Network function state backup method and device, electronic equipment and storage medium
Generate the optimal network function service chain through node clustering system and computing power model, solving the problems of low network function state backup efficiency and large computing overhead in the existing technology, realizing efficient network function state backup and reducing service interruption risk.
Patent Information
- Application Number
- CN202410983661.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-26
AI Technical Summary
The existing edge service technology cannot effectively consider the specific needs of each terminal, resulting in low network function status backup efficiency, large calculation overhead, and inability to handle network function failures in real time, resulting in service interruption.
The node clustering system and computing power model are adopted to determine the cluster initialization center through weighted probability distribution, perform edge node clustering and center of mass update, generate the optimal network functional service chain, and realize network functional status backup.
It improves the efficiency of network function status backup, reduces computing power calculation overhead, realizes efficient backup of network function status, and avoids service interruption.
Smart Images

Figure CN119025331A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of edge computing technology, and in particular to a network function status backup method, device, electronic device and storage medium. Background Art
[0002] At present, edge service technology is usually used to achieve network function backup. The existing edge service technology only selects edge nodes on the terminal side or the network side. However, the network side cannot consider the specific needs of each terminal; and the computing power on the terminal side cannot summarize the information of all possible edge nodes in the candidate edge computing system.
[0003] In addition to backing up virtual network functions, the status of network functions also needs to be backed up in real time. However, the time point when a network function fails cannot be estimated, so the backup of network status requires a large amount of storage resources. In virtual network functions, there are a large number of network functions that need to update their status in real time, such as the Network Function Address Translator (NAT) intrusion detection system. When a data packet passes through a network function, it is possible to change the current status of the network function. For network functions, if the backup of its functions is only implemented on different servers, it is often impossible to process data packets normally, resulting in service interruption. Summary of the invention
[0004] The main purpose of the embodiments of the present application is to propose a network function status backup method, device, electronic device and storage medium, which can effectively solve the network function status backup problem, improve the network function status backup efficiency, and reduce computing power calculation overhead.
[0005] On the one hand, an embodiment of the present application provides a method for backing up a network function status, the method comprising the following steps:
[0006] Obtain multiple network functions to be backed up and computing tasks to be processed;
[0007] Get multiple edge nodes in the edge network;
[0008] Initializing nodes of the edge network by using a node clustering system, and determining each cluster initialization center from a plurality of edge nodes by using a weighted probability distribution;
[0009] Clustering is performed in the edge network according to each of the cluster initialization centers by the node clustering system to obtain edge node clusters corresponding to each of the cluster initialization centers;
[0010] The node clustering system is used to update the centroid of each edge node cluster, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine a number of edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster;
[0011] Using a computing power quantification model and a virtual mapping system, an optimal network function service chain is generated according to the computing power tasks to be processed, each of the functional service nodes and each of the network functions to be backed up, and according to the optimal network function service chain, a target service node corresponding to each of the network functions to be backed up is determined from a plurality of the functional service nodes;
[0012] For each network function to be backed up, the target backup service nodes corresponding to the target service node are used to back up the network function status of the network function to be backed up according to a preset backup time interval, wherein the target service node is the backup service node in the edge node cluster corresponding to the target service node.
[0013] In some embodiments, the node initialization of the edge network by using a node clustering system and determining each cluster initialization center from a plurality of edge nodes by using a weighted probability distribution specifically includes:
[0014] Randomly selecting an initial node center from a plurality of edge nodes through the node clustering system, obtaining the initial node center as the current node center, obtaining the first round as the current iteration round number, and determining the current node center as the cluster initialization center;
[0015] Calculate the distance between each of the edge nodes except the cluster initialization center in the edge network and the current node center, and determine the next node center from the remaining edge nodes using weighted probability distribution according to the distance between each of the edge nodes and the current node center;
[0016] When the current iteration round number is less than the number corresponding to the network functions to be backed up, the next node center is determined as the cluster initialization center, the next node center is obtained as the current node center, the current iteration round number is incremented to obtain the next iteration round number, the next iteration round number is obtained as the current iteration round number, and then the distance between each edge node except the cluster initialization center and the current node center in the edge network is calculated, and the next node center is determined from the remaining edge nodes using weighted probability distribution according to the distance between each edge node and the current node center, until the current iteration round number is equal to the number corresponding to the network functions to be backed up;
[0017] When the current iteration round number is equal to the number corresponding to the network functions to be backed up, the node initialization of the edge network through the node clustering system is stopped.
[0018] In some embodiments, clustering in the edge network according to each cluster initialization center by the node clustering system to obtain edge node clusters corresponding to each cluster initialization center specifically includes:
[0019] Calculating the distance between each edge node to be clustered and each of the cluster initialization centers, wherein the edge nodes to be clustered are the remaining edge nodes in the edge network except for the plurality of cluster initialization centers;
[0020] For each edge node to be clustered, according to the distance between the edge node to be clustered and each cluster initialization center, determine the target cluster initialization center corresponding to the edge node to be clustered from the multiple cluster initialization centers, and add the edge node to be clustered to the edge node cluster corresponding to the target cluster initialization center, wherein, among the multiple cluster initialization centers, the target cluster initialization center is the closest to the edge node to be clustered.
[0021] In some embodiments, the node clustering system is used to update the centroid of each edge node cluster, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine several edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster, specifically including:
[0022] Get the termination condition of iterative correction;
[0023] For each of the edge node clusters, the cluster initialization center corresponding to the edge node cluster is iteratively corrected by the node clustering system until the corrected cluster initialization center meets the iterative correction termination condition, the corrected cluster initialization center is determined as the cluster center corresponding to the edge node cluster, and several edge nodes in the edge node cluster except the cluster center are determined as the backup service nodes corresponding to the edge node cluster.
[0024] In some embodiments, the use of a computing power quantization model and a virtual mapping system generates an optimal network function service chain according to the computing power tasks to be processed, each of the function service nodes and each of the network functions to be backed up, and determines a target service node corresponding to each of the network functions to be backed up from a plurality of the function service nodes according to the optimal network function service chain, specifically including:
[0025] Construct the computing power quantification model and obtain the computing power resource information corresponding to each of the functional service nodes;
[0026] Determine each optimal service node from a plurality of the functional service nodes using the computing power quantification model according to the computing power tasks to be processed and the computing power resource information corresponding to each of the functional service nodes;
[0027] Determine, by the virtual mapping system, the optimal node arrangement order corresponding to the plurality of the optimal service nodes according to the computing power tasks to be processed and the network functions to be backed up, and generate the optimal network function service chain according to the optimal node arrangement order and the optimal service nodes;
[0028] Determine multiple function node allocation schemes according to each of the network functions to be backed up and each of the optimal service nodes;
[0029] Allocate each of the network functions to be backed up to each of the optimal service nodes in the optimal network function service chain according to each of the functional node allocation schemes through the virtual mapping system, perform virtual network mapping, and calculate the virtual network mapping operation cost corresponding to each of the functional node allocation schemes;
[0030] According to the virtual network mapping operation cost corresponding to each of the function node allocation schemes, an optimal function node allocation scheme is determined from multiple function node allocation schemes, and according to the optimal function node allocation scheme, the target service node corresponding to each of the network functions to be backed up is determined.
[0031] In some embodiments, the computing quantization model is represented by the following formula:
[0032]
[0033] Among them, C br is the computing power requirement of the computing task to be processed, n is the number of functional service nodes that process computing tasks through logical computing capabilities, and a i The weight of the logical computing capability of the i-th functional service node is mapped to f(a i ) is the mapping function of the logical computing capability provided by the i-th functional service node, q 1i is the redundant logical computing capacity of the i-th functional service node, m is the number of functional service nodes that process computing tasks through parallel computing capacity, and b j The weight of the parallel computing capability of the jth function service node is mapped to f(b j ) is the mapping function of the parallel computing capability provided by the jth functional service node, q 2j is the redundant parallel computing capacity of the jth functional service node, p is the number of functional service nodes that process computing tasks through neural network acceleration capabilities, and yk The weight of the neural network acceleration capability mapping of the kth function service node, f(y k ) is the mapping function of the neural network acceleration capability provided by the kth function service node, q 3k Redundant neural network acceleration capabilities for the k-th function service node.
[0034] In some embodiments, the virtual mapping system allocates each of the network functions to be backed up to each of the optimal service nodes in the optimal network function service chain according to each of the function node allocation schemes, performs virtual network mapping, and calculates the virtual network mapping operation cost corresponding to each of the function node allocation schemes, specifically including:
[0035] For each of the function node allocation schemes, determine the optimal service node corresponding to each of the network functions to be backed up in the optimal network function service chain according to the function node allocation scheme, perform virtual network mapping, map the computing power tasks to be processed to the optimal network function service chain, and calculate the virtual network mapping operation cost corresponding to the function node allocation scheme.
[0036] On the other hand, an embodiment of the present application provides a network function status backup device, the device comprising:
[0037] The first module is used to obtain multiple network functions to be backed up and computing tasks to be processed;
[0038] The second module is used to obtain multiple edge nodes in the edge network;
[0039] The third module is used to initialize nodes of the edge network through a node clustering system, and determine each cluster initialization center from a plurality of edge nodes using a weighted probability distribution;
[0040] A fourth module is used to perform clustering in the edge network according to each cluster initialization center through the node clustering system to obtain edge node clusters corresponding to each cluster initialization center;
[0041] The fifth module is used to update the centroid of each edge node cluster through the node clustering system, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine several edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster;
[0042] The sixth module is used to generate an optimal network function service chain based on the computing power tasks to be processed, the functional service nodes and the network functions to be backed up by using the computing power quantization model and the virtual mapping system, and determine the target service node corresponding to each network function to be backed up from the multiple functional service nodes according to the optimal network function service chain;
[0043] The seventh module is used to back up the network function status of each network function to be backed up according to a preset backup time interval using each target backup service node corresponding to the target service node, wherein the target service node is the backup service node in the edge node cluster corresponding to the target service node.
[0044] On the other hand, an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the network function status backup method described above when executing the computer program.
[0045] On the other hand, an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the network function status backup method described above.
[0046] The embodiments of the present application include at least the following beneficial effects: a network function status backup method, device, electronic device and storage medium provided by the present application, which determines each cluster initialization center through a node clustering system, clusters in an edge network according to each cluster initialization center, obtains edge node clusters corresponding to each cluster initialization center, updates the centroid of each edge node cluster, determines the function service node and several backup service nodes corresponding to each edge node cluster, generates an optimal network function service chain according to the computing power task to be processed, each function service node and each network function to be backed up, and determines the target service node corresponding to each network function to be backed up from multiple function service nodes according to the optimal network function service chain, and for each network function to be backed up, uses each target backup service node corresponding to the target service node to perform network function status backup on the network function to be backed up according to a preset backup time interval. The present application can realize network function status backup, efficiently solve the problem of network function status backup, improve the efficiency of network function status backup, and reduce computing power calculation overhead. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 It is a flow chart of a network function status backup method provided by an embodiment of the present application;
[0048] Figure 2is a schematic diagram of data packet processing in an embodiment of the present application;
[0049] Figure 3 is a flowchart of step S103 in an embodiment of the present application;
[0050] Figure 4 is a schematic diagram of clustering multiple edge nodes in an embodiment of the present application;
[0051] Figure 5 is a flowchart of step S106 in an embodiment of the present application;
[0052] Figure 6 is a schematic diagram of processing computing tasks through network functions deployed on a service chain in an embodiment of the present application;
[0053] Figure 7 is another schematic diagram of processing computing tasks through network functions deployed on a service chain in an embodiment of the present application;
[0054] Figure 8 It is a schematic diagram of a network function processing computing task in an embodiment of the present application;
[0055] Fig. 9 It is a structural diagram of a network function status backup device provided in an embodiment of the present application;
[0056] Fig.10 It is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0057] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are only examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the attached claims.
[0058] It is understood that the terms "first", "second", etc. used in this application can be used to describe various concepts in this article, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another concept. For example, without departing from the scope of the embodiment of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein can be interpreted as "at the time of" or "when" or "in response to determination".
[0059] The terms "at least one", "multiple", "each", "any", etc. used in this application, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.
[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.
[0061] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.
[0062] Before describing the embodiments of the present application in detail, some nouns and terms involved in the embodiments of the present application are first described. The nouns and terms involved in the embodiments of the present application are subject to the following explanations.
[0063] 1) Computing power: Computing power is the computing ability to achieve target output results by processing information data.
[0064] 2) Edge nodes: Edge nodes are relative to cloud computing data centers and refer to network nodes with fewer intermediate links between them and the final access users. They can be a computer room or a physical device. Compared with direct access to the source station, the end user has better responsiveness and connection speed when accessing.
[0065] 3) k-means: k-means clustering algorithm is an iterative clustering analysis algorithm. Its steps are: divide the data into K groups in advance, randomly select K objects as the initial cluster centers, and then calculate the distance between each object and each seed cluster center, and assign each object to the cluster center closest to it. The cluster centers and the objects assigned to them represent a cluster. Every time a sample is assigned, the cluster center of the cluster will be recalculated based on the existing objects in the cluster. This process will be repeated until a certain termination condition is met. The termination condition can be that no (or a minimum number) of objects are reassigned to different clusters, no (or a minimum number) of cluster centers change again, and the sum of squared errors is locally minimized.
[0066] In the related technology, edge services are implemented based on edge nodes. Edge nodes are business platforms (such as servers and base stations) built at the edge of the network side close to the user side. They can provide storage, computing, network and other resources. By executing some applications (edge applications) of the terminal (User Equipment, UE) on the edge nodes, it is equivalent to "sinking" some services to the edge of the access network, which can reduce bandwidth occupancy and delay loss caused by multi-layer network transmission. The existing edge service discovery technology only selects edge nodes on the terminal side or the network side. However, the network side cannot consider the specific needs of each terminal; and the computing power of the terminal side cannot be developed, and it is obviously impossible to summarize the information of edge nodes in all possible candidate edge computing systems. In addition to backing up virtual network functions, the status of network functions also needs to be backed up in real time. However, the time point when the network function fails cannot be estimated, so the backup of network status requires a lot of storage resources.
[0067] In virtual network functions, there are a large number of network functions that require real-time status updates, such as the Network Function Address Translator (NAT) intrusion detection system. When a data packet passes through a network function, it is possible to change the current status of the network function. For such network functions, if their functions are only backed up on different servers, they often cannot process data packets normally, resulting in service interruption.
[0068] Based on this, the embodiments of the present application propose a network function status backup method, device, electronic device and storage medium, which can efficiently solve the problem of network function status backup, back up the network function status of computing tasks to the edge network, and use a virtual mapping system to obtain the optimal resource allocation strategy for the service chain, thereby improving the reliability of the edge network service chain and solving the problem of network function status backup.
[0069] Reference Figure 1 , Figure 1This is an optional flowchart of a network function status backup method provided by an embodiment of the present application. The method may include but is not limited to steps S101 to S107:
[0070] Step S101, obtaining multiple network functions to be backed up and computing tasks to be processed;
[0071] Step S102, obtaining multiple edge nodes in the edge network;
[0072] Step S103, initializing nodes of the edge network through the node clustering system, and determining each cluster initialization center from multiple edge nodes using weighted probability distribution;
[0073] Step S104, clustering is performed in the edge network according to each cluster initialization center by the node clustering system to obtain edge node clusters corresponding to each cluster initialization center;
[0074] Step S105, updating the centroid of each edge node cluster through the node clustering system, determining the cluster center corresponding to each edge node cluster, determining each cluster center as the corresponding functional service node, and determining several edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster;
[0075] Step S106, using the computing power quantization model and the virtual mapping system, generating an optimal network function service chain according to the computing power tasks to be processed, each functional service node and each network function to be backed up, and determining the target service node corresponding to each network function to be backed up from multiple functional service nodes according to the optimal network function service chain;
[0076] Step S107, for each network function to be backed up, using each target backup service node corresponding to the target service node to perform a network function status backup for the network function to be backed up according to a preset backup time interval, wherein the target service node is a backup service node in the edge node cluster corresponding to the target service node.
[0077] In some embodiments, a hot backup mechanism is adopted, that is, the data packet passes through the original network function and the backup network function. When the original network function fails, it can seamlessly switch to the backup network function; setting a certain time interval for the backup time point of the network function, that is, the above-mentioned backup time interval, can effectively reduce network transmission overhead.
[0078] In some embodiments, the above-mentioned network function status backup method combines a preset backup time interval and a hot backup mechanism on the software side to implement network function status backup, and combines a cold backup mechanism, a preset backup time interval and a hot backup mechanism on the hardware side to back up the network function status to a backup service node, thereby solving the problem of being unable to update the network function status in real time due to service interruption.
[0079] In some embodiments, the network function is a multi-threaded service running on a multi-core CPU. The network function uses a fixed number of threads to run, each thread reads data from its own receiving queue, and sends it to the output queue after processing. Each thread accesses or modifies a fixed shared state together. Therefore, the execution order of each thread will directly affect the value of the shared state.
[0080] Exemplarily, the process of processing each data packet is represented by a four-tuple: (p, n, v, s), where p represents a data packet, v represents a shared data unit, n represents the number of times the data packet p accesses the shared data unit v, and s represents the total number of times the shared data unit v is accessed. Figure 2 , there are four data packets A, B, C, and D processed in four threads respectively, (A, 1, X, 1) indicates that the number of times data packet A accesses the shared data unit X is 1, and the shared data unit X is accessed 1 time in total. Such a four-tuple represents the dependency between packets. This four-tuple record can effectively remove the uncertainty of data packets running in multiple threads. Combined with the preset backup time interval and hot backup mechanism, the network function status is backed up at the backup time interval, and the data packets to be processed are stored between the backup time intervals. When the network function fails, the network function status can be restored by restoring the status of the previous node and running all the data packets before the node crashes.
[0081] In some embodiments, reference Figure 3 , Figure 3 This is an optional flowchart of step S103 in the embodiment of the present application. Step S103 may include but is not limited to steps S201 to S204:
[0082] Step S201, randomly selecting an initial node center from multiple edge nodes through a node clustering system, obtaining the initial node center as the current node center, obtaining the first round as the current iteration round number, and determining the current node center as the cluster initialization center;
[0083] Step S202, calculating the distance between each edge node except the cluster initialization center and the current node center in the edge network, and determining the next node center from the remaining edge nodes using weighted probability distribution according to the distance between each edge node and the current node center;
[0084] Step S203, when the current iteration number is less than the number corresponding to the network functions to be backed up, the next node center is determined as the cluster initialization center, the next node center is obtained as the current node center, the current iteration number is incremented, the next iteration number is obtained, the next iteration number is obtained as the current iteration number, and then the distance between each edge node in the edge network except the cluster initialization center and the current node center is calculated, and the next node center is determined from the remaining edge nodes using weighted probability distribution according to the distance between each edge node and the current node center, until the current iteration number is equal to the number corresponding to the network functions to be backed up;
[0085] Step S204: when the current iteration round number is equal to the number corresponding to the network functions to be backed up, stop initializing the nodes of the edge network through the node clustering system.
[0086] In some embodiments, the node clustering system uses an improved k-means algorithm to accurately cluster the edge network to generate two data sets during the execution of the hot backup mechanism, thereby reducing the uncertainty of data flow operation. Optionally, the improved k-means algorithm is set to a K-means++ algorithm.
[0087] When the K-means algorithm selects K initial centroids (such as the cluster initialization centers mentioned above), it adopts a completely random selection method, which will cause the algorithm to converge very slowly. The K-means++ algorithm is an optimization of the K-means method of completely random selection of initialization centroids. The most essential difference between the K-means++ algorithm and the K-means algorithm is the initialization process of the K cluster centers.
[0088] The basic principle of the K-means++ algorithm in the process of initializing cluster centers is to make the distance between the initial cluster centers as far as possible, so as to avoid the above problems.
[0089] In some embodiments, illustratively, the node clustering system uses the K-means++ algorithm to perform node initialization, and the specific process of determining the initialization center of each cluster is as follows:
[0090] In the first step, a node is randomly selected from multiple edge nodes as the initial node center, the initial node center is determined as the cluster initialization center, and the initial node center is obtained as the current node center u;
[0091] The second step is to calculate the edge nodes x except the cluster initialization center. i The distance from the current node center u is obtained Among them, d(x i ,u) is the edge node x i The distance from the current node center u, x iis the i-th edge node, j is the number of edge nodes;
[0092] The third step is based on The weighted probability distribution is used to determine the next node center z from the remaining edge nodes, and z is determined as the cluster initialization center, where the probability of selecting z as the next node center is equal to is proportional to;
[0093] Step 4. When the number of cluster initialization centers is less than the number of network functions to be backed up, obtain z as the current node center u, and repeat the above steps 2 and 3 until the number of cluster initialization centers is equal to the number of network functions to be backed up. When the number of cluster initialization centers is equal to the number of network functions to be backed up, stop node initialization and obtain k cluster initialization centers, where k is the number of network functions to be backed up.
[0094] In some embodiments, step S104 may include but is not limited to steps S301 to S302:
[0095] Step S301, calculating the distance between each edge node to be clustered and each cluster initialization center, wherein the edge nodes to be clustered are the remaining edge nodes in the edge network except for a plurality of cluster initialization centers;
[0096] Step S302: for each edge node to be clustered, determine the target cluster initialization center corresponding to the edge node to be clustered from multiple cluster initialization centers according to the distance between the edge node to be clustered and each cluster initialization center, and add the edge node to be clustered to the edge node cluster corresponding to the target cluster initialization center, wherein, among the multiple cluster initialization centers, the distance between the target cluster initialization center and the edge node to be clustered is the shortest.
[0097] In some embodiments, the node clustering system uses standard K-means clustering to determine the backup nodes corresponding to each cluster initialization center and obtain the edge node clusters corresponding to each cluster initialization center.
[0098] For example, refer to Figure 4 , Figure 4 This is an optional schematic diagram of clustering multiple edge nodes in an embodiment of the present application. Assume that there is a cluster initialization center u in the edge network. 1 、u 2 and u 3 , edge nodes x to be clustered 1 to x 9 , calculate the edge nodes x to be clustered 1 Initialize the center u of each cluster separately 1 、u 2 and u 3 The distance is denoted as d 11,d 12 and d 13 , due to d 12 <d 11 <d 13 , represents the edge node x to be clustered 1 Initialize the cluster center u 2 The distance is closest, and the cluster initialization center u 2 Determine the edge node x to be clustered 1 The corresponding target cluster initialization center, that is, the edge node to be clustered x 1 Initialize the center u for the cluster 2 The corresponding backup node, and so on, determines the cluster initialization center u 2 The corresponding backup nodes include x 1 、x 3 and x 5 , cluster initialization center u 1 The corresponding backup nodes include x 2 、x 4 、x 6 and x 9 , cluster initialization center u 3 The corresponding backup nodes include x 7 and x 8 , then u 2 、x 1 、x 3 and x 5 The corresponding edge node cluster C 2 ,u 1 、x 2 、x 4 、x 6 and x 9 The corresponding edge node cluster C 1 ,u 3 、x 7 and x 8 The corresponding edge node cluster C 3 .
[0099] In some embodiments, step S105 may include but is not limited to steps S401 to S402:
[0100] Step S401, obtaining an iterative correction termination condition;
[0101] Step S402, for each edge node cluster, iteratively correct the cluster initialization center corresponding to the edge node cluster through the node clustering system until the corrected cluster initialization center meets the iterative correction termination condition, determine the corrected cluster initialization center as the cluster center corresponding to the edge node cluster, and determine several edge nodes in the edge node cluster except the cluster center as backup service nodes corresponding to the edge node cluster.
[0102] In some embodiments, the centroid of each edge node cluster is updated through a node clustering system, and an iterative correction termination condition is set. When the corrected cluster initialization center meets the iterative correction termination condition, the centroid update is stopped. Optionally, the iterative correction termination condition can be set so that the corrected cluster initialization center can minimize the sum of squared deviations of the edge node clusters under the classification of other data points at the time of iteration.
[0103] In some embodiments, reference Figure 5 , Figure 5 This is an optional flowchart of step S106 in the embodiment of the present application. Step S106 may include but is not limited to steps S501 to S506:
[0104] Step S501, constructing a computing power quantification model to obtain computing power resource information corresponding to each of the functional service nodes;
[0105] Step S502, using a computing power quantification model to determine each optimal service node from multiple functional service nodes according to the computing power tasks to be processed and the computing power resource information corresponding to each functional service node;
[0106] Step S503, determining the optimal node arrangement order corresponding to the multiple optimal service nodes according to the computing tasks to be processed and the network functions to be backed up through the virtual mapping system, and generating the optimal network function service chain according to the optimal node arrangement order and the optimal service nodes;
[0107] Step S504, determining multiple function node allocation schemes according to each network function to be backed up and each optimal service node;
[0108] Step S505, allocating each network function to be backed up to each optimal service node in the optimal network function service chain according to each function node allocation scheme through the virtual mapping system, performing virtual network mapping, and calculating the virtual network mapping operation cost corresponding to each function node allocation scheme;
[0109] Step S506, determining an optimal function node allocation scheme from multiple function node allocation schemes according to the virtual network mapping operation cost corresponding to each function node allocation scheme, and determining a target service node corresponding to each network function to be backed up according to the optimal function node allocation scheme.
[0110] In some embodiments, an optimal network function service chain is generated through a computing power quantization model and a virtual mapping system, and the target service node corresponding to each network function to be backed up in the optimal network function service chain is determined, so that the computing resource utilization rate of processing computing power tasks is optimized.
[0111] In some embodiments, the computing power quantification model is used to calculate the idle computing power of the node chip of the node that passes through the shortest hop point from the computing power task to the edge node, and the optimal service nodes that meet the computing power requirements and have the best computing resource utilization are determined from multiple functional node allocation schemes. The node arrangement order corresponding to the multiple optimal service nodes is determined according to the computing power task through the virtual mapping system, and the above-mentioned optimal network function service chain is generated. Optionally, the computing power quantification model is expressed by the following formula:
[0112]
[0113] Among them, C br is the computing power requirement of the computing task to be processed, n is the number of functional service nodes that process computing tasks through logical computing capabilities, and a i The weight of the logical computing capability of the i-th functional service node is mapped to f(a i ) is the mapping function of the logical computing capability provided by the i-th functional service node, q 1i is the redundant logical computing capacity of the i-th functional service node, m is the number of functional service nodes that process computing tasks through parallel computing capacity, and b j The weight of the parallel computing capability of the jth function service node is mapped to f(b j ) is the mapping function of the parallel computing capability provided by the jth functional service node, q 2j is the redundant parallel computing capacity of the jth functional service node, p is the number of functional service nodes that process computing tasks through neural network acceleration capabilities, and y k The weight of the neural network acceleration capability mapping of the kth function service node, f(y k ) is the mapping function of the neural network acceleration capability provided by the kth function service node, q 3k Redundant neural network acceleration capabilities for the k-th function service node.
[0114] In step S505 of some embodiments, specifically, the virtual mapping system determines the optimal service node corresponding to each network function to be backed up in the optimal network function service chain according to the functional node allocation plan for each functional node allocation plan, performs virtual network mapping, maps the computing power tasks to be processed to the optimal network function service chain, and calculates the virtual network mapping operation cost corresponding to the functional node allocation plan.
[0115] In some embodiments, the virtual mapping system adopts network function virtualization technology, which implements specific services through a series of network functions. The service chain specifically includes two parts: the order of network functions in the service chain and the division and merging of network functions in the service chain. The computing resource utilization is optimized through a reasonable order of network functions and the division and merging of network functions. The network function arrangement order with the best computing resource utilization is obtained through the relative order between the network functions in the service chain.
[0116] In some embodiments, considering the possible multi-virtual resource utilization of the nodes between the edge node and the computing task initiation point, the physical network resource unit of the link passing through the node is priced, and the node is evaluated from the cost operation input-output ratio, that is, the above-mentioned virtual network mapping operation cost, which is determined by the physical network CPU resources and link bandwidth resources used to map the virtual network. Optionally, the virtual network mapping operation cost is expressed by the following formula:
[0117]
[0118] Among them, Cost(M(G V )) is the cost of virtual network mapping operation, H(l v ) refers to the number of underlying physical network links required for successful virtual network link mapping, or called the number of hops. v ) is larger, the cost of virtual network mapping operation is greater; BW(l v ) refers to the bandwidth constraint of the virtual network link; CPU(n v ) refers to virtual network node n v CPU constraints; L V and N V Respectively represent the link set and node set of the virtual network; α C and β C It is a weight adjustment parameter used to balance the consumption of physical network node CPU and link bandwidth. In practical applications, it can be used to represent the unit pricing of physical network resource costs.
[0119] In some embodiments, it is assumed that the computing task to be processed needs to flow through the node v deployed 1 and v 2 Network functions 1 and m 2 Processing, where the network function m 1 and m 2It is a special type of network function. When the data flow passes through the network function, it will affect the size of the data flow. For example, the network optimization function of Citrix Cloudbridge will compress the data before sending it to the next network function, with a compression ratio of up to 80%. Another example is the network coding function. In order to improve the reliability of satellite communications, redundant checksum bits are added to increase the size of each data packet. For example, BCH (63,48) encoding will increase the size of the data packet by 31%.
[0120] Reference Figure 6 , Figure 6 There is a network function 1 and m 2 And the function service node v 1 、v 2 and v 3 When the initial flow of the computing task is set to 1, the computing task passes through the first network function m 1 After that, the data flow doubles to 2, through the service chain v 1 -v 2 -v 3 Transport network function 1 , the bandwidth resource occupied at this time is 2.
[0121] In the service chain, different network functions may not have dependencies, that is, the order of network functions can be adjusted. For example, there is no fixed requirement for the relative position relationship between the VPN proxy network function and the firewall network function. If the above Figure 7 Medium Service Chain v 1 -v 2 -v 3 The order of network functions in Figure 7 , the network function m 2 Deployed on the function service node v 1 In the case where the initial flow of the computing task is set to 1, when the computing task first passes through the deployment on the function service node v 1 Network functions on m 2 After that, the data flow is reduced to 0.5, and then through the service chain v 1 -v 2 -v 3 Transport network function 2 , the bandwidth resources occupied at this time are 0.5. Therefore, the bandwidth resources of the edge network can be optimized by adjusting the relative positions of network functions in the service chain.
[0122] In some embodiments, a parameter ratio(m) is defined as the ratio of the traffic size of the data flow after passing through the network function m to the traffic size before passing through the network function m, and the network functions are divided into two categories based on this:
[0123] 1) ratio(m)<1, the network function of scaling data flow;
[0124] 2)ratio(m)>1, expanding the network function of data flow.
[0125] Optionally, when the deployment of the network function meets the following characteristics, the bandwidth resource usage of the service chain is minimal, as follows:
[0126] 1) Deploy the network function that scales the data flow before the network function that amplifies the data flow;
[0127] 2) Among the network functions m with ratio(m)<1, sort the network functions in ascending order;
[0128] 3) Among the network functions with ratio(m)>1, sort the network functions in descending order.
[0129] In some embodiments, software-based network functions such as firewalls, NAT, proxies, deep packet inspection, and WAN optimization functions can be run on commercial off-the-shelf (COTS) hardware with VNFs. However, network functions are usually linked together. In this case, data packets need to be processed by a series of network functions during forwarding. If the computing resources of the upstream and downstream network functions in the service chain are not allocated reasonably, such as the throughput of the upstream network function is greater than the throughput of the downstream network function, the downstream network function may discard the data packets that have been processed by the upstream network function, which may cause serious performance degradation and waste the computing resources of the upstream edge node. Therefore, the virtual mapping system is used to reasonably process the data packet flow, adjust the deployment order of the network functions on each optimal service node, and generate the optimal service chain.
[0130] Reference Figure 8 , assuming that the computing task to be processed will pass through three network functions in sequence: VNF 1 、VNF 2 and VNF 3 , C 1 Represents the network function VNF 1 The computational overhead required to process unit traffic, C 2 Represents the network function VNF 2 The computational overhead required to process unit traffic, C 3 Represents the network function VNF 3 The computing overhead required to process a unit of traffic. The computing overhead is related to the type and complexity of network functions.
[0131] When the request rate of computing tasks is F, VNF 1 The required computing resources are FC 1, for VNF 2 Optimal computing resource allocation is not simply FC 2 This is because the upstream network function does not simply forward data like a router or switch, but affects the size of the data flow. Assuming that VNF 1 The compression ratio of the data stream is r 1 , then flows into the network function VNF 2 The data size is Fr 1 , Network Function VNF 2 The optimal computing resources required are Fr 1 C 2 According to this computing resource allocation scheme, the upstream and downstream network functions of the service chain can cooperate with each other without obvious bottlenecks. When the request rate of computing tasks changes, it is only necessary to change the size of computing resources allocated to each network function in the same proportion. For example, in the above example, the optimal proportion of computing resources allocated to each network function is VNF 1 :VNF 2 :VNF 3 =C 1 :r 1 C 2 :r 1 r 2 C 3 .
[0132] In some embodiments, the solution for computing resource allocation in the service chain is obtained by calculating the network function computing complexity and the traffic compression ratio. However, this method is based on the premise that the performance of each data packet is predictable, and there may be errors in actual situations:
[0133] ① Before the service chain is deployed and run, the network function virtualization management scheduler cannot understand the prior capabilities, capacity or processing requirements of each network function;
[0134] ② The packet cost of the same type of request is variable (for example, some packets may need to trigger a DNS lookup, while other packets may only require a simple header match);
[0135] ③ In different hardware environments, allocating the same computing resources may result in different throughputs.
[0136] Therefore, the above virtual mapping system can adjust the resource allocation strategy in real time and provide fair and effective resource allocation for the optimal network function service chain, with the following characteristics:
[0137] ① Automatically adjust the CPU allocation strategy to reasonably allocate strategic computing resources based on the packet arrival rate and the required computing resources;
[0138] ② The service chain-level congestion control is realized by using a reasonable resource allocation strategy. The real-time feedback mechanism of the downstream network function to the upstream network function can avoid the packet loss that may be caused by the downstream network;
[0139] ③ Congestion control can not only be applied to adjacent network functions in the service chain, but also extend the entire service chain and use ECN (Explicit Congestion Notification) to manage congestion between hosts.
[0140] Reference Fig. 9 , Fig. 9 : is an optional structural diagram of a network function status backup device provided in an embodiment of the present application, the device is used to implement the above-mentioned network function status backup method, and the device may include:
[0141] The first module is used to obtain multiple network functions to be backed up and computing tasks to be processed;
[0142] The second module is used to obtain multiple edge nodes in the edge network;
[0143] The third module is used to initialize nodes of the edge network through the node clustering system, and determine each cluster initialization center from multiple edge nodes using weighted probability distribution;
[0144] The fourth module is used to perform clustering in the edge network according to each cluster initialization center through the node clustering system to obtain edge node clusters corresponding to each cluster initialization center;
[0145] The fifth module is used to update the centroid of each edge node cluster through the node clustering system, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine several edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster;
[0146] The sixth module is used to generate an optimal network function service chain based on the computing power tasks to be processed, each functional service node and each network function to be backed up by using the computing power quantization model and the virtual mapping system, and determine the target service node corresponding to each network function to be backed up from multiple functional service nodes according to the optimal network function service chain;
[0147] The seventh module is used to back up the network function status of each network function to be backed up according to a preset backup time interval using each target backup service node corresponding to the target service node, wherein the target service node is a backup service node in the edge node cluster corresponding to the target service node.
[0148] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0149] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above network function status backup method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.
[0150] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0151] See also Fig.10 , Fig.10 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:
[0152] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;
[0153] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the network function status backup method of the embodiment of the present application;
[0154] Input / output interface 903, used to implement information input and output;
[0155] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);
[0156] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);
[0157] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .
[0158] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned network function status backup method is implemented.
[0159] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0160] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0161] The embodiments of the present application provide a network function status backup method, device, electronic device and storage medium, which determine each cluster initialization center through a node clustering system, cluster in the edge network according to each cluster initialization center, obtain the edge node cluster corresponding to each cluster initialization center, update the centroid of each edge node cluster, determine the function service node and several backup service nodes corresponding to each edge node cluster, use the computing power quantization model and virtual mapping system to generate the optimal network function service chain according to the computing power task to be processed, each function service node and each network function to be backed up, and determine the target service node corresponding to each network function to be backed up from multiple function service nodes according to the optimal network function service chain. For each network function to be backed up, use each target backup service node corresponding to the target service node to perform network function status backup on the network function to be backed up according to a preset backup time interval. The embodiments of the present application can realize network function status backup, efficiently solve the problem of network function status backup, improve the efficiency of network function status backup, and reduce computing power calculation overhead.
[0162] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.
[0163] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0164] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.
[0165] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0166] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.
Claims
1. A network function status backup method, It is characterized in that The method comprises the following steps: Obtain multiple network functions to be backed up and computing tasks to be processed; Get multiple edge nodes in the edge network; Initializing nodes of the edge network by using a node clustering system, and determining each cluster initialization center from a plurality of edge nodes by using a weighted probability distribution; Clustering is performed in the edge network according to each of the cluster initialization centers by the node clustering system to obtain edge node clusters corresponding to each of the cluster initialization centers; The node clustering system is used to update the centroid of each edge node cluster, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine a number of edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster; Using a computing power quantification model and a virtual mapping system, an optimal network function service chain is generated according to the computing power tasks to be processed, each of the functional service nodes and each of the network functions to be backed up, and according to the optimal network function service chain, a target service node corresponding to each of the network functions to be backed up is determined from a plurality of the functional service nodes; For each network function to be backed up, the target backup service nodes corresponding to the target service node are used to back up the network function status of the network function to be backed up according to a preset backup time interval, wherein the target service node is the backup service node in the edge node cluster corresponding to the target service node.
2. The network function status backup method according to claim 1, It is characterized in that The node initialization of the edge network by the node clustering system and the determination of each cluster initialization center from the plurality of edge nodes by weighted probability distribution specifically include: Randomly selecting an initial node center from a plurality of edge nodes through the node clustering system, obtaining the initial node center as the current node center, obtaining the first round as the current iteration round number, and determining the current node center as the cluster initialization center; Calculate the distance between each of the edge nodes except the cluster initialization center in the edge network and the current node center, and determine the next node center from the remaining edge nodes using weighted probability distribution according to the distance between each of the edge nodes and the current node center; When the current iteration round number is less than the number corresponding to the network functions to be backed up, the next node center is determined as the cluster initialization center, the next node center is obtained as the current node center, the current iteration round number is incremented to obtain the next iteration round number, the next iteration round number is obtained as the current iteration round number, and then the distance between each edge node except the cluster initialization center and the current node center in the edge network is calculated, and the next node center is determined from the remaining edge nodes using weighted probability distribution according to the distance between each edge node and the current node center, until the current iteration round number is equal to the number corresponding to the network functions to be backed up; When the current iteration round number is equal to the number corresponding to the network functions to be backed up, the node initialization of the edge network through the node clustering system is stopped.
3. The network function status backup method according to claim 1, It is characterized in that The node clustering system performs clustering in the edge network according to each cluster initialization center to obtain edge node clusters corresponding to each cluster initialization center, specifically including: Calculating the distance between each edge node to be clustered and each of the cluster initialization centers, wherein the edge nodes to be clustered are the remaining edge nodes in the edge network except for the plurality of cluster initialization centers; For each edge node to be clustered, according to the distance between the edge node to be clustered and each cluster initialization center, determine the target cluster initialization center corresponding to the edge node to be clustered from the multiple cluster initialization centers, and add the edge node to be clustered to the edge node cluster corresponding to the target cluster initialization center, wherein, among the multiple cluster initialization centers, the target cluster initialization center is the closest to the edge node to be clustered.
4. The network function status backup method according to claim 1, It is characterized in that The node clustering system is used to update the centroid of each edge node cluster, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine a number of edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster, specifically including: Get the termination condition of iterative correction; For each of the edge node clusters, the cluster initialization center corresponding to the edge node cluster is iteratively corrected by the node clustering system until the corrected cluster initialization center meets the iterative correction termination condition, the corrected cluster initialization center is determined as the cluster center corresponding to the edge node cluster, and several edge nodes in the edge node cluster except the cluster center are determined as the backup service nodes corresponding to the edge node cluster.
5. The network function status backup method according to claim 1, It is characterized in that The use of the computing power quantization model and the virtual mapping system, generating an optimal network function service chain according to the computing power tasks to be processed, each of the functional service nodes and each of the network functions to be backed up, and determining a target service node corresponding to each of the network functions to be backed up from a plurality of the functional service nodes according to the optimal network function service chain, specifically includes: Construct the computing power quantification model and obtain the computing power resource information corresponding to each of the functional service nodes; Determine each optimal service node from a plurality of the functional service nodes using the computing power quantification model according to the computing power tasks to be processed and the computing power resource information corresponding to each of the functional service nodes; Determine, by the virtual mapping system, the optimal node arrangement order corresponding to the plurality of the optimal service nodes according to the computing power tasks to be processed and the network functions to be backed up, and generate the optimal network function service chain according to the optimal node arrangement order and the optimal service nodes; Determine multiple function node allocation schemes according to each of the network functions to be backed up and each of the optimal service nodes; Allocate each of the network functions to be backed up to each of the optimal service nodes in the optimal network function service chain according to each of the functional node allocation schemes through the virtual mapping system, perform virtual network mapping, and calculate the virtual network mapping operation cost corresponding to each of the functional node allocation schemes; According to the virtual network mapping operation cost corresponding to each of the function node allocation schemes, an optimal function node allocation scheme is determined from multiple function node allocation schemes, and according to the optimal function node allocation scheme, the target service node corresponding to each of the network functions to be backed up is determined.
6. The network function status backup method according to claim 5, It is characterized in that The computational power quantization model is expressed by the following formula: Among them, C br is the computing power requirement of the computing task to be processed, n is the number of functional service nodes that process computing tasks through logical computing capabilities, and a i The weight of the logical computing capability of the i-th functional service node is mapped to f(a i ) is the mapping function of the logical computing capability provided by the i-th functional service node, q 1i is the redundant logical computing capacity of the i-th functional service node, m is the number of functional service nodes that process computing tasks through parallel computing capacity, and b j The weight of the parallel computing capability of the jth function service node is mapped to f(b j ) is the mapping function of the parallel computing capability provided by the jth functional service node, q 2j is the redundant parallel computing capacity of the jth functional service node, p is the number of functional service nodes that process computing tasks through neural network acceleration capabilities, and y k The weight of the neural network acceleration capability mapping of the kth function service node, f(y k ) is the mapping function of the neural network acceleration capability provided by the kth function service node, q 3k Redundant neural network acceleration capabilities for the k-th function service node.
7. The network function status backup method according to claim 5, It is characterized in that The virtual mapping system allocates each of the network functions to be backed up to each of the optimal service nodes in the optimal network function service chain according to each of the functional node allocation schemes, performs virtual network mapping, and calculates the virtual network mapping operation cost corresponding to each of the functional node allocation schemes, specifically including: For each of the function node allocation schemes, determine the optimal service node corresponding to each of the network functions to be backed up in the optimal network function service chain according to the function node allocation scheme, perform virtual network mapping, map the computing power tasks to be processed to the optimal network function service chain, and calculate the virtual network mapping operation cost corresponding to the function node allocation scheme.
8. A network function status backup device, It is characterized in that The device comprises: The first module is used to obtain multiple network functions to be backed up and computing tasks to be processed; The second module is used to obtain multiple edge nodes in the edge network; The third module is used to initialize nodes of the edge network through a node clustering system, and determine each cluster initialization center from a plurality of edge nodes using a weighted probability distribution; A fourth module is used to perform clustering in the edge network according to each cluster initialization center through the node clustering system to obtain edge node clusters corresponding to each cluster initialization center; The fifth module is used to update the centroid of each edge node cluster through the node clustering system, determine the cluster center corresponding to each edge node cluster, determine each cluster center as the corresponding functional service node, and determine several edge nodes in each edge node cluster except the cluster center as the backup service nodes corresponding to each edge node cluster; The sixth module is used to generate an optimal network function service chain based on the computing power tasks to be processed, the functional service nodes and the network functions to be backed up by using the computing power quantization model and the virtual mapping system, and determine the target service node corresponding to each network function to be backed up from the multiple functional service nodes according to the optimal network function service chain; The seventh module is used to back up the network function status of each network function to be backed up according to a preset backup time interval using each target backup service node corresponding to the target service node, wherein the target service node is the backup service node in the edge node cluster corresponding to the target service node.
9. An electronic device, It is characterized in that The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the network function status backup method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium storing a computer program. It is characterized in that When the computer program is executed by a processor, the network function status backup method according to any one of claims 1 to 7 is implemented.
