node and the first network function node on the mth data center node. 01 indicates that the initial network function node of the SFC goes to the first network The state transition of the function node, σm represents the transition from the initial data center node to the mth data center node;

The second preset expression is:

In the formula, Indicates status transfer to status The transition probability of the state Indicates the state corresponding to the process of deploying the i-1th network function node in the nth data center node, state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the transfer bandwidth cost of deploying the (i-1)th network function node in data center node n, and deploying the i-th network function node in data center m, (i-1)i represents the (i-1)th network The transfer from the function node to the i-th network function node, nm means the transfer from the n-th data center node to the m-th data center node, Represents the unit bandwidth cost from data center node n to data center node m, N represents the number of network topologies of data center nodes, Indicates the amount of requested bandwidth between the (i-1)th network function node of the p-th SFC and the i-th network function node;

The third preset expression is:

In the formula, Indicates status The following output network function type probability of state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the function type of the i-th network function node of the p-th SFC, Indicates status The network function type is not output under The probability, Indicates the overhead generated by deploying the i-th network function node to the s-th VNF instance in the m-th data center node, Indicates the reliability of the sth VNF instance in the mth data center node, Indicates the function type of the sth VNF instance in the mth data center node.

Optionally, the plurality of SFCs to be edited is a set of SFCs; using the initial probability, the transition probability and the output probability in the hidden Markov model to calculate the network in the SFC The hidden state probability corresponding to the functional node deployed in the data center node, and based on the hidden state probability, arranges each SFC in the plurality of SFCs, obtains the hidden state subsequence corresponding to each of the SFCs, and obtains the Multiple hidden state subsequences corresponding to multiple SFCs are:

Using the initial probability, the transition probability, and the output probability in the hidden Markov model, calculate the hidden state probability corresponding to deploying the network function node in the SFC in the data center node, and based on the The hidden state probability arranges each SFC in the plurality of SFCs, obtains hidden state subsequences corresponding to each of the SFCs, and obtains a hidden state sequence set corresponding to the set of SFCs.

Optionally, using the initial probability, the transition probability, and the output probability in the hidden Markov model to calculate the hidden state probability, and arrange each SFC in the multiple SFCs based on the hidden state probability, obtain the hidden state subsequence corresponding to each of the SFCs, and obtain the hidden state sequence set corresponding to the SFC set, including :

Judging whether there is an SFC to be edited in the SFC set;

If there is an SFC to be edited in the SFC set, select the longest SFC in the SFC set as the current edited SFC;

Using the initial probability, the transition probability, and the output probability in the hidden Markov model, calculate the hidden state probability corresponding to deploying each network function node in the current orchestration SFC in a data center node , and arrange the current arrangement SFC based on the hidden state probability, and obtain the hidden state subsequence corresponding to the current arrangement SFC;

If there is no SFC to be programmed in the SFC set, then output the hidden state sequence set corresponding to the SFC set.

Optionally, using the initial probability, the transition probability, and the output probability in the hidden Markov model to calculate and deploy each network function node in the current orchestration SFC in the data center node The corresponding hidden state probability, and based on the hidden state probability, arrange the current arrangement SFC, and obtain the step of hidden state subsequence corresponding to the current arrangement SFC, including:

Judging whether the current arrangement of the SFC is completed;

If the current programming SFC has not been programmed, then judging whether the current programming network function node of the current programming SFC is the first network function node of the current programming SFC;

If the current orchestration network function node of the current orchestration SFC is the first network function node of the current orchestration SFC, then calculate the current orchestration network function node based on the initial probability in the hidden Markov model The initial hidden state probability deployed in each data center node;

If the current orchestration network function node of the current orchestration SFC is not the first network function node of the current orchestration SFC, then use the transition probability and the output probability in the hidden Markov model to calculate the Currently arrange the maximum hidden state probability of the network function node deployed in each data center node, and record the position of the preceding data center node corresponding to the maximum hidden state probability;

Using the network function node next to the currently programmed network function node of the currently programmed SFC as the current programmed network function node of the currently programmed SFC, performing the step of judging whether the currently programmed SFC is completed;

If the arrangement of the current arrangement SFC is completed, the last network function node of the current arrangement SFC is used as the current network function node, and the fourth data center node corresponding to the maximum hidden state probability of the current network function node is selected to deploy the The current network function node, and storing the fourth data center node in the hidden state subsequence corresponding to the current orchestration SFC;

Determining the preceding data center node corresponding to the maximum hidden state probability of the current network function node as the fifth data center node corresponding to the previous network function node of the current network function node, in the fifth data Deploying the previous network function node of the current network function node in the central node, storing the fifth data center node in the hidden state subsequence corresponding to the current orchestration SFC, and storing the previous network function node of the current network function node A network function node as the current network function node;

judging whether the previous network function node of the current network function node is the first network function node of the current programmed SFC;

If the previous network function node of the current network function node is not the first network function node of the current orchestration SFC, execute the preceding data center node corresponding to the maximum hidden state probability of the current network function node, A step of determining the fifth data center node corresponding to the previous network function node of the current network function node;

If the previous network function node of the current network function node is the first network function node of the current composed SFC, then delete the current composed SFC from the SFC set.

Optionally, the method also includes:

Based on the hidden state sequence set, the reliability value corresponding to each of the SFCs after arrangement, and the cost-effectiveness value of the VNF corresponding to each network function node in the SFC corresponding to each hidden state subsequence in the hidden state sequence set, The VNF is backed up.

Optionally, based on the hidden state sequence set, the reliability value corresponding to each of the SFCs after arrangement, and the reliability value corresponding to each network function node in the SFC corresponding to each hidden state subsequence in the hidden state sequence set The cost-benefit value of the VNF, the steps of backing up the VNF, including:

Traverse each hidden state subsequence in the hidden state sequence set, and use the SFC corresponding to the hidden state subsequence as the current SFC;

calculating the reliability value of the current SFC;

judging whether the reliability value of the current SFC is less than a preset reliability value;

When the reliability value of the current SFC is less than a preset reliability value, placing the current SFC in a first set, and placing the VNF passing through the current SFC in a second set;

judging whether the first set is empty;

When the first set is not empty, calculating a cost-benefit value for each VNF in the second set;

Back up the VNF corresponding to the maximum cost-benefit value;

calculating the reliability value of each SFC in the first set after backup;

If the reliability value of the SFC in the first set after backup is not less than the preset reliability value, then delete the SFC from the first set, and perform the step of judging whether the first set is empty .

In a second aspect, an embodiment of the present invention provides a device for orchestrating a cross-domain SFC, the device comprising:

An acquisition module, configured to acquire a plurality of service function chains SFCs to be arranged and a plurality of data center nodes, each of the SFCs includes a plurality of network function nodes, and the data center nodes are used to deploy the network in the SFC A function node, one of the network function nodes is deployed in one of the data center nodes;

The first building block is used to determine the initial probability of deploying different network function nodes in different data center nodes based on the virtual network function VNF overhead and bandwidth overhead generated when different network function nodes are deployed in different data center nodes, and based on A plurality of the determined initial probabilities construct an initial probability set;

The second building block is configured to construct a transition probability matrix based on the transition probability determined by transitioning from the first state to the second state; the first state is the deployment of the first network function node in the SFC in the first data center node The state corresponding to the process, the second state is the state corresponding to the process of deploying the second network function node in the SFC in the second data center node, the second network function node being the first network function node adjacent nodes;

The third building block is configured to construct an output probability matrix based on the output probability of the function type corresponding to the third network function node in the output SFC in the third state; the third state is to deploy the third network function node in the SFC a state corresponding to a process in the third data center node;

A fourth construction module, configured to construct a hidden Markov model based on the constructed initial probability set, the transition probability matrix, and the output probability matrix;

An orchestration module, configured to use the initial probability, the transition probability, and the output probability in the hidden Markov model to calculate the hidden state corresponding to deploying the network function node in the SFC in the data center node probability, and arrange each SFC in the multiple SFCs based on the hidden state probability, obtain hidden state subsequences corresponding to each of the SFCs, and obtain multiple hidden state subsequences corresponding to the multiple SFCs; The elements included in the hidden state subsequence are: the data center nodes deployed by the network function nodes in the SFC.

In the third aspect, the embodiment of the present invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein, the processor, the communication interface, and the memory complete communication with each other through the communication bus;

memory for storing computer programs;

The processor is used to implement the cross-domain SFC programming method described in the first aspect when executing the program stored in the memory.

In the fourth aspect, the embodiment of the present invention also provides a computer-readable storage medium, where instructions are stored in the computer-readable storage medium, and when the computer-readable storage medium is run on a computer, the computer executes one of the above-mentioned first aspects. A cross-domain SFC orchestration method.

A cross-domain SFC arrangement method, device, electronic equipment, and storage medium provided by the embodiments of the present invention cannot be directly observed due to the hidden state of the HMM, but can be observed through observable sequences, and each observable sequence is obtained through The probability density distribution represents various states, and each observable sequence is generated by a state sequence with a corresponding probability density distribution. In the embodiment of the present invention, the cross-domain SFC arrangement is modeled as a hidden Markov model, and then Use the hidden Markov model to arrange each SFC in multiple SFCs to obtain the hidden state subsequence corresponding to each SFC, because it is based on the hidden state subsequence corresponding to the deployment of the network function nodes in the SFC in the data center nodes obtained through calculation. State probability, deploying SFC network function nodes to realize SFC orchestration, can reduce the cross-domain bandwidth overhead of SFC after orchestration, and then reduce the bandwidth resources consumed by SFC after orchestration.

Of course, implementing any product or method of the present invention does not necessarily need to achieve all the above-mentioned advantages at the same time.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

FIG. 1 is a schematic flowchart of a cross-domain SFC orchestration method provided by an embodiment of the present invention;

Fig. 2 is a schematic flow chart of an implementation manner of arranging SFC provided by an embodiment of the present invention;

Fig. 3 is a schematic flow chart of an embodiment of S203 in Fig. 2;

FIG. 4 is a schematic structural diagram of an SFC provided by an embodiment of the present invention;

FIG. 5 is a schematic flowchart of an implementation manner of backing up a VNF provided by an embodiment of the present invention;

Figure 6a is a simulation diagram of the relationship between VNF overhead and SFC demand provided by the embodiment of the present invention;

Figure 6b is a simulation diagram of the relationship between cross-domain bandwidth overhead and SFC demand provided by the embodiment of the present invention;

Figure 6c is a simulation diagram of the relationship between backup overhead and SFC demand provided by the embodiment of the present invention;

Figure 6d is a simulation diagram of the relationship between the total cost and the SFC demand provided by the embodiment of the present invention;

FIG. 7 is a schematic structural diagram of a cross-domain SFC orchestration device provided by an embodiment of the present invention;

FIG. 8 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

As shown in FIG. 1, FIG. 1 is a schematic flowchart of a cross-domain SFC orchestration method provided by an embodiment of the present invention. The method may include:

S101. Obtain multiple service function chains SFCs to be orchestrated and multiple data center nodes.

In the embodiment of the present invention, when the SFC is arranged, multiple SFCs to be arranged and multiple data center nodes can be obtained. Each SFC includes multiple network function nodes, the data center node is used to deploy the network function nodes in the SFC, one network function node is deployed in one data center node, and one data center node represents a data center network.

S102, based on the VNF overhead and bandwidth overhead generated when different network function nodes are deployed in different data center nodes, determine the initial probability of deploying different network function nodes in different data center nodes, and construct a network based on the determined multiple initial probabilities set of initial probabilities.

In practical applications, when deploying network function nodes on data center nodes, it is necessary to create a VNF instance, and then deploy the network function nodes on the created VNF instance, which will generate VNF overhead. Exemplarily, the VNF instance may be a vFW (Virtualized Fire Wall, virtual firewall), vLB (Virtualized LoadBalance, virtual traffic balancer) and the like. When the network between different data center nodes performs data interaction, it needs to occupy corresponding bandwidth, so that when nodes with different network functions are deployed on different data center nodes, corresponding bandwidth overhead will be generated.

Because the hidden Markov model can calculate the hidden state sequence with the highest probability based on the known state sequence, and in the process of SFC arrangement, the sequence of network function nodes in SFC is known, and the sequence of positions deployed by network function nodes needs to be obtained is unknown, so in the embodiment of the present invention, the cross-domain SFC orchestration problem is modeled as a hidden Markov model. Based on the VNF overhead and bandwidth overhead generated when different network function nodes are deployed in different data center nodes, the initial probability of deploying different network function nodes in different data center nodes is determined.

As an optional implementation of the embodiment of the present invention, the initial probability of deploying different network function nodes in different data center nodes may be determined by using a first preset expression, and the first preset expression may be:

In the formula, π _m represents the initial probability of deploying the network function node on the mth data center node, M represents the number of data center nodes, Indicates the sth VNF instance in the mth data center node, Indicates that the first network function node is deployed on the overhead incurred, Indicates the transfer bandwidth overhead generated by deploying the initial network function node on the initial data center node and the first network function node on the mth data center node. 01 indicates that the initial network function node of the SFC goes to the first network The state transition of the function node, σm represents the transition from the starting data center node to the mth data center node.

After determining initial probabilities of deploying different network function nodes in different data center nodes, an initial probability set may be constructed based on the determined multiple initial probabilities. Exemplarily, the constructed initial probability set may be expressed as: Π={π ₁ , π ₂ ,...π _M }, where π _M represents the initial probability of deploying the network function node on the Mth data center node.

S103. Construct a transition probability matrix based on the determined transition probability from the first state to the second state.

In the embodiment of the present invention, the transition probability from the first state to the second state may be determined, and then a transition probability matrix may be constructed based on the determined transition probability. Wherein, the first state is the state corresponding to the process of deploying the first network function node in the SFC to the first data center node, and the second state is the state corresponding to the process of deploying the second network function node in the SFC to the second data center node In the state corresponding to the process, the second network function node is an adjacent node of the first network function node. In the embodiment of the present invention, the order of the network function nodes in the SFC is an observable sequence, and the data center nodes deployed by the network function nodes in the SFC cannot be directly observed, but can be expressed as various state. Exemplarily, the process of deploying the first network function node in the first data center node in the SFC may be transformed into a state, which is denoted as the first state.

As an optional implementation of the embodiment of the present invention, a second preset expression may be used to determine the transition probability of the transition from the first state to the second state, and the second preset expression may be:

In the formula, Indicates status transfer to state The transition probability of the state Indicates the state corresponding to the process of deploying the i-1th network function node in the nth data center node, state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the transfer bandwidth cost of deploying the (i-1)th network function node in data center node n, and deploying the i-th network function node in data center m, (i-1)i represents the (i-1)th network The transfer from the function node to the i-th network function node, nm means the transfer from the n-th data center node to the m-th data center node, Represents the unit bandwidth cost from data center node n to data center node m, N represents the number of network topologies of data center nodes, Indicates the amount of requested bandwidth between the (i-1)th network function node of the p-th SFC and the i-th network function node.

Exemplarily, the constructed transition probability matrix can be expressed as:

Among them, A represents the constructed transition probability matrix, Indicates status transfer to state The transition probability of the state Indicates the state corresponding to the process of deploying the i-1th network function node in the first data center node, state Indicates the state corresponding to the process of deploying the i-th network function node in the M-th data center node.

S104. Construct an output probability matrix based on the output probability of the function type corresponding to the third network function node in the output SFC in the third state.

In the embodiment of the present invention, the output probability of the function type corresponding to the third network function node in the output SFC in the third state may be determined, and then an output probability matrix is constructed according to the determined output probability. Wherein, the third state is a state corresponding to the process of deploying the third network function node in the SFC in the third data center node.

As an optional implementation of the embodiment of the present invention, the third preset expression can be used to determine the output probability of the function type corresponding to the third network function node in the output SFC in the third state, the third preset expression Can be:

In the formula, Indicates status The following output network function type probability of state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the function type of the i-th network function node of the p-th SFC, Indicates status The network function type is not output under The probability, Indicates the sth VNF instance in the mth data center node, Indicates that the i-th network function node is deployed on the overhead incurred, express the reliability value of express type of function.

Exemplarily, the constructed output probability matrix can be expressed as:

Among them, B represents the constructed output probability matrix, Indicates status The following output network function type probability of state Indicates the state corresponding to the process of deploying the i-th network function node in the M-th data center node, Indicates the function type corresponding to the i-th network function node of the p-th SFC.

S105. Construct a hidden Markov model based on the constructed initial probability set, transition probability matrix, and output probability matrix.

As an optional implementation of the embodiment of the present invention, based on the constructed initial probability set Π, transition probability matrix A, and output probability matrix B, a triplet model of hidden Markov model can be constructed, the triplet The model can be expressed as (Π, A, B).

S106. Using the initial probability, transition probability, and output probability in the hidden Markov model, calculate the hidden state probability corresponding to deploying the network function node in the SFC in the data center node, and based on the hidden state probability One SFC is arranged to obtain hidden state subsequences corresponding to each SFC, and multiple hidden state subsequences corresponding to multiple SFCs are obtained.

In the embodiment of the present invention, the order of the network function nodes in the SFC is an observable sequence, and the data center nodes deployed by the network function nodes in the SFC cannot be directly observed, but can be observed through the observable sequence, and each observable Sequences are expressed as various states through probability density distributions, and each observable sequence is generated by a state sequence with a corresponding probability density distribution. In the embodiment of the present invention, the hidden state probability corresponding to deploying the network function node in the SFC in the data center node is calculated, and then each SFC in the multiple SFCs can be arranged based on the hidden state probability. Wherein, the elements included in the obtained hidden state subsequence may be: the data center nodes deployed by the network function nodes in the SFC.

As an optional implementation manner of the embodiment of the present invention, multiple SFCs to be edited may be expressed as an SFC set. The above step S106 may specifically be:

Using the initial probability, transition probability, and output probability in the hidden Markov model, calculate the hidden state probability corresponding to the deployment of the network function node in the SFC in the data center node, and based on the hidden state probability for each SFC in multiple SFCs Perform arrangement to obtain the hidden state subsequence corresponding to each SFC, and obtain the hidden state sequence set corresponding to the SFC set.

Exemplarily, the SFC set can be expressed as: SFC={S ¹ , S ² ,...S ^q }, the hidden state sequence set corresponding to the SFC set can be expressed as: Q={Q ¹ ,Q ² ,...Q ^q }, Among them, S ^q represents the qth SFC, and Q ^q represents the hidden state subsequence corresponding to the qth SFC.

As an optional implementation of the embodiment of the present invention, use the initial probability, transition probability and output probability in the hidden Markov model to calculate the hidden state probability corresponding to deploying the network function node in the SFC in the data center node, And based on the hidden state probability, arrange each SFC in the multiple SFCs, obtain the hidden state subsequence corresponding to each SFC, and obtain the hidden state sequence set corresponding to the SFC set. Refer to FIG. 2, and this embodiment may include:

S201. Determine whether there is an SFC to be programmed in the SFC set.

For the SFC set to be edited, it is judged whether there is an SFC to be edited in the SFC set, if there is an SFC to be edited in the SFC set, it means that the SFC to be edited is being edited, or the SFC to be edited is about to be edited, and the step of S202 is executed ; If there is no SFC to be programmed in the SFC set, it means that the SFC to be programmed has been programmed, and the step S204 is executed.

S202. If there is an SFC to be edited in the SFC set, select the longest SFC in the SFC set as the currently edited SFC.

As an optional implementation of the embodiment of the present invention, if there is an SFC to be arranged in the SFC set, in order to simplify the complexity of SFC arrangement, the longest SFC in the SFC set is selected as the current arrangement SFC, and the longest SFC can be : Contains the SFC corresponding to the largest number of network function nodes. Exemplarily, the longest SFC can be expressed as: in, Indicates the Kth network function node of the pth SFC.

S203. Using the initial probability, transition probability, and output probability in the hidden Markov model, calculate the hidden state probability corresponding to deploying each network function node in the current orchestration SFC in the data center node, and based on the hidden state probability Arrange the SFC to perform the arrangement, and obtain the hidden state subsequence corresponding to the currently arranged SFC.

After the current orchestration SFC is determined, the current orchestration SFC is arranged, and the specific implementation process is described in detail below.

S204, if there is no SFC to be programmed in the SFC set, output a hidden state sequence set corresponding to the SFC set.

If there is no SFC to be programmed in the SFC set, it indicates that the SFC to be programmed has been programmed, and at this time, the hidden state sequence set corresponding to the SFC set is output.

As an optional implementation manner of the embodiment of the present invention, the specific implementation manner of the above step S203 can be referred to FIG. 3 , and this implementation manner may include:

S2031, judging whether the programming of the currently programmed SFC is completed.

For the currently edited SFC, it can be judged whether the currently edited SFC is edited, if yes, execute step S2032; if not, execute step S2036.

S2032. If the current programming of the SFC has not been completed, determine whether the current programming network function node of the current programming SFC is the first network function node of the current programming SFC.

When the current layout SFC is not completed, it can be judged whether the current layout network function node of the current layout SFC is the first network function node of the current layout SFC, if yes, then perform the step of S2033; if not, then perform the step of S2034 .

S2033, if the current orchestration network function node of the current orchestration SFC is the first network function node of the current orchestration SFC, calculate and deploy the current orchestration network function node in each data center based on the initial probability in the hidden Markov model Initial hidden state probabilities in nodes.

As an optional implementation of the embodiment of the present invention, the fourth preset expression can be used to calculate the deployment of the current orchestration network function node in each data center node based on the initial probability in the hidden Markov model Initial hidden state probability, because the current orchestration network function node is the first network function node of the current orchestration SFC, the calculated initial hidden state probability corresponding to the first network function node can also be its maximum hidden state probability . The fourth preset expression may be:

In the formula, Indicates the initial hidden state probability that the first network function node of the current orchestration SFC is deployed on the data center node m, π _m represents the initial probability of deploying the network function node on the mth data center node, Indicates status The following output network function type probability of state Indicates the state corresponding to the process of deploying the first network function node in the mth data center node, Indicates the function type of the first network function node of the p-th SFC.

S2034. If the current orchestration network function node of the current orchestration SFC is not the first network function node of the current orchestration SFC, calculate and deploy the current orchestration network function node in each The maximum hidden state probability in a data center node, and the position of the preceding data center node corresponding to the maximum hidden state probability is recorded.

As an optional implementation of the embodiment of the present invention, when the current network function node of the current SFC layout is not the first network function node of the current SFC layout, the fifth preset expression can be used to calculate the current Arranging the maximum hidden state probability of network function nodes deployed in each data center node, and recording the position of the front data center node corresponding to the maximum hidden state probability. The fifth preset expression may be:

In the formula, Indicates the maximum hidden state probability of deploying the i-th network function node of the currently orchestrated SFC in the data center node m, Indicates the maximum hidden state probability of deploying the i-1th network function node of the currently orchestrated SFC in the data center node x, Indicates status transfer to status The transition probability of the state Indicates the state corresponding to the process of deploying the i-1th network function node in the xth data center node, state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates status The following output network function type probability of state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the function type of the i-th network function node of the p-th SFC.

S2035. Use the next network function node of the currently programmed network function node of the currently programmed SFC as the current programmed network function node of the currently programmed SFC, and perform the step of judging whether the currently programmed SFC is completed.

After calculating the maximum hidden state probability of deploying the current orchestration network function node in each data center node, and recording the location of the front data center node corresponding to the maximum hidden state probability, the current orchestration network function of the current orchestration SFC The next network function node of the node is used as the current network function node of the current program SFC, and then the step of S2031 is executed until the program of the current program SFC is completed.

S2036. If the current arrangement of the SFC is completed, use the end network function node of the current arrangement SFC as the current network function node, select the fourth data center node corresponding to the largest hidden state probability of the current network function node to deploy the current network function node, and Store the fourth data center node in the hidden state subsequence corresponding to the current orchestration SFC.

When the arrangement of the current arranged SFC is completed, the last network function node of the currently arranged SFC, that is, the last network function node of the currently arranged SFC is taken as the current network function node, and the fourth data corresponding to the maximum hidden state probability of the current network function node is selected The central node deploys the current network function node, and stores the fourth data center node in the hidden state subsequence corresponding to the current orchestration SFC. Exemplarily, the current orchestration SFC is the pth SFC, and the end network function node of the current orchestration SFC is Make the current network function node The fourth data center node corresponding to the largest hidden state probability is d _m , and d _m is stored in the hidden state subsequence corresponding to the current orchestration SFC.

S2037, determine the preceding data center node corresponding to the maximum hidden state probability of the current network function node as the fifth data center node corresponding to the previous network function node of the current network function node, and deploy in the fifth data center node The network function node preceding the current network function node, storing the fifth data center node in the hidden state subsequence corresponding to the current orchestration SFC, and using the network function node preceding the current network function node as the current network function node.

Exemplarily, make the current network function node The fourth data center node corresponding to the largest hidden state probability is d _m , so that the current network function node The previous data center node corresponding to the largest hidden state probability of , that is, the previous network function node of the current network function node The corresponding fifth data center node d _x is deployed in d _x Store d _x in the hidden state subsequence corresponding to the current orchestration SFC, and set As the current network function node.

S2038, judging whether the previous network function node of the current network function node is the first network function node for currently programming the SFC.

Determine whether the previous network function node of the current network function node is the first network function node of the current orchestration SFC, if yes, it means that all the network function nodes of the current orchestration SFC have been deployed, and the hidden state corresponding to the current orchestration SFC is obtained subsequence, execute the step of S2039, if not, it means that the deployment of all the network function nodes currently orchestrating the SFC has not been completed, then return to execute the step of S2037. Exemplarily, the current programmed SFC is ^Sp , and the obtained hidden state subsequence corresponding to the previous programmed SFC can be expressed as: in, Indicates the data center node d _K deployed by the Kth network function node in the previous orchestration S ^p .

S2039. If the previous network function node of the current network function node is the first network function node of the current composed SFC, delete the current composed SFC from the SFC set.

If the previous network function node of the current network function node is the first network function node of the currently programmed SFC, it means that all the network function nodes of the current programmed SFC have been deployed, and the hidden state subsequence corresponding to the current programmed SFC has been obtained, then Delete the currently programmed SFC from the SFC collection, and continue to program the next SFC to be programmed.

The embodiment of the present invention provides a cross-domain SFC orchestration method, because the hidden state of the HMM cannot be directly observed, but can be observed through the observable sequence, and each observable sequence is expressed as various states through the probability density distribution Each observable sequence is generated by a state sequence with a corresponding probability density distribution. In the embodiment of the present invention, the arrangement of cross-domain SFC is modeled as a hidden Markov model, and then the hidden Markov model is used to Each SFC in multiple SFCs is arranged to obtain the hidden state subsequence corresponding to each SFC, because it is based on the calculated hidden state probability corresponding to the network function node in the SFC deployed in the data center node, and the network function of the SFC Deploying nodes to implement SFC orchestration can reduce the cross-domain bandwidth overhead of SFC after orchestration, thereby reducing the bandwidth resources consumed by SFC after orchestration.

As an optional implementation of the embodiment of the present invention, after the above step S106, the arrangement of the cross-domain SFC in the embodiment of the present invention may also include: based on the hidden state sequence set, the reliability value corresponding to each SFC after arrangement, and the hidden The cost-benefit value of the VNF corresponding to each network function node in the SFC corresponding to each hidden state subsequence in the state sequence set is used to back up the VNF.

In practical applications, users will have reliability requirements for each SFC. Exemplarily, as shown in FIG. 4 , there are two SFCs, namely service function chain 1 and service function chain 2 . Service function chain 1 starts from client A, and the service flow needs to pass through FW (Fire Wall, firewall), LB (Load Balance, traffic balancer) and NAT (Network Address Translatio, network address translator) in sequence, and finally flows to client C. Service function chain 2 starts from client B, and the service flow passes through FW, LB and GW (Gateway, gateway) in sequence, and finally flows to client D. Among them, the reliability requirement of service function chain 1 is R ₁ =80%, and the reliability requirement of service function chain 2 is R ₂ =85%.

Exemplarily, there are 5 data center nodes, namely: DC1, DC2, DC3, DC4 and DC5. After orchestrating the service function chain 1 and service function chain 2, the orchestration result is: the service flow of service function chain 1 passes through the vFW deployed in DC1, the vLB deployed in DC3, and the vNAT (Virtualized Network Address Translatio , virtual network address translator), to client C. Among them, the reliability value of vFW deployed on DC1 is 0.92, the reliability value of vLB deployed on DC3 is 0.82, and the reliability value of vNAT deployed on DC4 is 0.93, then the reliability value of service function chain 1 deployment is: 0.92×0.82×0.93=70.2%<R ₁ =80%.

The service flow of the service function chain 2 passes through the vFW deployed on DC2, the vLB deployed on DC3, and the vGW (Virtualized Gateway, virtual gateway) deployed on DC5 in sequence, and finally reaches the client D. Among them, the reliability value of vFW deployed on DC2 is 0.97, the reliability value of vLB deployed on DC3 is 0.82, the reliability value of vGW deployed on DC5 is 0.74, and the reliability value of service function chain 2 is 0.97× 0.82×0.74=58.86%<R ₂ =85%.

It can be seen that in practical applications, after the arrangement of SFCs, the reliability requirements of SFCs may not be able to meet the needs of users. Therefore, in the embodiment of the present invention, the reliability values corresponding to each SFC after arrangement can be based on the above hidden state sequence set. And the cost-benefit value of the VNF corresponding to each network function node in the SFC corresponding to each hidden state subsequence in the hidden state sequence set, back up the VNF to improve the reliability value of the SFC after the VNF backup.

As an optional implementation of the embodiment of the present invention, as shown in FIG. 5, the above steps of backing up the VNF may specifically include:

S301. Traverse each hidden state subsequence in the hidden state sequence set, and use the SFC corresponding to the hidden state subsequence as the current SFC.

For each SFC programmed above, traverse each hidden state subsequence in the hidden state sequence set, and use the SFC corresponding to the hidden state subsequence as the current SFC.

S302. Calculate the reliability value of the current SFC.

As an optional implementation manner of the embodiment of the present invention, for each current SFC, calculating the reliability value of the current SFC may be: calculating the reliability value of the current SFC by using a sixth preset expression. The sixth preset expression may be:

In the formula, R ^p represents the reliability value of the p-th SFC, K represents the number of network function nodes of the p-th SFC, Indicates the reliability value of the i-th network function node of the p-th SFC. in, It is determined by the position of the data center node deployed by the i-th network function node and the function type of the i-th network function node in the p-th SFC arrangement process, that is, when the SFC arrangement is completed, the i-th network function node is deployed in the data center Node, the reliability value of the VNF in the data center node is determined by the function type of the VNF, that is, when the SFC orchestration is completed, get ok.

S303, judging whether the reliability value of the current SFC is smaller than a preset reliability value.

After completing the calculation of reliability values of all SFCs after arrangement, for each SFC, it may be determined whether the reliability value of the current SFC is less than a preset reliability value. The preset reliability value is a user's reliability requirement value for the SFC.

S304. When the reliability value of the current SFC is less than the preset reliability value, place the current SFC in the first set, and place the VNFs passing through the current SFC in the second set.

When the reliability value of the current SFC is less than the preset reliability value, it means that the reliability value of the current SFC does not meet the user's needs after arrangement, and the current SFC is placed in the first set, and the VNF that has passed the current SFC is placed in the second set. In the second set.

S305. Determine whether the first set is empty.

The first set contains the SFCs whose reliability values do not meet the user’s requirements. If the first set is empty, it means that the reliability values of all SFCs after SFC arrangement meet the user’s needs. If the first set is not empty, it means After the SFC is programmed, the reliability value of the SFC does not meet the user's requirements.

S306. When the first set is not empty, calculate the cost-benefit value of each VNF in the second set.

The first set is not empty, which means that the reliability value of the SFC after SFC arrangement does not meet the user's needs. At this time, for each VNF in the second set, the sixth preset expression above can be used to calculate the backup value of the VNF. A reliability value for each SFC in the first set. Then, based on the reliability value of each SFC in the first set after the backup VNF, the first preset reliability value, the processing capacity requirement of the backup VNF and the unit processing capacity cost of the backup VNF, the seventh preset expression is used to calculate A cost-benefit value for each VNF in the second set. The seventh preset expression may be:

in,

In the formula, Indicates the sth VNF instance in the mth data center node, means backup The cost-benefit value of , p represents the p-th SFC, q represents the number of SFCs, means backup The improvement value of the p-th SFC reliability, R ^p means backup The reliability value of the last p-th SFC, φ ^p represents the first preset reliability value, which can be the reliability value set by the user, means backup For all SFC reliability enhancements, express The cost generated, α _m represents the unit processing power cost of the backup VNF, express The amount of processing power required, express type of function.

when backup When the reliability value R ^p of the p-th SFC is greater than the first preset reliability value, the Take the result as 1, making the backup resulting in higher reliability gains and lower cost Has a high cost-benefit value.

S307. Back up the VNF corresponding to the maximum cost benefit value.

After the cost-benefit value of each VNF in the second set is calculated, the VNF corresponding to the maximum cost-benefit value may be backed up. In practical applications, one VNF corresponding to the maximum cost-benefit value may be backed up, or multiple VNFs corresponding to the maximum cost-benefit value may be backed up, and the specific embodiments of the present invention are not limited here.

S308. Calculate the reliability value of each SFC in the first set after backup.

In the embodiment of the present invention, after backing up the VNF corresponding to the maximum cost-benefit value, calculate the reliability value of each SFC in the first set after backup. For details, refer to the calculation of the reliability value of the current SFC in step S302 The method is used to calculate the reliability value of each SFC in the first set after backup, which will not be described in detail here in this embodiment of the present invention.

S309, if the reliability value of the SFC in the first set after backup is not less than the preset reliability value, delete the SFC from the first set, and perform the step of judging whether the first set is empty.

After calculating the reliability value of each SFC in the first set after backup, the reliability value of the SFC can be judged to determine whether it is not less than the preset reliability value. If the reliability of the SFC in the first set after backup is The value is not less than the preset reliability value, indicating that the reliability value of the SFC meets the user's needs at this time, delete the SFC from the first set, and return to the step of S305 until the reliability value of all SFCs in the first set All meet the needs of users.

The embodiment of the present invention provides a cross-domain SFC orchestration method, because the hidden state of the HMM cannot be directly observed, but can be observed through the observable sequence, and each observable sequence is expressed as various states through the probability density distribution Each observable sequence is generated by a state sequence with a corresponding probability density distribution. In the embodiment of the present invention, the arrangement of cross-domain SFC is modeled as a hidden Markov model, and then the hidden Markov model is used to Each SFC in multiple SFCs is arranged to obtain the hidden state subsequence corresponding to each SFC, because it is based on the calculated hidden state probability corresponding to the network function nodes in the SFC deployed in the data center nodes, and the network function of the SFC Deploying nodes to implement SFC orchestration can reduce the cross-domain bandwidth overhead of SFC after orchestration, thereby reducing the bandwidth resources consumed by SFC after orchestration. In addition, the VNF is backed up, and the reliability value of the SFC after backup is improved, so that the reliability value of each SFC after backup can meet the user's requirements.

Exemplarily, in the embodiment of the present invention, the SFC is arranged using different arrangement methods, and the obtained simulation diagrams are shown in Fig. 6a to Fig. 6d respectively. Among them, the prior art method 1 is the prior art described in the background technology, and the prior art method 2 is: modeling the SFC arrangement problem as an integer linear programming problem, and then realizing the arrangement. For the specific implementation process, please refer to the existing The implementation of the technology, the embodiment of the present invention will not be repeated here. In the embodiment of the present invention, the multi-data center network is selected from the large-scale and accurate network topology topology-zoo, the cross-data center bandwidth is set to 200Gbps, and the cross-data center bandwidth unit cost is uniformly distributed in [0.01,0.02]$/Mbps A random value between , the unit IT resource cost of the data center is [0.05,0.10]$/unit, where the IT resources requested by different virtual network function nodes are randomly selected between [1,3], the reliability of virtual network function nodes The reliability value is set to [0.8,0.99], the length of SFC is uniformly distributed between [2,6], the requested bandwidth of SFC satisfies the uniform distribution of [10,100] Mbps, and the reliability value of SFC is between [0.95,0.98,0.99, 0.995,0.999] randomly selected.

6a is a simulation diagram of the relationship between VNF overhead and SFC demand provided by the embodiment of the present invention. When the number of SFC requests is less than 400, the prior art method 1 uses a smaller VNF instance to process incoming SFC requests. However, when the number of SFC requests is greater than 600, the VNF overhead of method 1 in the prior art will increase significantly. The embodiment of the present invention uses more VNFs in the case of small-scale SFC requests. This is because the embodiments of the present invention process the SFCs in descending order of length, ignoring other correlations of the SFCs to reduce the complexity of the problem, resulting in more VNF overhead. When dealing with large-scale SFC requests, the VNF overhead of the embodiment of the present invention is not much different from that of the prior art method 2. This is because the embodiment of the present invention converts SFC programming into the decoding problem of the hidden Markov model, fully considering the VNF Higher usage efficiency and lower cost VNF overhead.

Figure 6b is a simulation diagram of the relationship between cross-domain bandwidth overhead and SFC demand provided by the embodiment of the present invention. When arranging the same number of SFCs, prior art method 1 uses more cross-domain bandwidth overhead, which is due to the prior art method 1 Merge SFCs into the graph to reduce VNF instance usage, but this results in higher bandwidth consumption. The embodiment of the present invention considers both the bandwidth overhead and the VNF overhead, and compared with method 1 in the prior art, the bandwidth cost used is reduced by about 26.2%. When the number of SFC requests is less than 400, the bandwidth cost of the embodiment of the present invention is equivalent to that of method 2 in the prior art. When the number is greater than 600, compared with method 2 of the prior art, the bandwidth cost between data center nodes used in the embodiment of the present invention increases by about 11.3%, which is because the output probability in the hidden Markov model of the embodiment of the present invention is both Considering the VNF overhead, the reliability of the VNF is also considered. Therefore, the embodiment of the present invention sacrifices part of the cross-domain bandwidth overhead to obtain a highly reliable SFC orchestration result.

FIG. 6c is a simulation diagram of the relationship between backup overhead and SFC demand provided by the embodiment of the present invention. When the number of SFC requests is less than 600, the embodiment of the present invention uses a backup overhead comparable to that of method 2 in the prior art. When the number of SFC requests is greater than 800, the backup overhead used by the embodiment of the present invention is reduced by 13.2%. This is because the embodiment of the present invention considers the reliability requirements when using the hidden Markov model to arrange the SFC, and the embodiment of the present invention sacrifices the deployment Cross-domain bandwidth enables SFC orchestration with higher reliability.

Figure 6d is a simulation diagram of the relationship between total overhead and SFC demand provided by the embodiment of the present invention. The total overhead includes three parts: VNF overhead (as shown in Figure 6a), cross-domain bandwidth overhead (as shown in Figure 6b) and VNF backup overhead (as shown in Figure 6c). Obviously, compared with the results obtained by the embodiment of the present invention and the method 2 of the prior art, the method 1 of the prior art consumes more overhead. Compared with the results obtained by the method 2 of the prior art and the method of the embodiment of the present invention, the cost of the method 1 of the prior art is about 20.4% and 15.6% higher respectively. When the number of SFCs is greater than 600, the total overhead of the embodiment of the present invention is about 11.4% higher than that of method 2 of the prior art.

Although the overhead of the embodiment of the present invention is relatively high, the VNF is backed up after the SFC is orchestrated in the embodiment of the present invention, which can improve the reliability of the SFC and better meet the needs of users.

Corresponding to the foregoing method embodiments, an embodiment of the present invention provides a cross-domain SFC orchestration device, as shown in FIG. 7 , the device may include:

The obtaining module 401 is used to obtain multiple service function chains SFCs to be arranged and multiple data center nodes, each SFC includes multiple network function nodes, and the data center nodes are used to deploy network function nodes in the SFC, a network Function nodes are deployed in a data center node.

The first building block 402 is configured to determine the initial probability of deploying different network function nodes in different data center nodes based on the virtual network function VNF overhead and bandwidth overhead generated when different network function nodes are deployed in different data center nodes, and An initial probability set is constructed based on the determined plurality of initial probabilities.

The second construction module 403 is configured to construct a transition probability matrix based on the transition probability determined by transitioning from the first state to the second state; the first state is the process of deploying the first network function node in the SFC to the first data center node The corresponding state, the second state is the state corresponding to the process of deploying the second network function node in the SFC in the second data center node, and the second network function node is an adjacent node of the first network function node.

The third construction module 404 is configured to construct an output probability matrix based on the output probability of the function type corresponding to the third network function node in the output SFC in the third state; the third state is to deploy the third network function node in the SFC on the third The state corresponding to the process in the data center node.

The fourth construction module 405 is configured to construct a hidden Markov model based on the constructed initial probability set, transition probability matrix, and output probability matrix.

The orchestration module 406 is used to use the initial probability, transition probability and output probability in the hidden Markov model to calculate the hidden state probability corresponding to the deployment of the network function node in the SFC in the data center node, and based on the hidden state probability. Each SFC in the SFC is arranged to obtain the hidden state subsequence corresponding to each SFC, and multiple hidden state subsequences corresponding to multiple SFCs are obtained; the elements contained in the hidden state subsequence are: deployed by the network function nodes in the SFC Data center nodes.

It should be noted that the device in the embodiment of the present invention is a device corresponding to a cross-domain SFC layout method shown in FIG. 1 , and all embodiments of a cross-domain SFC layout method shown in FIG. 1 are applicable to The device, and all can achieve the same or similar beneficial effects.

A cross-domain SFC orchestration device provided by the embodiment of the present invention cannot be directly observed due to the hidden state of HMM, but can be observed through observable sequences, and each observable sequence is expressed as various states through probability density distribution Each observable sequence is generated by a state sequence with a corresponding probability density distribution. In the embodiment of the present invention, the arrangement of cross-domain SFC is modeled as a hidden Markov model, and then the hidden Markov model is used to Each SFC in multiple SFCs is arranged to obtain the hidden state subsequence corresponding to each SFC, because it is based on the calculated hidden state probability corresponding to the network function nodes in the SFC deployed in the data center nodes, and the network function of the SFC Deploying nodes to implement SFC orchestration can reduce the cross-domain bandwidth overhead of SFC after orchestration, thereby reducing the bandwidth resources consumed by SFC after orchestration.

Optionally, the initial probabilities of deploying different network function nodes in different data center nodes are determined by using a first preset expression.

Using the second preset expression, a transition probability from the first state to the second state is determined.

Using the third preset expression, determine the output probability of the function type corresponding to the third network function node in the output SFC in the third state.

The first preset expression is:

In the formula, π _m represents the initial probability of deploying the network function node on the mth data center node, M represents the number of data center nodes, Indicates the sth VNF instance in the mth data center node, Indicates the overhead generated by deploying the first network function node to the sth VNF instance in the mth data center node, Indicates the transfer bandwidth overhead generated by deploying the initial network function node on the initial data center node and the first network function node on the mth data center node. 01 indicates that the initial network function node of the SFC goes to the first network The state transition of the function node, σm represents the transition from the starting data center node to the mth data center node.

The second preset expression is:

In the formula, Indicates status transfer to state The transition probability of the state Indicates the state corresponding to the process of deploying the i-1th network function node in the nth data center node, state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the transfer bandwidth cost of deploying the (i-1)th network function node in data center node n, and deploying the i-th network function node in data center m, (i-1)i represents the (i-1)th network The transfer from the function node to the i-th network function node, nm means the transfer from the n-th data center node to the m-th data center node, Represents the unit bandwidth cost from data center node n to data center node m, N represents the number of network topologies of data center nodes, Indicates the amount of requested bandwidth between the (i-1)th network function node of the p-th SFC and the i-th network function node.

The third preset expression is:

In the formula, Indicates status The following output network function type probability of state Indicates the state corresponding to the process of deploying the i-th network function node in the m-th data center node, Indicates the function type of the i-th network function node of the p-th SFC, Indicates status The network function type is not output under The probability, Indicates the overhead generated by deploying the i-th network function node to the s-th VNF instance in the m-th data center node, Indicates the reliability of the sth VNF instance in the mth data center node, Indicates the function type of the sth VNF instance in the mth data center node.

Optionally, multiple SFCs to be programmed are SFC sets; use the initial probability, transition probability and output probability in the hidden Markov model to calculate the hidden state probability corresponding to the network function nodes in the SFC deployed in the data center nodes , and arrange each SFC in multiple SFCs based on the hidden state probability, and obtain the hidden state subsequences corresponding to each SFC, and obtain the multiple hidden state subsequences corresponding to multiple SFCs as:

Using the initial probability, transition probability, and output probability in the hidden Markov model, calculate the hidden state probability corresponding to the deployment of the network function node in the SFC in the data center node, and based on the hidden state probability for each SFC in multiple SFCs Perform arrangement to obtain the hidden state subsequence corresponding to each SFC, and obtain the hidden state sequence set corresponding to the SFC set.

Optionally, the orchestration module 406 includes:

The judging sub-module is used to judge whether there is an SFC to be edited in the SFC set.

The selection sub-module is used to select the longest SFC in the SFC set as the currently programmed SFC when there are SFCs to be edited in the SFC set.

The orchestration sub-module is used to use the initial probability, transition probability and output probability in the hidden Markov model to calculate the hidden state probability corresponding to the deployment of each network function node in the current orchestration SFC in the data center node, and based on the hidden state The state probability arranges the current arranged SFC, and obtains the hidden state subsequence corresponding to the current arranged SFC.

The output sub-module is used to output the hidden state sequence set corresponding to the SFC set when there is no SFC to be programmed in the SFC set.

Optionally, orchestrate submodules, specifically for:

Determine whether the current programming SFC is completed;

If the current programming SFC has not been programmed, it is judged whether the current programming network function node of the current programming SFC is the first network function node of the current programming SFC;

If the current orchestration network function node of the current orchestration SFC is the first network function node of the current orchestration SFC, calculate and deploy the current orchestration network function node in each data center node based on the initial probability in the hidden Markov model The initial hidden state probability of ;

If the current orchestration network function node of the current orchestration SFC is not the first network function node of the current orchestration SFC, use the transition probability and output probability in the hidden Markov model to calculate the deployment of the current orchestration network function node in each data The maximum hidden state probability in the central node, and record the position of the front data center node corresponding to the maximum hidden state probability;

Using the next network function node of the currently programmed network function node of the currently programmed SFC as the current programmed network function node of the currently programmed SFC, the step of judging whether the current programmed SFC is completed is executed;

If the arrangement of the current SFC is completed, the network function node at the end of the currently arranged SFC is used as the current network function node, and the fourth data center node corresponding to the maximum hidden state probability of the current network function node is selected to deploy the current network function node, and the second The four data center nodes are stored in the hidden state subsequence corresponding to the current orchestration SFC;

The preceding data center node corresponding to the maximum hidden state probability of the current network function node is determined as the fifth data center node corresponding to the previous network function node of the current network function node, and the current network is deployed in the fifth data center node The previous network function node of the function node, storing the fifth data center node in the hidden state subsequence corresponding to the current orchestration SFC, and using the previous network function node of the current network function node as the current network function node;

Judging whether the previous network function node of the current network function node is the first network function node of the current SFC arrangement;

If the previous network function node of the current network function node is not the first network function node that currently arranges the SFC, the execution will maximize the hidden state probability of the current network function node. The corresponding previous data center node is determined as the current network function node The step of the fifth data center node corresponding to the preceding network function node;

If the previous network function node of the current network function node is the first network function node of the current composed SFC, the current composed SFC is deleted from the SFC set.

Optionally, the device in the embodiment of the present invention further includes: a backup module, configured to use the hidden state sequence set, the reliability value corresponding to each SFC after arrangement, and each of the SFCs corresponding to each hidden state subsequence in the hidden state sequence set The cost-benefit value of the VNF corresponding to the network function node is used to back up the VNF.

Optionally, backup modules, specifically for:

Traverse each hidden state subsequence in the hidden state sequence set, and use the SFC corresponding to the hidden state subsequence as the current SFC;

Calculate the reliability value of the current SFC;

Judging whether the reliability value of the current SFC is less than the preset reliability value;

When the reliability value of the current SFC is less than the preset reliability value, the current SFC is placed in the first set, and the VNF that has passed the current SFC is placed in the second set;

Determine whether the first collection is empty;

When the first set is not empty, calculate the cost-benefit value of each VNF in the second set;

Back up the VNF corresponding to the maximum cost-benefit value;

Calculate the reliability value of each SFC in the first set after backup;

If the reliability value of the SFC in the first set after backup is not less than the preset reliability value, then delete the SFC from the first set, and perform the step of judging whether the first set is empty.

The embodiment of the present invention also provides an electronic device, as shown in FIG. complete the mutual communication,

Memory 503, used to store computer programs;

The processor 501 is configured to implement a cross-domain SFC arrangement provided by the embodiment of the present invention when executing the program stored in the memory 503 .

An electronic device provided by an embodiment of the present invention cannot be directly observed due to the hidden state of the HMM, but it can be observed through an observable sequence, and each observable sequence is expressed as various states through a probability density distribution, and each observable sequence The observation sequence is generated by a state sequence with a corresponding probability density distribution. In the embodiment of the present invention, the arrangement of cross-domain SFC is modeled as a hidden Markov model, and then the hidden Markov model is used to calculate each One SFC is arranged to obtain the hidden state subsequence corresponding to each SFC, because it is based on the calculated hidden state probability corresponding to the network function nodes in the SFC deployed in the data center nodes, and the network function nodes of the SFC are deployed to realize The orchestration of the SFC can reduce the cross-domain bandwidth overhead of the orchestrated SFC, thereby reducing the bandwidth resources consumed by the orchestrated SFC.

The communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus or the like. The communication bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface is used for communication between the electronic device and other devices.

The memory may include a random access memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far away from the aforementioned processor.

Above-mentioned processor can be general-purpose processor, comprises central processing unit (Central Processing Unit, CPU), network processor (Network Processor, NP) etc.; Can also be Digital Signal Processor (Digital Signal Processing, DSP), ASIC (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In yet another embodiment provided by the present invention, a computer-readable storage medium is also provided, and a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any one of the above cross-connected Steps of the orchestration method of domain SFC.

In yet another embodiment provided by the present invention, a computer program product including instructions is also provided, and when it is run on a computer, it causes the computer to execute any cross-domain SFC orchestration method in the above embodiments.

It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is a relationship between these entities or operations. any such actual relationship or order exists between them. Furthermore, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article or apparatus comprising a set of elements includes not only those elements, but also includes elements not expressly listed. other elements of or also include elements inherent in such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising said element.

Each embodiment in this specification is described in a related manner, the same and similar parts of each embodiment can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device/electronic device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for relevant parts, please refer to part of the description of the method embodiment.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present invention are included in the protection scope of the present invention.

Claims (10)

1. A method for orchestrating cross-domain SFCs, the method comprising:

acquiring a plurality of Service Function Chains (SFCs) to be arranged and a plurality of data center nodes, wherein each SFC comprises a plurality of network function nodes, each data center node is used for deploying the network function nodes in the SFC, and one network function node is deployed in one data center node;

determining initial probabilities of deploying different network function nodes in different data center nodes based on Virtual Network Function (VNF) overheads and bandwidth overheads generated when the different network function nodes are deployed in different data center nodes, and constructing an initial probability set based on the determined initial probabilities;

constructing a transition probability matrix based on the transition probability determined by the transition of the first state to the second state; the first state corresponds to a process of deploying a first network function node in the SFC in a first data center node, the second state corresponds to a process of deploying a second network function node in the SFC in a second data center node, and the second network function node is a neighboring node of the first network function node;

constructing an output probability matrix based on the output probability of the function type corresponding to the third network function node in the SFC under the third state; the third state is a state corresponding to a process of deploying a third network function node in the SFC in a third data center node;

constructing a hidden Markov model based on the constructed initial probability set, the transition probability matrix, and the output probability matrix;

calculating the hidden state probability corresponding to the network function nodes in the SFCs deployed in the data center nodes by using the initial probability, the transition probability and the output probability in the hidden Markov model, and arranging each SFC in the plurality of SFCs based on the hidden state probability to obtain hidden state subsequences corresponding to each SFC and obtain a plurality of hidden state subsequences corresponding to the plurality of SFCs; the hidden state subsequence comprises the following elements: and the data center node is deployed on the network function node in the SFC.

2. The method of claim 1, wherein an initial probability of deploying the different network function nodes in the different data center nodes is determined using a first preset expression;

determining the transition probability of the first state to the second state by using a second preset expression;

determining the output probability of the function type corresponding to the third network function node in the output SFC under the third state by using a third preset expression;

the first preset expression is as follows:

in the formula, pi_mRepresenting an initial probability of deploying the network function node to an mth data center node, M representing a number of data center nodes,representing the s-th VNF instance in the m-th data center node,represents the overhead incurred by deploying the first network function node to the s-th VNF instance in the m-th data center node,representing the bandwidth transfer overhead generated by deploying an initial network function node to an initial data center node, deploying a first network function node to an m-th data center node, 01 representing the state transfer from the initial network function node to the first network function node of the SFC, and σ m representing the transfer from the initial data center node to the m-th data center node;

the second preset expression is as follows:

in the formula, presentation formState of the artTransition to StateTransition probability, state ofRepresents the state corresponding to the process of deploying the (i-1) th network function node in the nth data center nodeRepresents the state corresponding to the process of deploying the ith network function node in the mth data center node,represents the bandwidth overhead of the (i-1) th network function node deployed in the data center node n and the ith network function node deployed in the data center m, wherein (i-1) i represents the transfer from the (i-1) th network function node to the ith network function node, nm represents the transfer from the nth data center node to the mth data center node,representing the unit bandwidth cost from data center node N to data center node m, N representing the number of network topologies of data center nodes,representing the requested bandwidth amount between the (i-1) th network function node and the ith network function node of the p-th SFC;

the third preset expression is as follows:

in the formula,indicating a stateLower output network function typeProbability, state ofRepresents the state corresponding to the process of deploying the ith network function node in the mth data center node,indicates the function type of the i-th network function node of the p-th SFC,indicating a stateType of network function not outputThe probability of (a) of (b) being,represents the overhead incurred by deploying the ith network function node to the s-th VNF instance in the m-th data center node,indicating the reliability of the s-th VNF instance in the m-th data center node,indicating the function type of the s-th VNF instance in the m-th data center node.

3. The method of claim 2, wherein the plurality of SFCs to be orchestrated are a set of SFCs; calculating, by using the initial probability, the transition probability, and the output probability in the hidden markov model, a hidden state probability corresponding to a network function node in the SFC deployed in a data center node, and arranging each of the SFCs based on the hidden state probability to obtain a hidden state subsequence corresponding to each SFC, where the obtained multiple hidden state subsequences corresponding to the multiple SFCs are:

calculating the hidden state probability corresponding to the network function nodes in the SFCs deployed in the data center nodes by using the initial probability, the transition probability and the output probability in the hidden Markov model, and arranging each SFC in the plurality of SFCs based on the hidden state probability to obtain the hidden state subsequence corresponding to each SFC and obtain the hidden state sequence set corresponding to the SFC set.

4. The method according to claim 3, wherein the step of calculating a hidden state probability corresponding to a network function node in the SFC deployed in a data center node by using the initial probability, the transition probability and the output probability in the hidden markov model, and arranging each SFC in the SFCs based on the hidden state probability to obtain a hidden state subsequence corresponding to each SFC, and obtain a hidden state sequence set corresponding to the SFC set, comprises:

judging whether the SFC set contains SFCs to be arranged or not;

if the SFC set comprises the SFCs to be programmed, selecting the longest SFC in the SFC set as the current SFC to be programmed;

calculating the hidden state probability corresponding to each network function node in the current layout SFC deployed in a data center node by using the initial probability, the transition probability and the output probability in the hidden Markov model, and arranging the current layout SFC based on the hidden state probability to obtain a hidden state subsequence corresponding to the current layout SFC;

and if the SFC set does not have the SFC to be arranged, outputting a hidden state sequence set corresponding to the SFC set.

5. The method according to claim 4, wherein the step of calculating a hidden state probability corresponding to each network function node deployed in a data center node in the currently deployed SFC by using the initial probability, the transition probability and the output probability in the hidden Markov model, and deploying the currently deployed SFC based on the hidden state probabilities to obtain a hidden state subsequence corresponding to the currently deployed SFC comprises:

judging whether the current layout SFC is finished or not;

if the current layout SFC is not completely laid, judging whether the current layout network function node of the current layout SFC is the first network function node of the current layout SFC;

if the current arranging network function node of the current arranging SFC is the first network function node of the current arranging SFC, calculating the initial hidden state probability of arranging the current arranging network function node in each data center node based on the initial probability in the hidden Markov model;

if the current arranging network function node of the current arranging SFC is not the first network function node of the current arranging SFC, calculating the maximum hidden state probability of arranging the current arranging network function node in each data center node by using the transition probability and the output probability in the hidden Markov model, and recording the position of a preposed data center node corresponding to the maximum hidden state probability;

taking the next network function node of the current arranging SFC as the current arranging network function node of the current arranging SFC, and executing the step of judging whether the current arranging SFC is arranged;

if the current SFC is arranged completely, taking the last network function node of the current SFC as the current network function node, selecting a fourth data center node which corresponds to the maximum hidden state probability of the current network function node to deploy the current network function node, and storing the fourth data center node in a hidden state subsequence corresponding to the current SFC;

determining a pre-data center node which corresponds to the maximum hidden state probability of the current network function node as a fifth data center node corresponding to a network function node which is before the current network function node, deploying the network function node which is before the current network function node in the fifth data center node, storing the fifth data center node in a hidden state subsequence which corresponds to the current layout SFC, and taking the network function node which is before the current network function node as the current network function node;

judging whether a network function node before the current network function node is the first network function node of the current SFC;

if the previous network function node of the current network function node is not the first network function node of the current SFC, executing a step of determining a preposed data center node corresponding to the hidden state probability of the current network function node to be the maximum as a fifth data center node corresponding to the previous network function node of the current network function node;

deleting the current orchestrated SFC from the set of SFCs if a previous network function node to the current network function node is a first network function node of the current orchestrated SFC.

6. The method of claim 3, further comprising:

and backing up the VNF based on the hidden state sequence set, the reliability value corresponding to each SFC after arrangement and the cost benefit value of the VNF corresponding to each network function node in the SFC corresponding to each hidden state subsequence in the hidden state sequence set.

7. The method of claim 6, wherein the step of backing up the VNF based on the set of hidden-state sequences, the reliability value corresponding to each SFC after the arranging, and the cost-benefit value of the VNF corresponding to each network function node in the SFC corresponding to each hidden-state subsequence in the set of hidden-state sequences comprises:

traversing each hidden state subsequence in the hidden state sequence set, and taking the SFC corresponding to the hidden state subsequence as the current SFC;

calculating a reliability value of the current SFC;

judging whether the reliability value of the current SFC is smaller than a preset reliability value or not;

when the reliability value of the current SFC is smaller than a preset reliability value, placing the current SFC in a first set, and placing VNFs passing through the current SFC in a second set;

determining whether the first set is empty;

calculating a cost-benefit value for each VNF in the second set when the first set is not empty;

backing up the VNF corresponding to the maximum cost benefit value;

calculating the reliability value of each SFC in the first set after backup;

and if the reliability value of the SFC in the first set is not less than the preset reliability value after backup, deleting the SFC from the first set, and executing the step of judging whether the first set is empty or not.

8. An apparatus for orchestration of cross-domain SFCs, the apparatus comprising:

the system comprises an acquisition module, a scheduling module and a scheduling module, wherein the acquisition module is used for acquiring a plurality of Service Function Chains (SFCs) to be scheduled and a plurality of data center nodes, each SFC comprises a plurality of network function nodes, each data center node is used for deploying the network function nodes in the SFC, and one network function node is deployed in one data center node;

the first construction module is used for determining initial probabilities of deploying different network function nodes in different data center nodes based on Virtual Network Function (VNF) overheads and bandwidth overheads generated when the different network function nodes are deployed in different data center nodes, and constructing an initial probability set based on the determined initial probabilities;

the second construction module is used for constructing a transition probability matrix based on the transition probability determined by the transition of the first state to the second state; the first state corresponds to a process of deploying a first network function node in the SFC in a first data center node, the second state corresponds to a process of deploying a second network function node in the SFC in a second data center node, and the second network function node is a neighboring node of the first network function node;

the third construction module is used for constructing an output probability matrix based on the output probability of the function type corresponding to the third network function node in the SFC under the third state; the third state is a state corresponding to a process of deploying a third network function node in the SFC in a third data center node;

a fourth construction module, configured to construct a hidden markov model based on the constructed initial probability set, the transition probability matrix, and the output probability matrix;

the arranging module is used for calculating the hidden state probability corresponding to the network function nodes in the SFCs deployed in the data center nodes by utilizing the initial probability, the transition probability and the output probability in the hidden Markov model, arranging each SFC in the plurality of SFCs based on the hidden state probability, obtaining the hidden state subsequences corresponding to the SFCs and obtaining a plurality of hidden state subsequences corresponding to the SFCs; the hidden state subsequence comprises the following elements: and the data center node is deployed on the network function node in the SFC.

9. An electronic device is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing mutual communication by the memory through the communication bus;

a memory for storing a computer program;

a processor for implementing the method steps of any of claims 1 to 7 when executing a program stored in the memory.

10. A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, which computer program, when being executed by a processor, carries out the method steps of any one of claims 1 to 7.