A task intent driven networking service dynamic migration method

Abstract

The application discloses a task intention driven networking service dynamic migration method, comprising the following steps: S101, obtaining intention triples contained in a task issued from a control center; S102, making a decision on whether to perform service migration according to the obtained intention triples and a current network environment; and S103, finding an optimal migration node according to a service migration algorithm and in combination with the description of the requirements for various resources in the intention triples. The application provides continuous and stable networking services for specific tasks by establishing the description of the intention triples for the task, making the service migration decision and selecting the node around the intention requirements. Meanwhile, the optimal migration node is determined in combination with the resource requirements and the resource utilization balance of each node in the service migration algorithm, so that the resource utilization rate in the network is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to a task intent-driven dynamic migration method for networking services. Background Art

[0002] In recent years, with the rapid development of new-generation information technology, including smart cities, intelligent transportation, and other Internet of Things (IoT) applications, as well as new service models such as mobile services, the rapid growth of network traffic has led to higher requirements for data center transmission latency. Simultaneously, the number of connected devices and the resulting amount of data are also growing exponentially, posing significant challenges to traditional networks and cloud computing. These challenges have left cloud computing and networks facing challenges of "poor transmission, inefficient computing, and insufficient storage." This has driven computing to move from the cloud to the edge, closer to the data source, creating distributed computing resources within the network. This shifts the traditional single, centralized model of data centers, enabling large amounts of data to be processed at the edge. This enables network structures to support latency-sensitive and computationally intensive services, effectively reducing the resource consumption of services on the core backbone network and significantly improving wireless network utilization.

[0003] At the same time, new technologies also bring new problems. Because the resources of each node in the network are limited, when providing corresponding services for different computing tasks, multiple different services may be running simultaneously on the same node to provide services for related tasks. As the number of tasks increases, the traffic and resource usage of certain nodes will increase dramatically, and the service quality cannot be guaranteed. To obtain and guarantee high-quality services, service migration is necessary to solve this problem. The process of service migration is to move programs, data, and execution status from one node to another suitable node, thereby alleviating the computing pressure on the original node, improving resource utilization, and ensuring service quality.

[0004] Prior art 1: A method and system for dynamically allocating service network nodes. This technology includes: receiving access requests from user clients, allocating service network nodes based on the access requests, and accessing a behavior monitoring module for the allocated service network nodes; the behavior monitoring module collects behavior information data periodically at preset time periods, and sends the collected behavior information data to the information storage module; and determines whether to start service migration based on the load situation of the load data information. Although this technology realizes the dynamic allocation of services between various network nodes based on load information, it passively triggers service migration during service operation, resulting in long response time and recovery time for service migration, and it is difficult to dynamically migrate networking services for specific task intentions.

[0005] Prior art two: a multi-instance microservice migration method for mobile edge computing. The technology includes: determining the migration order of the microservices to be migrated according to the storage resource large priority migration strategy, then selecting the server node that meets the resource constraint and has the minimum migration delay and communication delay as the target migration node for each microservice according to the migration order, and migrating in turn, and finally adjusting the nodes introducing additional delay to eliminate the additional delay. Although this technology effectively reduces the delay in the service migration process, it is difficult to ensure the running quality of the service on the new node after the service migration only by taking the delay as the optimization target, and the resource utilization of each node in the network is not considered.

[0006] The existing service migration trigger condition is mainly passive triggering, and the conditions considered in the migration decision are single, resulting in long response time and recovery time of service migration; for the calculation of the migration node, the shortest communication distance between nodes is often considered as the optimal strategy, and then the delay and energy consumption are optimized, resulting in low resource utilization and slow convergence speed; the decentralized management system hinders the flexible scheduling and cannot provide protection for specific tasks. SUMMARY

[0007] In order to overcome the defects of the above prior art, the purpose of the present application is to provide a task intention driven networking service dynamic migration method, which evaluates the network state and resources of each node, and dynamically migrates the networking service required by the task according to the decision rule prepared in advance; the resource utilization of each node is improved through service migration. The present application obtains intention elements from tasks, makes service migration decision and node selection around intention demand, which is convenient for providing continuous and stable networking service for specific tasks; at the same time, the optimal migration node is determined by combining resource demand and resource utilization balance of each node in the service migration algorithm, which is beneficial to improve the resource utilization of each node.

[0008] In order to achieve the above purpose, the technical scheme adopted by the present application is:

[0009] A task intention driven networking service dynamic migration method, comprising the following steps:

[0010] S101: extracting key elements from the task issued by the management and control center to generate intention tuples; Figure 3

[0011] S102: making a decision on whether to migrate the service according to the generated intention tuples combined with the current network environment; Figure 3

[0012] S103: finding the optimal migration node according to the service migration algorithm combined with the description of resource demand in the intention tuples. Figure 3

[0013] ​​​In S101, the intention elements contained in the task issued by the control center are obtained, specifically:

[0014] The intention elements extracted from the tasks issued by the control center contain constraint information, which is used for the subsequent migration decision. Different tasks have different requirements for node resources and network environment. The intention of the network service migration task is determined according to the intention. Figure 3 The form of the tuple is described as:

[0015] <region> <resources> <attribute>

[0016] in, <region>to describe the task identity, i.e. the specific service type; <resources>Describe the resources required for the task, including computing, storage, forwarding, and security resources; <attribute>Describe the task's requirements for the network environment, including latency and load.

[0017] The specific process of the decision on service migration in S102 is as follows:

[0018] In order to obtain the specific task Figure 3 Based on the tuple, if the task attributes require low latency and low load, the current network environment is combined to determine whether to perform service migration. The judgment criteria are as follows:

[0019]

[0020] In the table, in terms of service load: L indicates low load, M indicates medium load, and H indicates high load; in terms of latency: LD indicates low latency, MD indicates medium latency, and HD indicates high latency (evaluation criteria are as follows); F indicates service migration, and Y indicates no migration is required.

[0021] The decision on service migration requires evaluating the latency and load of each node in the current network, specifically:

[0022] Based on pre-defined decision rules, the current network environment needs to be evaluated. By quantifying network latency and load, a description of the current network latency and load is obtained to make a service migration decision.

[0023] Latency quantification: The node's latency reflects the current node's network connection status. The latency between the current node and its adjacent routing nodes is quantified. This is measured by extending the timestamp field in broadcast messages and broadcasting them regularly.

[0024] Load quantification: Network load is another important indicator for measuring the current network status. The data passing through a node per unit time includes both data sent by the node as a source node and data forwarded by the node as a forwarding node. The amount of data passing through the node per unit time is used as a representation of the node's business load.

[0025] Each node in the network periodically exchanges its own status information with neighboring nodes to evaluate the current network environment. The specific process is as follows:

[0026] First, the node evaluates the network environment periodically. When a node reaches the detection period, it detects its own network status parameters, including the delay DE from the node to the neighboring node and the amount of data passing through the node per unit time DA.

[0027] If the node reaches the period of sending status information, it combines the status parameters detected in the previous N detection cycles, averages the above detected network status parameters, and calculates the node's delay parameter and load parameter;

[0028] A node sends status information to interact with other nodes. The TTL value of this status information is 1, and only the neighboring nodes of this node can receive this information. This status information contains fields such as the sending node identifier, sequence number, network load parameters of the sending node, and delay parameters of the sending node. The purpose of the status information sequence number is to distinguish between new and old status information.

[0029] The node receives the neighbor status information, extracts the delay and load parameters, updates the node's status information, and calculates the average delay and average load of the network in the current sensing cycle;

[0030] The node weights the state information calculated over two consecutive cycles to obtain the final network state information. Based on the results, the current network latency and load are described as L, M, and H, providing data support for migration decisions. The judgment criteria are as follows: For latency, if the latency is 1-60ms, the current latency is low (LD); if the latency is 61-100ms, the current latency is poor (MD); and if the latency is 100ms-200ms, the current latency is very poor (HD). For load, if the latency is less than 30%, it is L; if it is 31%-60%, it is M; and if it exceeds 61%, it is H.

[0031] In the S103, according to the intention Figure 3 Find the optimal migration node for each resource requirement in the tuple, specifically:

[0032] After making a service migration decision, if the task needs to be migrated, it is necessary to follow the service migration algorithm. The algorithm process is as follows: Figure 3 The description of resource requirements in the tuple is used to find the optimal migration node;

[0033] First, the resource utilization of each node (Current resource usage divided by total resource usage) is calculated. If the resource utilization of a certain type of node exceeds the set threshold value, the node is excluded, and finally several candidate nodes that meet the requirements are obtained;

[0034] Furthermore, since nodes have computing and storage resources, the standard deviation of resource utilization is defined to indicate the balance of resource utilization of each node. The smaller this indicator is, the more reasonable the resource utilization of the node is. The standard deviation of resource utilization of each node is:

[0035]

[0036] where λ = 1 / K, Indicates the utilization rate of the kth resource of the node (the current resource occupancy value divided by the total resource amount), Represents the average resource utilization on node i.

[0037] The resource balance at a node is measured by the ratio of the average utilization of each resource at the node to the maximum utilization of each resource:

[0038]

[0039] in Indicates the maximum resource utilization among multiple resource types. The closer the node's resource balance value is to 1, the better the utilization control of a particular resource type is.

[0040] The formula for calculating the resource utilization balance of a node is:

[0041]

[0042] Furthermore, the resource utilization balance after migrating the service to a node is calculated based on the service's resource demand and the node's available resources. The improvement in resource utilization balance after service migration is measured by the ratio of the change in resource utilization balance before and after service migration to the resource utilization balance before service migration.

[0043] Service migration benefit analysis: Before a service migration, metrics are needed to evaluate the migration's effectiveness. The benefits of the migration must be defined to determine whether the migration costs meet expectations. The benefits of service migration are primarily reflected in the improvement in resource utilization balance before and after the migration.

[0044]

[0045] If the value is greater than zero, it means that migrating the service to the node will improve the resource utilization balance of the node. Otherwise, the opposite is true. The node with a larger change range is more likely to be selected as the node for service migration in the candidate node set;

[0046] Service migration cost analysis: The cost of service migration mainly comes from the service data and node status data transmitted during the migration process. If the number of hops from the source node to the target node is n, the service migration cost is the sum of the costs of each path segment. The service migration cost calculation formula is:

[0047] cost(i)=cost(l i )×n

[0048] Where cost(l i ) is the weight of each hop in the path.

[0049] Finally, by weighting the migration benefits and migration costs, we get the overall evaluation value for service migration, and the node with the largest value is selected as the best target node for migration. The overall evaluation function for service migration is as follows:

[0050] G(i)=Benefit(i)×α-cost(i)×β

[0051] α and β are correction coefficients for latency and cost, respectively, which dynamically adjust demand for different services. α is weighted more heavily for services that prioritize revenue, while β is weighted more heavily for services that prioritize cost.

[0052] Based on the above description, the algorithm flow is as follows:

[0053]

[0054]

[0055] A task intent-driven dynamic migration system for networking services, comprising:

[0056] a memory for storing instructions executable by the processor;

[0057] A processor is configured to execute the instructions to implement the above method.

[0058] A computer-readable medium stores computer program codes, which implement the above-mentioned method when executed by a processor.

[0059] Beneficial effects of the present invention:

[0060] The present invention discloses a task intent-driven dynamic migration method for networking services. The task intent-driven dynamic migration method for networking services includes: extracting intent elements contained in tasks issued by a control center; making a decision on whether to perform service migration based on the obtained intent elements in combination with the current network environment; if service migration is required, finding the optimal migration node based on the description of various resource requirements in the intent elements according to the service migration algorithm. The present invention facilitates providing continuous and stable networking services for specific tasks by establishing an intent element description for the task and making service migration decisions and node selections based on the intent requirements; at the same time, the service migration algorithm combines resource requirements and the resource utilization balance of each node to determine the optimal migration node, which is beneficial to improving the resource utilization of each node. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a flow chart of the method for dynamic migration of networking services driven by the task intent of the present invention.

[0062] Figure 2 For the purpose of the present invention Figure 3 Tuple diagram.

[0063] Figure 3 This is a schematic diagram of the service migration decision flow chart.

[0064] Figure 4 This is a diagram of the service migration process. DETAILED DESCRIPTION

[0065] The present invention will be further described in detail below with reference to the accompanying drawings.

[0066] like Figure 1 The flowchart of the task intent-driven dynamic migration method of networking services provided by the present invention includes the following steps:

[0067] S101: Extract key elements from tasks issued by the control center to generate intentions Figure 3 Tuple;

[0068] S102: According to the obtained Figure 3 The tuple is combined with the current network environment to determine whether to perform service migration;

[0069] S103: Based on the service migration algorithm, Figure 3 The description of various resource requirements in the tuple is used to find the optimal migration node.

[0070] In S101, the intention elements contained in the task issued by the control center are obtained, specifically:

[0071] The tasks issued by the control center, on the one hand, refer to the processing of certain specific data packets during their forwarding process to ensure the security and integrity of the data packets during transmission; on the other hand, if a node router is damaged during network operation, the status of the node needs to be migrated to other nodes to ensure the continuous provision of networking services.

[0072] The intention elements extracted from the task mainly contain some constraint information, which is used for the subsequent migration decision. Different tasks have different requirements for node resources and network environment. Therefore, the intention of the network service migration task can be determined according to the intention. Figure 3 It is described in the form of a tuple, such as Figure 2 As shown:

[0073] <region> <resources> <attribute>

[0074] in, <region>Describe the task identifier, i.e. the specific service type, etc. <resources>Describe the resources required for the task, including computing, storage, forwarding, and security resources; <attribute>Describe the task's requirements for the network environment, including latency, load, etc.

[0075] meaning Figure 3 The basic description of the tuple is shown in the following table:

[0076] Tuple Basic Description Region Signature verification / Message cache / etc. Resources Calculate / store / forward / security / etc. Attribute Delay / load / etc.

[0077] The specific process of the service migration decision in S102 is as follows:

[0078] In order to obtain the specific task Figure 3 Based on the tuple, if the task has high requirements for task attributes (delay, load), it is necessary to make a decision on whether to migrate the service in combination with the current network environment. The decision process is as follows: Figure 4 For tasks with high latency and load requirements, the judgment criteria are as follows:

[0079]

[0080] In the table, service load is represented by L, M, and H, respectively. Latency is represented by LD, MD, and HD, respectively, indicating low latency, medium latency, and high latency (evaluation criteria are as follows). F indicates service migration is required, while Y indicates no migration is required.

[0081] Based on pre-defined decision rules, the current network environment needs to be evaluated. By quantifying network latency and network load, a description of the current network latency and network load is obtained, which is used to make service migration decisions.

[0082] Latency quantification: The latency of a node reflects the network connection status of the current node. When the network condition is poor, the latency is often higher. We consider quantifying the latency between the current node and its adjacent routing nodes. We test the latency by extending the timestamp field in the broadcast message and broadcasting it regularly.

[0083] Load quantification: Network load is another important indicator for measuring the current network status. The data passing through a node per unit time includes both data sent by the node as a source node and data forwarded by the node as a forwarding node. Therefore, the amount of data passing through the node per unit time is used to represent the node's business load.

[0084] Each node in the network periodically exchanges its own status information with neighboring nodes to evaluate the current network environment. The specific process is as follows:

[0085] First, the node evaluates the network environment periodically. When a node reaches the detection period, it detects its own network status parameters, including the delay DE from the node to the neighboring node and the amount of data passing through the node per unit time DA.

[0086] Furthermore, if the node reaches the period for sending status information, the state parameters detected in the previous N detection periods are combined, the network state parameters detected above are averaged, and the node's delay parameters and load parameters are calculated;

[0087] Furthermore, the node sends status information to interact with other nodes. The TTL value of this status information is 1, and only the neighboring nodes of this node can receive this information. This status information contains fields such as the sending node identifier, sequence number, network load parameters of the sending node, and delay parameters of the sending node. The purpose of the status information sequence number is to distinguish the new and old status information.

[0088] Furthermore, the node receives neighbor status information, extracts delay and load parameters, updates the node's status information table, and calculates the average delay and average load of the network in the current sensing cycle;

[0089] Furthermore, the node weights the state information calculated over two consecutive cycles to obtain the final network state information. Based on the results, the current network latency and load are described using L, M, and H metrics, providing data support for migration decisions. The judgment criteria are as follows: For latency, if the latency is 1-60ms, the current latency is low (LD); if the latency is 61-100ms, the current latency is poor (MD); and if the latency is 100ms-200ms, the current latency is very poor (HD). For load, if the latency is less than 30%, it is L; if it is 31%-60%, it is M; and if it exceeds 61%, it is H.

[0090] In the S103, according to the intention Figure 3 Find the optimal migration node for each resource requirement in the tuple, specifically:

[0091] After making a service migration decision, if the task needs to be migrated, it is necessary to follow the service migration algorithm. The algorithm process is as follows: Figure 3 The description of resource requirements in the tuple is used to find the optimal migration node;

[0092] Since nodes have computing, storage and other resources, the standard deviation of resource utilization is defined to indicate the balance of resource utilization of each node. The smaller this indicator is, the more reasonable the resource utilization of the node is. The standard deviation of resource utilization of each node is:

[0093]

[0094] Where λ=1 / K, Indicates the utilization rate of the kth resource of the node (the current resource occupancy value divided by the total resource amount), Represents the average utilization of resources on node i.

[0095] The resource balance at a node is measured by the ratio of the average utilization of each resource at the node to the maximum utilization of each resource:

[0096]

[0097] in Indicates the maximum resource utilization among multiple resource types. The closer the node's resource balance value is to 1, the better the utilization control of a particular resource type is.

[0098] Furthermore, the formula for calculating the resource utilization balance of a node is:

[0099]

[0100] At the same time, the resource utilization balance after the service is migrated to a certain node is calculated based on the service's demand for resources and the available resources of the node. The improvement in resource utilization balance caused by service migration is measured by the ratio of the change in resource utilization balance before and after service migration to the resource utilization balance before service migration.

[0101] Service Migration Benefit Analysis: Before migrating services, metrics are needed to evaluate the effectiveness of the migration. The benefits of the migration must be defined to determine whether the migration costs meet expectations. The benefits of service migration are primarily reflected in the improvement in resource utilization balance before and after the migration.

[0102]

[0103] If the value is greater than zero, it means that migrating the service to the node will improve the resource utilization balance of the node. Otherwise, the opposite is true. The node with a larger change range is more likely to be selected as the node for service migration in the candidate node set;

[0104] Service migration cost analysis: The cost of service migration mainly comes from the service data and node status data transmitted during the migration process. If the number of hops from the source node to the target node is n, the service migration cost is the sum of the costs of each path segment. The service migration cost calculation formula is:

[0105] cost(i)=cost(l i )×n

[0106] Where cost(l i ) is the weight of each hop in the path.

[0107] Finally, by weighting the migration benefits and migration costs, we get the overall evaluation value for service migration, and the node with the largest value is selected as the best target node for migration. The overall evaluation function for service migration is as follows:

[0108] G(i)=Benefit(i)×α-cost(i)×β

[0109] α and β are correction coefficients for latency and cost, respectively, which dynamically adjust demand for different services. α is weighted more heavily for services that prioritize revenue, while β is weighted more heavily for services that prioritize cost.

[0110] like Figure 4 FIG2 is a flow chart of service migration in an application scenario of the present invention. For a specific data packet, in order to ensure its security and integrity during transmission, certain specific nodes in its forwarding process need to provide it with verification services.

[0111] In combination with the above technical solutions, the present invention obtains intent elements from the task, makes service migration decisions and node selections based on the intent requirements, and facilitates the provision of continuous and stable networking services for specific tasks. At the same time, the service migration algorithm combines resource requirements and the resource utilization balance of each node to determine the optimal migration node, effectively improving resource utilization in the network.< / attribute> < / resources> < / region> < / attribute> < / resources> < / region> < / attribute> < / resources> < / region> < / attribute> < / resources> < / region>

Claims

1. A task intent driven dynamic migration method for networking services, characterized in that: The following steps are included: S101: Tasks issued from the control center process specific data packets during forwarding to ensure the security and integrity of data packets during transmission. Furthermore, if a node router fails during network operation, the node's state needs to be migrated to another node to ensure continuous network service. The intent triplet contained in the task issued from the control center is obtained. S102: Based on the generated intent triplet for a specific task, if the task has high requirements for task attributes, a decision is made based on the current network environment whether to perform service migration. S103: After making a service migration decision, if the task needs to be migrated, the optimal migration node is found based on the service migration algorithm and the description of various resource requirements in the task intent triplet. In S101, the intention elements contained in the task issued by the control center are obtained, specifically: The intent elements extracted from the tasks issued by the control center contain constraint information, which is used for subsequent migration decisions. Different tasks have different requirements for node resources and network environments. The intent of the networking service migration task is described in the form of an intent triplet: <region> <resources> <attribute>< / attribute> < / resources> < / region> in, <region>Describe the task identifier, that is, the specific service type; <resources>Describe the resources required for the task, including computing, storage, forwarding, and security resources; <attribute> Describe the task's requirements for the network environment, including latency and load;< / attribute> < / resources> < / region> The specific process of the decision on service migration in S102 is as follows: Based on the triplet of intent for a specific task, if the task attributes require low latency and low load, a decision is made on whether to perform service migration based on the current network environment. The judgment criteria are as follows: In the table, in terms of service load: L indicates low load, M indicates medium load, and H indicates high load. In terms of latency: LD indicates low latency, MD indicates medium latency, and HD indicates high latency. F indicates service migration, and Y indicates no migration is required. The decision on service migration requires evaluating the latency and load of each node in the current network, specifically: Based on pre-defined decision rules, the current network environment needs to be evaluated. By quantifying network latency and load, a description of the current network latency and load is obtained to make a service migration decision. Latency quantification: The node's latency reflects the current node's network connection status. The latency between the current node and its adjacent routing nodes is quantified. This is measured by extending the timestamp field in broadcast messages and broadcasting them regularly. Load quantification: Network load is another important indicator for measuring the current network status. The data passing through a node per unit time includes both data sent by the node as a source node and data forwarded by the node as a forwarding node. The amount of data passing through the node per unit time is used to represent the node's service load. In S103, the optimal migration node is found according to the requirements for various resources in the intention triple, specifically: After making a service migration decision, if the task needs to be migrated, the service migration algorithm is used. The algorithm process is as follows: combining the description of various resource requirements in the task intent triplet to find the optimal migration node; First, the resource utilization of each node Calculation is performed. If the resource utilization of a certain type of node exceeds the set threshold value, the node is excluded, and finally several candidate nodes that meet the requirements are obtained; The standard deviation of resource utilization is defined to indicate the balanced utilization of various types of resources at a node. The smaller this indicator is, the more reasonable the resource utilization of the node is. The standard deviation of resource utilization at each node is: where λ = 1 / K, represents the utilization rate of the kth resource of the node, represents the average utilization of resources on node i; The resource balance at a node is measured by the ratio of the average utilization of each resource at the node to the maximum utilization of each resource: in Indicates the maximum value of resource utilization among multiple types of resources; The formula for calculating the resource utilization balance of a node is:

2. The method for dynamic migration of network services driven by task intent according to claim 1, characterized in that: Each node in the network periodically exchanges its own status information with neighboring nodes to evaluate the current network environment. The specific process is as follows: First, the node evaluates the network environment periodically. When a node reaches the detection period, it detects its own network status parameters, including the delay DE from the node to the neighboring node and the amount of data passing through the node per unit time DA. If the node reaches the period of sending status information, it combines the status parameters detected in the previous N detection cycles, averages the above detected network status parameters, and calculates the node's delay parameter and load parameter; A node sends status information to interact with other nodes. The TTL value of this status information is 1, and only the neighboring nodes of this node can receive this information. This status information contains fields such as the sending node identifier, sequence number, network load parameters of the sending node, and delay parameters of the sending node. The purpose of the status information sequence number is to distinguish between new and old status information. The node receives the neighbor status information, extracts the delay and load parameters, updates the node's status information table, and calculates the average delay and average load of the network in the current sensing cycle; The node weights the status information calculated in two adjacent cycles to obtain the final network status information. Based on the results, the current network delay and load are described as L, M, and H, providing data support for migration decisions. The judgment criteria are as follows: For delay judgment, if the delay is 1-60ms, the current delay is low (LD); if the delay is 61-100ms, the current delay is poor (MD); and if the delay is 100ms-200ms, the current delay is very poor (HD). For load judgment, if it is less than 30%, it is L; if it is 31%-60%, it is M; and if it exceeds 61%, it is H.

3. The method for dynamic migration of network services driven by task intent according to claim 1, characterized in that: Service migration benefit analysis: Before a service migration, metrics are needed to evaluate the migration's effectiveness. The benefits of the migration must be defined to determine whether the migration costs meet expectations. The benefits of service migration are primarily reflected in the improvement in resource utilization balance before and after the migration. If the value is greater than zero, it means that migrating the service to the node will improve the resource utilization balance of the node. Otherwise, the opposite is true. The node with a larger change range is more likely to be selected as the node for service migration in the candidate node set; Service migration cost analysis: The cost of service migration mainly comes from the service data and node status data transmitted during the migration process. If the number of hops from the source node to the target node is n, the service migration cost is the sum of the costs of each path segment. The service migration cost calculation formula is: cost(i)=cost(l i )×n Where cost(l i ) is the weight of each hop in the path; Finally, by weighting the migration benefits and migration costs, we get the overall evaluation value of service migration. The node with the largest value is taken as the best target node for migration. The overall evaluation function of service migration is as follows: G(i)=Benefit(i)×α-cost(i)×β Among them, α and β are correction coefficients for delay and cost, which can dynamically adjust the demand for different services. For services that value benefits, α has a larger weight; for services that value costs, β has a larger weight.

4. A task intent driven networking service dynamic migration system, characterized in that: include: a memory for storing instructions executable by the processor; A processor is configured to execute the instructions to implement the method according to any one of claims 1 to 3.

5. A computer-readable medium, characterized in that The computer program code is stored, and when the computer program code is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images