A network configuration automatic generation and dynamic conflict checking method and system based on service migration

Abstract

This invention relates to the field of digital information transmission technology and discloses a method for automatic generation and dynamic conflict verification of network configuration based on service migration. The method includes: acquiring basic data of service migration resources; characterizing multi-dimensional resource request vectors using a hash and Laplace feature mapping mechanism; constructing a multi-domain resource conflict graph using a hash domain index and graph model mechanism; determining the conflict domain level results using a Hearst exponent dynamic conflict verification mechanism; forming resource reservation results using a weighted independent set resolution and two-phase lock atomic reservation mechanism; and performing pre-simulation verification and transactional distribution using a Monte Carlo shadow deduction mechanism to obtain incremental network configuration distribution results. Because this application achieves dynamic verification and atomic reservation of concurrent conflicts through the hierarchical connection of multi-dimensional resource request vectors, multi-domain resource conflict graphs, conflict domain level results, and resource reservation results, it improves the reliability and consistency of network configuration distribution.

Description

Technical Field

[0001] This invention relates to the field of digital information transmission technology, and in particular to a method and system for automatic generation and dynamic conflict verification of network configuration based on service migration. Background Technology

[0002] Currently, in data centers, campus networks, and carrier bearer networks, service migration is often accompanied by the synchronous update of numerous network device configurations, involving multiple resource adaptation items such as link bandwidth reservation, port queue resource allocation, VLAN / VNI identifier segment division, and access control policy priority adjustment. When multiple service migration tasks are executed concurrently within the same scheduling window, the orchestration system typically generates incremental network configurations for each migration task and checks the legality of the internal configuration parameters of each task through a serialized verification process before distributing the configurations to the target devices one by one. For requests for shared resources such as bandwidth, queues, and identifiers, existing processing paths mostly adopt a first-come, first-served approach or perform pre-checks according to the task submission order. Resource conflict detection is mainly concentrated in the generation stage of a single task and the static comparison stage before distribution. This single-task-based processing method lacks a global resource occupancy view for a set of concurrent tasks. The configuration generation, conflict verification, and resource reservation stages are isolated from each other, making it difficult to perceive the resource competition relationship between multiple tasks while adapting resources.

[0003] For example, in scenarios where multiple service migration tasks are executed concurrently, different migration tasks may simultaneously reserve remaining bandwidth on the same core link, or initiate overlapping requests for the same VLAN / VNI identifier segment, or contend for limited queue resources on the same port or adjacent policy priority ranges. The serial verification mechanism can only detect issues such as inconsistent path parameters with the configuration template, mismatched port attributes, and parameter out-of-bounds errors within a single task. It cannot detect the resource reservation status of other concurrent tasks, let alone discover resource reservation overlaps between concurrent tasks during the configuration generation phase. Since the verification processes of each task are independent, both tasks can determine their status through local conditions during separate checks. However, resource reservation overlaps can occur after actual submission, leading to later-submitted configurations overwriting earlier-submitted configurations, queue resource exceeding limits, identifier segment conflicts, or policy order disorder, thereby causing traffic interruptions, service switching failures, or configuration inconsistencies. Especially when migration task time windows overlap, dependent paths share multiple network edges, and network resource utilization is high, concurrent conflicts are more easily amplified. The lag in static comparison and serial pre-check further exacerbates the risk of configuration overlap and service interruption.

[0004] Therefore, there is an urgent need for a method for automatic generation and dynamic conflict verification of network configuration based on service migration. This method should be able to uniformly characterize resource adaptation items, model conflict relationships, and determine conflict levels among concurrent tasks under dynamic conditions where multiple service migration tasks are executed concurrently, and multiple domain resources such as bandwidth, queues, VLAN / VNI, and policy priorities are applied for simultaneously with overlapping time windows. Furthermore, it should be able to atomically reserve bandwidth, queues, and identification resources during the resource adaptation phase to avoid reservation overlap and configuration overlay between concurrent applications, thereby improving configuration consistency, resource utilization efficiency, and network reliability during service migration. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the purpose of this invention is to propose an automatic network configuration generation and dynamic conflict verification method based on service migration. This method aims to solve the technical problem that in the existing technology, serial verification only detects conflicts within a single task. In particular, when multiple service migration tasks concurrently request bandwidth, queues, VLAN / VNI, and policy priorities, it fails to detect overlapping reservations between tasks, leading to configuration overwriting and service interruption, and thus cannot achieve dynamic conflict verification between concurrent tasks.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for automatic generation and dynamic conflict verification of network configuration based on service migration.

[0007] The method for automatic generation and dynamic conflict verification of network configuration based on service migration includes: Step S10: Obtain basic data of business migration resources. Based on the basic data of business migration resources, perform a multi-dimensional resource request characterization task using a hash and Laplace feature mapping mechanism, and output a multi-dimensional resource request vector. Step S20: Based on the multi-dimensional resource request vector, a concurrent conflict relationship characterization task is performed using a hash domain index and graph model mechanism to output a multi-domain resource conflict graph; Step S30: Based on the multi-domain resource conflict graph, the conflict level determination task is performed using the Hearst exponent dynamic conflict verification mechanism, and the conflict domain level result is output. Step S40: Based on the conflict domain level results, the resource reservation task is executed using weighted independent set resolution and a two-phase locked atomic reservation mechanism, and the resource reservation results are output. Step S50: Based on the resource reservation results, the Monte Carlo shadow simulation mechanism is used to perform the pre-simulation verification and transactional distribution tasks, and the incremental network configuration distribution results are output.

[0008] Preferably, step S10, which involves obtaining basic data on business migration resources, and based on this data, performing a multi-dimensional resource request characterization task using a hash and Laplacian feature mapping mechanism to output a multi-dimensional resource request vector, specifically includes: Step S101: Obtain the set of concurrent migration tasks and the real-time network resource view within the current scheduling period, and encapsulate the set of concurrent migration tasks and the real-time network resource view into the basic data of the service migration resources. Each migration task in the set of concurrent migration tasks includes a task identifier, service type, task priority, planned execution time window, dependency path, and traffic parameters. The real-time network resource view includes the remaining bandwidth capacity of the link, the upper limit of the number of available queues on the port, the upper limit of the queue depth that the shared cache can accommodate, the idle segment of the identifier pool, and the range of policy priority values. Step S102: Based on the basic data of the business migration resources, the planned execution time window is discretized according to the migration stage, and a stage time slot matrix is ​​constructed according to the preset resource occupancy coefficient, traffic burst coefficient, basic application volume and business confidence coefficient of each migration stage to form stage resource description data. Step S103: Based on the stage resource description data, hash the sequence of device interfaces traversed by the migration task to obtain the first-level path fingerprint, perform Laplace eigenmap on the adjacency matrix of the subgraph of the migration task on the path coupling graph to obtain the second-level path fingerprint, and organize resource request data according to task identifier, resource domain, resource key, request quantity, planned execution time window, task priority, first-level path fingerprint, second-level path fingerprint, traffic burst coefficient, service confidence coefficient, and request identifier. Associate the resource request data with the current occupied quantity of the corresponding resource domain, the occupied quantity of resources that have passed pre-verification but have not yet been submitted, the available quantity, and the total quantity of resources to obtain the multi-dimensional resource request vector.

[0009] Preferably, step S20, which involves using a hash-based domain indexing and graph model mechanism to perform a concurrent conflict relationship characterization task based on the multi-dimensional resource request vector and outputting a multi-domain resource conflict graph, specifically includes: Step S201: Perform hash domain indexing on the multidimensional resource request vector according to resource domain, resource key and time slot, and map the non-zero resource requests of each migration stage to the corresponding time slot to form domain time slot index data; Step S202: Based on the domain time slot index data, extract the task dependency path fingerprint, determine the time window overlap rate by the ratio of the number of overlapping slots corresponding to the time slots of the two task requests to the number of non-zero slots of the lesser one, and establish a conflict edge when the two task requests meet at least one of the following conditions: the resource key is the same, they belong to the same shared resource group, the dependency path has a shared edge, or the strategy matching domain has an inclusion relationship, and the time window overlap rate is greater than zero. Step S203: Based on the initial resource conflict graph, according to the preset resource domain weights of each resource domain, the product of resource coverage overlap, time window overlap rate and resource scarcity coefficient is weighted and summarized. The resource scarcity coefficient is determined according to the ratio of the current occupied amount, the occupied amount of resources that have passed pre-verification but have not yet been submitted, and the predicted new occupied amount to the total resource amount of the resource domain. The conflict edges are updated in combination with the joint occupied potential of dependent paths to obtain the multi-domain resource conflict graph.

[0010] Preferably, in step S203, the calculation step of the joint occupancy potential of the dependency paths includes determining the shared edge set of the task dependency paths based on the task dependency path fingerprint, and calculating the joint occupancy potential of the dependency paths according to the following formula:

[0011] in, For the dependent path to jointly occupy potential energy, Shared edge set for task-dependent paths For shared edge indexes, For bandwidth domain weights, For time window overlap rate, For task indexing, For the task Dependency path, For the task Traffic burst coefficient, For the task In shared edge Basic bandwidth requirements, For shared edges capacity, For shared edges The resource scarcity coefficient; the bandwidth domain weight, the time window overlap rate, the resource scarcity coefficient and the joint occupancy potential of the dependent path are all dimensionless quantities, and the basic bandwidth requirement and the capacity of the shared edge use the same bandwidth unit; the joint occupancy potential of the dependent path is merged with the multi-domain conflict potential of the corresponding conflict edge to obtain the updated conflict edge potential.

[0012] Preferably, step S30, which involves using the Hearst exponent dynamic conflict verification mechanism to perform conflict level determination based on the multi-domain resource conflict graph and outputting the conflict domain level result, specifically includes: Step S301: Based on the multi-domain resource conflict graph, extract the remaining bandwidth capacity of the corresponding link, the upper limit of the number of available queues at the port, and the upper limit of the queue depth that the shared cache can accommodate from the multi-dimensional resource request vector. Summarize the concurrent bandwidth demand according to the request quantity and planned execution time window in the multi-dimensional resource request vector, and correct the concurrent bandwidth demand using a Hearst exponent greater than 0.5 and less than 1 and a preset self-similarity penalty coefficient to obtain the equivalent concurrent bandwidth. Calculate the queue number pressure of the number of task request queues relative to the upper limit of the number of available queues at the port and the queue depth pressure of the task request queue depth relative to the upper limit of the queue depth that the shared cache can accommodate. Determine the larger of the two values ​​as the comprehensive queue pressure, and combine it with the equivalent concurrent bandwidth to form the first conflict verification data. Step S302: Based on the first conflict verification data, determine the overlap of the two task application flag segments according to the overlapping interval of the flag segments, and determine the flag conflict potential energy by combining the time window overlap rate and the flag resource scarcity coefficient of the corresponding flag domain in the resource scarcity coefficient; obtain the preset policy priority value range, determine the rule matching domain coverage according to the ratio of the intersection of the two rule matching spaces to the smaller one's matching space, and determine the priority distance according to the difference between the two rule priority values ​​relative to the preset policy priority value range; combine the flag conflict potential energy, the rule matching domain coverage, and the priority distance to form the second conflict verification data. Step S303: Based on the second conflict verification data, determine bandwidth conflict when the equivalent concurrent bandwidth exceeds the product of the remaining bandwidth capacity of the link and the preset link security level; determine queue conflict when the comprehensive queue pressure is greater than 1; determine strong identifier conflict when the identifier conflict potential is greater than zero, the identifier segment overlap exceeds the preset identifier segment overlap threshold, and the time window overlap rate exceeds the preset time window overlap rate threshold; determine policy priority conflict when the rule matching domain coverage exceeds the preset matching domain coverage threshold, the priority distance is less than the preset priority distance threshold, and at least one of the following conditions is met: different rule actions or rules are located in the same access control list chain; summarize the conflict potential of each resource domain, and determine the conflict domain level result according to the preset conflict level threshold.

[0013] Preferably, step S40, based on the conflict domain level result, involves performing a resource reservation task using weighted independent set resolution and a two-phase locked atomic reservation mechanism, and outputting the resource reservation result. This step specifically includes: Step S401: Based on the conflict domain level results, lock the non-conflicting resource requests in the order of the preset partial order relationship between the identifier domain, bandwidth domain, queue domain and policy priority domain, lock the resource domains that have no dependency relationship in parallel, and release all the acquired locks and roll back the reservation table when the reservation of any resource domain fails, forming the initial resource reservation data. Step S402: Based on the initial resource reservation data, construct a weighted independent set optimization problem for conflicting resource requests. Determine the retention weight of resource requests based on task priority, business confidence coefficient, and task duration. Determine the conflict penalty for retaining adjacent resource requests based on conflict edge potential energy. Under the constraint that the request amount in each resource domain does not exceed the corresponding available amount, use greedy pre-selection and local search to determine the retained resource requests, forming conflict resolution data. Step S403: Based on the conflict resolution data, for tasks that have not obtained the original resource application right, candidate alternative resources that simultaneously meet the preset satisfaction threshold, the preset new conflict potential threshold, and the preset switching cost threshold are selected from the available resources belonging to the same resource domain as the original resource application. The alternative resources are determined according to the descending order of satisfaction, the ascending order of new conflict potential, and the ascending order of switching cost. Reserved tokens are generated for the retained resource application and the resource application using the alternative resources. When the simulation is not completed, the renewal is determined according to the change ratio of the current resource scarcity coefficient relative to the resource scarcity coefficient at the initial reservation, thus forming the resource reservation result.

[0014] Preferably, step S50, based on the resource reservation result, involves using a Monte Carlo shadow simulation mechanism to perform pre-simulation verification and transactional deployment tasks, and outputting incremental network configuration deployment results. This step specifically includes: Step S501: Based on the resource reservation results, write the reserved bandwidth resources, identifier resources, queue resources, and policy order resources into the corresponding shadow resource tables. Overlay the shadow incremental configurations of all migration tasks corresponding to the resource reservation results onto the simulation topology to construct a forwarding behavior model. Obtain the set of service flows to be verified corresponding to the resource reservation results, and select a set of key flows from the set of service flows to be verified. This forms shadow projection data; Step S502: Based on the shadow projection data, determine the link congestion aggregation utilization rate according to the ratio of the equivalent concurrent bandwidth to the corresponding link capacity, and calculate the multidimensional reachability loss according to the following formula:

[0015] in, For multidimensional accessibility loss, For the set of key flows, Index for business flow, For business flow The weight, , and To preset the loss weight, For business flow Changes in packet loss rate For business flow The increase in latency, For business flow The preset latency tolerance limit, For business flow The amount of vibration increases. For business flow The preset jitter tolerance limit; the weight of the service flow, the preset loss weight, the packet loss rate change, the multidimensional reachability loss, the ratio of the increase in latency to the preset latency tolerance limit, and the ratio of the increase in jitter to the preset jitter tolerance limit are all dimensionless quantities; the link congestion superposition utilization rate and the multidimensional reachability loss are combined to form a simulation verification index; Step S503: Based on the simulation verification index, when the multidimensional reachability loss exceeds a preset reachability loss threshold or the link congestion superposition utilization exceeds a preset congestion utilization threshold, a secondary resolution or rollback is triggered; otherwise, an incremental configuration difference set is generated. The order of distribution of each configuration operation in the incremental configuration difference set is determined according to the resource dependency relationship. A compensation operation and configuration fingerprint are generated for each configuration operation. A two-phase transaction protocol is used to perform pre-commit and commit to the target device. If any target device fails to commit or times out, a rollback is broadcast. If a device fails to execute after commit, an idempotent rollback is performed according to the compensation operation and the configuration fingerprint to obtain the incremental network configuration distribution result.

[0016] This invention also provides a network configuration automatic generation and dynamic conflict verification system based on service migration, comprising: The multi-dimensional resource request characterization module is used to obtain basic data of business migration resources, and based on the basic data of business migration resources, it performs the multi-dimensional resource request characterization task using a hash and Laplace feature mapping mechanism, and outputs a multi-dimensional resource request vector. The concurrent conflict relationship characterization module is used to perform concurrent conflict relationship characterization tasks based on the multi-dimensional resource request vector, using hash domain indexing and graph model mechanism, and output a multi-domain resource conflict graph; The multi-domain dynamic conflict verification module is used to perform conflict level determination tasks based on the multi-domain resource conflict graph and using the Hearst exponent dynamic conflict verification mechanism, and output the conflict domain level results. The resource reservation and conflict resolution module is used to perform resource reservation tasks based on the conflict domain level results, using weighted independent set resolution and a two-phase locked atomic reservation mechanism, and output resource reservation results. The pre-simulation and transactional deployment module is used to perform pre-simulation verification and transactional deployment tasks based on the resource reservation results, using the Monte Carlo shadow simulation mechanism, and output incremental network configuration deployment results.

[0017] The present invention also provides a network configuration automatic generation and dynamic conflict verification device based on service migration. The network configuration automatic generation and dynamic conflict verification device based on service migration includes: a memory, a processor, and a network configuration automatic generation and dynamic conflict verification program based on service migration stored in the memory and executable on the processor. When the network configuration automatic generation and dynamic conflict verification program based on service migration is executed by the processor, it implements the above method.

[0018] The present invention also provides a computer program product, the computer program product including a network configuration automatic generation and dynamic conflict verification program based on service migration, the network configuration automatic generation and dynamic conflict verification program based on service migration implementing the above method when executed by a processor.

[0019] The beneficial effects of this invention are as follows: By using a hash and Laplace feature mapping mechanism, two-level fingerprint extraction is performed on the migration task path and device interface sequence. The stage time slot matrix is ​​associated with the resource request data, so that the multi-dimensional resource requirements of concurrent tasks are uniformly characterized as an indexable request vector. This allows a global resource occupancy view to be obtained during the resource adaptation stage, reducing the risk of resource reservation overlap and configuration overlay caused by a single task perspective.

[0020] A multi-domain resource conflict graph is constructed using a hash-based domain index and graph model mechanism. The Hearst exponent dynamic conflict verification mechanism is used to determine the conflict level of bandwidth, queue, identifier, and policy priority. This allows the resource competition relationship between concurrent tasks to be explicitly modeled and quantified during the configuration generation phase. This enables the identification of potential conflicts and the differentiation of conflict intensity before resource reservation, avoiding the overwriting of earlier-submitted configurations or the exceeding of queue resource limits by later-submitted configurations.

[0021] By using weighted independent set resolution and two-phase lock atomic reservation mechanism to resolve conflicting resource requests and atomically reserve them, and combining Monte Carlo shadow simulation mechanism to verify the impact of reservation configuration on forwarding behavior in simulation environment, the resource reservation results are pre-verified before formal distribution, thereby reducing the probability of traffic interruption and service switching failure caused by configuration conflicts, and improving configuration consistency and network reliability during service migration. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the first embodiment of a method for automatic generation and dynamic conflict verification of network configuration based on service migration according to the present invention.

[0023] Figure 2 This is a schematic diagram of the stage time slot matrix of the migration task in the first embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of the two-level path fingerprint and shared path edge relationship in the first embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of a multi-domain resource conflict graph in the first embodiment of the present invention.

[0026] Figure 5 This is a schematic diagram illustrating the resource scarcity coefficients of each resource domain in the first embodiment of the present invention.

[0027] Figure 6 This is a schematic diagram illustrating the correction of the Hearst exponent to the equivalent concurrent bandwidth and the comparison of safety levels in the first embodiment of the present invention.

[0028] Figure 7 This is a schematic diagram showing the comparison of dynamic conflict thresholds for the bandwidth domain, queue domain, identifier domain, and policy priority domain in the first embodiment of the present invention.

[0029] Figure 8 This is a schematic diagram illustrating the conflict relationship of the weighted independent set and the application retention results in the first embodiment of the present invention.

[0030] Figure 9 This is a schematic diagram of the normal path and failure rollback branch reserved for the two-stage lock atoms in the first embodiment of the present invention.

[0031] Figure 10 This is a schematic diagram of the link congestion superposition utilization rate distribution before and after the second resolution in the first embodiment of the present invention.

[0032] Figure 11 This is a schematic diagram comparing the utilization rate, multidimensional reachability loss, and key flow quality indicators before and after secondary congestion resolution in the first embodiment of the present invention.

[0033] Figure 12 This is a schematic diagram of the normal commit path and failure rollback branch for multi-device two-phase transaction issuance in the first embodiment of the present invention.

[0034] Explanation of reference numerals in the attached diagram: 01, Core switch; 02, Spine switch; 03, Spine switch; 11, Leaf switch; 18, Leaf switch; 25, Leaf switch. Detailed Implementation

[0035] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0036] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the method for automatic generation and dynamic conflict verification of network configuration based on service migration according to the present invention. The first embodiment of the method for automatic generation and dynamic conflict verification of network configuration based on service migration according to the present invention is presented.

[0038] In the first embodiment, the method for automatic generation and dynamic conflict verification of network configuration based on service migration includes: This embodiment illustrates the technical effects of each step in a computer simulation verification scenario of business migration. Figures 2 to 12 The diagrams illustrate the verification results or execution relationships for staged resource usage, path sharing, multi-domain conflicts, dynamic verification, resource resolution and reservation, shadow simulation, and transactional deployment. The number of tasks, parameter groups, and statistical definitions for each diagram are set according to the corresponding verification scenario; data points not labeled with specific values ​​are explained using approximate readings.

[0039] Step S10: Obtain basic data of business migration resources. Based on the basic data of business migration resources, perform a multi-dimensional resource request characterization task using a hash and Laplace feature mapping mechanism, and output a multi-dimensional resource request vector. The scheduling system assigns the set of concurrent migration tasks accepted but not yet configured within the current cycle to this step, while the network controller provides a real-time network resource view. Each migration task includes at least a task identifier, service type, task priority, planned execution time window, dependent paths, and traffic parameters. The real-time network resource view covers the remaining bandwidth capacity of the link, the maximum number of available queues on the port, the maximum queue depth that the shared cache can accommodate, the idle segment of the identifier pool, and the policy priority value range. These contents are encapsulated together as the basic data for service migration resources. The planned execution time window is then discretized according to the preparation phase, synchronization phase, switching phase, and verification phase, forming a phase time slot matrix. The matrix elements are jointly determined by the preset resource occupancy coefficient, traffic burst coefficient, basic request volume, and service confidence coefficient for the corresponding phase, enabling the same task to differentiate the occupancy pressure on the same resource in different migration phases. The sequence of device interfaces traversed by the migration task is hashed to obtain the first-level path fingerprint, and the adjacency matrix of the subgraph of the migration task on the path coupling graph is Laplace eigenmapped to obtain the second-level path fingerprint. Finally, resource request data is organized according to task identifier, resource domain, resource key, request quantity, planned execution time window, task priority, first-level path fingerprint, second-level path fingerprint, traffic burst coefficient, business confidence coefficient, and request identifier. This data is then associated with the current occupied quantity, pre-validated but not yet submitted resource usage, available quantity, and total resource quantity of the corresponding resource domain, forming a multi-dimensional resource request vector. This vector unifies the time structure, path structure, and resource domain information into a single object, allowing for direct reading of subsequent concurrent conflict characterization.

[0040] like Figure 2 As shown, within the same scheduling cycle, migration tasks M01, M02, and M03 start at different times, but their planned execution time windows overlap. The preparation, synchronization, switching, and verification phases of each task use resource occupancy coefficients of 0.18, 0.72, 1.00, and 0.28, respectively. For requests with the same basic request volume, traffic burst coefficient, and service confidence coefficient, the resource request intensity formed during the switching phase is higher than that during the preparation and verification phases, while the synchronization phase maintains a consistently high occupancy. In time slots approximately 19 to 25, M01 is in the switching phase, while M02 and M03 are still in the synchronization phase; therefore, the shared resources bear the combined occupancy of one switching and two synchronization phases. Entering time slots approximately 25 to 28, M01 transitions to a low-occupancy verification phase, M02 begins switching, and M03 continues synchronization, thus changing the resource contention combination. The phase time slot matrix preserves these changes, allowing subsequent verification to determine which requests need to be superimposed on a slot-by-slot basis, rather than treating the entire overlapping time window as having the same peak occupancy. This scenario illustrates that step S10 can identify the competition between high-occupancy switching and continuous synchronization, and retain the decrease in resource demand after the task enters the verification phase, providing a specific basis for staggering the switching time window and arranging resource reservations.

[0041] like Figure 3 As shown, in a network consisting of leaf switches, spine switches, core switches, and egress routers, two migration tasks can originate from different leaf switches, pass through spine switch 02, and then converge into core switch 01. The complete device interface sequences of the two tasks are different, and their corresponding first-level path fingerprints are also different. However, the interface edge from spine switch 02 to core switch 01 may still be occupied simultaneously. Step S10, while recording the interface sequence hash digest, retains the Laplace feature description of the path coupling subgraph, enabling step S20 to combine resource keys and actual shared edge relationships to continue verifying shared paths for tasks with different complete paths. Path fingerprints are used to organize and retrieve path information. The specific attribution of shared edges is still determined by the task-dependent paths; two paths cannot be considered identical solely based on similarity in dimensionality reduction features. For this convergence scenario, the multi-dimensional resource request vector simultaneously carries time slots, path structure, and resource status. Subsequent steps can locate the competition at the convergence interface edge and its overlapping time slots, avoiding treating all interfaces on the two complete paths as common bottlenecks. This reduces ineffective restrictions on non-shared links and lowers the risk of missing local shared link competition. Figure 3 Leaf switch 25 is connected to core switch 01 via spine switch 03. Spine switch 03 is used to aggregate the branch where leaf switch 25 is located. This branch, together with leaf switches 11 and 18, forms a branch via spine switch 02, which converges at core switch 01, but does not pass through the interface edge from spine switch 02 to core switch 01. The above connection relationship is used to distinguish different device interface sequences and corresponding path edges. The path sharing relationship is still determined according to the interface edge actually traversed by the migration task.

[0042] Step S20: Based on the multi-dimensional resource request vector, a concurrent conflict relationship characterization task is performed using a hash domain index and graph model mechanism to output a multi-domain resource conflict graph; After receiving the multi-dimensional resource request vector generated in step S10, the scheduling system does not directly compare all requests pairwise. Instead, it first categorizes non-zero requests into hash-based domain indexes based on resource domain, resource key, and time slot. The time slot occupancy of each migration task at each migration stage is mapped to a time slot bitmap. The time window overlap rate of two task requests is obtained by dividing the number of set slots after a bitwise AND operation of the two bitmaps by the number of set slots of the smaller node. If two requests match at least one of the following: resource key, shared resource group, dependency path shared edge, or policy matching domain inclusion relationship, and the time window overlap rate is greater than zero, the scheduling system establishes a conflict edge between the corresponding nodes. The initial weight of the conflict edge depends on the combination of the above matching conditions, and is then weighted and summarized by combining the preset weights of each resource domain, resource coverage overlap, time window overlap rate, and resource scarcity coefficient. The resource scarcity coefficient is determined by the current occupied amount, the amount of resources that have passed pre-verification but have not yet been submitted within the current scheduling period, and the proportion of the short-term predicted new occupied amount relative to the total resource amount of the resource domain. The joint occupancy potential of dependent paths is calculated based on the shared edge set of task dependent paths. It integrates bandwidth domain weights, time window overlap rates, traffic burst coefficients of each task on the shared edges, basic bandwidth requirements, shared edge capacity, and shared edge resource scarcity coefficients into a bandwidth superposition pressure on the shared path, and merges it with the multi-domain conflict potential of the corresponding conflicting edges. The final output multi-domain resource conflict graph uses all resource requests as nodes, and the potential information on the edges directly describes the resource competition relationship between concurrent tasks.

[0043] like Figure 4 As shown, migration tasks M01 and M02 have candidate competition relationships in the bandwidth domain, queue domain, identifier domain, and policy priority domain, with corresponding conflict edge potential energies of 0.91, 0.67, 0.74, and 0.58, respectively; the bandwidth domain conflict edge potential energy between M02 and M03 is 0.34. These values ​​represent the combined results of resource coverage, time overlap, resource scarcity, and the impact of shared paths; they do not represent the probability of failure and do not directly replace the domain threshold judgment in step S30. In this scenario, the hash domain index established by resource domain, resource key, and time slot first categorizes applications that may compete for the same resource into their corresponding index ranges, and then establishes conflict edges for applications with a time window overlap rate greater than zero and that meet the resource association conditions. Therefore, the bandwidth competition between M01 and M02 and the bandwidth competition between M02 and M03 can be retained separately, and will not be mixed into an unlocatable task-level marker because M02 participates in both sets of competition simultaneously. After reading the graph in step S30, the bandwidth requests of M01 and M02 with higher potential energy can be verified first, while the competition information of other resource domains is retained. This process assigns the resource domains, requests, and time relationships that need to be verified to specific nodes and edges, so that subsequent resolution can be carried out on the actual competing objects.

[0044] like Figure 5As shown, the resource scarcity coefficients for the bandwidth domain, queue domain, identifier domain, and policy priority domain are 0.81, 0.71, 0.86, and 0.56, respectively. Each column consists of currently occupied, pre-verified but unsubmitted, and short-term predicted new occupancy. Taking a scenario where multiple migration tasks have passed pre-verification but have not yet been submitted to the device as an example, the current resource occupancy in the device's resource view does not yet include these pending submissions. If subsequent tasks are judged only based on the device's current idle capacity, they may re-submit the same identifier segment or bandwidth share that has already been used by other task plans. Step S20 incorporates pending submissions and predicted new occupancy into the scarcity coefficient, allowing this type of competition to enter the conflict edge potential energy before the actual device configuration changes. The identifier domain scarcity coefficient of 0.86 in the figure is higher than that of the bandwidth domain of 0.81, indicating that identifier resources under this parameter group also need to be carefully verified. The fact that there is still remaining bandwidth on the link does not presume that the overall migration task is executable. The reference line of 0.80 in the figure is used to assist in observing the resource tension level and does not replace the conflict judgment threshold of each resource domain. This scenario demonstrates that a multi-domain conflict graph can simultaneously reflect both active occupancy and concurrent requests that have not yet taken effect, reducing the likelihood of resource overlaps being exposed only when multiple tasks are checked separately and then submitted in a centralized manner.

[0045] Step S30: Based on the multi-domain resource conflict graph, the conflict level determination task is performed using the Hearst exponent dynamic conflict verification mechanism, and the conflict domain level result is output. After receiving the multi-domain resource conflict graph generated in step S20, the scheduling system first splits the conflicting edges and node request information according to the bandwidth domain, queue domain, identifier domain, and policy priority domain. The bandwidth domain verification extracts the request volume and planned execution time window of each task on the same link from the multi-dimensional resource request vector, and summarizes the original concurrent bandwidth according to the overlap of time windows. Network traffic exhibits self-similar burst characteristics in time; simple summation easily underestimates peak overlap. Therefore, the original concurrent bandwidth is corrected using the Hearst exponent and a preset self-similar penalty coefficient to obtain the equivalent concurrent bandwidth. The Hearst exponent is greater than 0.5 and less than 1; the closer it is to 1, the more significant the burst overlap. The queue domain verification considers both the upper limit of the number of available queues on the port and the upper limit of the queue depth that the shared cache can accommodate. It calculates the queue pressure of the number of task request queues relative to the upper limit of the number of available queues on the port, and the queue depth pressure of the task request queue depth relative to the upper limit of the queue depth that the shared cache can accommodate, taking the larger of the two as the comprehensive queue pressure. The identifier domain verification compares the start and end positions of the identifier segments requested by two tasks to obtain the identifier segment overlap. Then, it combines the time window overlap rate and the identifier resource scarcity coefficient of the corresponding identifier domain to calculate the identifier conflict potential energy. The policy priority domain verification determines the rule matching domain coverage based on the ratio of the intersection of the two rule matching spaces to the smaller of the two matching spaces, and determines the priority distance based on the relative magnitude of the difference between the two rule priority values ​​to the preset policy priority value range. The verification results of each resource domain are summarized into a comprehensive conflict potential energy. The conflict domain level results are then classified according to a preset conflict level threshold and output to step S40.

[0046] like Figure 6 As shown, in a simulation scenario with a link safety level capacity of 1640Mbps, when using three sets of self-similar penalty coefficients of 0.22, 0.38, and 0.54, the equivalent concurrent bandwidth increases with the increase of the Hearst exponent. When the Hearst exponent is approximately 0.78, the curve corresponding to the penalty coefficient of 0.38 is close to the safety level, the curve corresponding to the penalty coefficient of 0.54 exceeds the level, and the curve corresponding to the penalty coefficient of 0.22 is still below the level. This comparison illustrates that the same basic bandwidth application may create different reserved risks under different burst durations and correction parameters; the parameter groups used for comparison in the figure cannot be arbitrarily interchanged, and the actual judgment should use the Hearst exponent corresponding to the service traffic and the preset penalty coefficient. Step S30 compares the corrected equivalent concurrent bandwidth with the link remaining bandwidth capacity multiplied by the safety level, so that the continuous traffic bursts are included in the pre-deployment verification. Burst margin is reserved for concurrent migration close to the safety boundary, and application combinations exceeding the safety level are sent to subsequent resolution, rather than allowing configuration submission simply because the basic application volume has not exceeded the remaining capacity.

[0047] like Figure 7As shown, in another set of threshold verification scenarios in the bandwidth domain, the ratio of the original concurrent bandwidth to the remaining link bandwidth capacity is 0.72, which is lower than the link safety level of 0.82. After correction by the Hearst exponent, the equivalent concurrent bandwidth ratio is 0.88, exceeding the level of 0.06, thus indicating a bandwidth conflict. This set of data specifically presents the situation where the original demand can be met, but the corrected demand needs to be resolved. For the queue application in the same figure, the queue quantity pressure is 0.83, which has not yet exceeded 1.00, while the queue depth pressure is 1.12, which has exceeded the maximum depth that the shared cache can accommodate. Taking the larger value of the two pressures, the overall queue pressure is 1.12, and the queue domain should still be judged as a conflict. Taking the scenario where two migration tasks occupy different queues but share the same cache group as an example, the availability of queue numbers does not necessarily mean that the cache depth is sufficient; at the same time, verifying the quantity and depth can detect this combination of compliant quantity and insufficient shared cache, avoiding the risk of queue backlog only being exposed after resource reservation is completed. The denominator of the above bandwidth ratio is the remaining bandwidth capacity, which is different from the congestion superposition utilization rate calculated using the total link capacity in step S50.

[0048] like Figure 7 As shown, the overlap of the identifier segments of the two tasks is 0.52, exceeding the threshold of 0.18; the rule matching domain coverage is 0.76, exceeding the threshold of 0.55. Taking two service migration tasks applying for partially identical VLAN or VNI identifier segments within an overlapping time window as an example, a strong identifier conflict is only determined when the identifier conflict potential is greater than zero, the identifier segment overlap exceeds the threshold, and the time window overlap rate also exceeds the corresponding threshold. If the two applications are completely staggered in time, they cannot be considered to be competing simultaneously simply because the identifier segments overlap. For two access control rules covering the same service flow, it is also necessary to combine the priority distance and the conditions of different rule actions or being located in the same access control list chain to determine the policy priority conflict; a coverage of 0.76 is insufficient to complete this determination. Step S30 outputs the conflict information of the above resource domains respectively, enabling step S40 to distinguish between duplicate identifier allocation, insufficient cache depth, and policy execution order competition, and select alternative resources or resolution measures for the corresponding application items, reducing the risk of failing to locate the specific cause of the conflict by relying solely on a single total score.

[0049] Step S40: Based on the conflict domain level results, the resource reservation task is executed using weighted independent set resolution and a two-phase locked atomic reservation mechanism, and the resource reservation results are output. Step S40, based on the conflict domain level results obtained in step S30, has already marked the conflict levels according to the bandwidth domain, queue domain, identifier domain, and policy priority domain, and includes the conflict edge potential energy between resource requests within each conflict domain. The scheduling system first divides the resource requests into a conflict-free set and a conflict-prone set. Requests in the conflict-free set directly undergo two-phase lock atomic reservation, locking sequentially according to the preset partial order relationship between the identifier domain, bandwidth domain, queue domain, and policy priority domain. Resource domains without dependencies can be locked in parallel. The reservation state machine transitions between idle, reserved, under verification, committed, and released. If reservation fails in any resource domain, the system releases all locks acquired by the task and rolls back the reservation table, returning the task to the unreserved state. For conflict-prone requests, a weighted independent set optimization problem is constructed. The retention weight is jointly determined by the task priority, business confidence coefficient, and task duration. The conflict edge potential energy serves as a penalty for simultaneously retaining adjacent requests, and the request quantity for each resource domain does not exceed the corresponding available quantity as a constraint. In the greedy pre-selection phase, the value density of each conflicting application is calculated. The value density is determined by the retention weight and the potential energy of the conflict edges connected to the node. Pre-selection is completed by selecting nodes in descending order of value density and judging strong conflict edges. In the local search phase, unselected nodes are replaced with their selected neighbor nodes according to their weights and the objective function value after the swap. Tasks that have not obtained the original resource application right filter from available resources in the same resource domain for candidate alternative resources that simultaneously meet the preset satisfaction requirements, preset new conflict potential energy requirements, and preset switching cost requirements. Alternative resources are determined in descending order of satisfaction, ascending order of new conflict potential energy, and ascending order of switching cost. Reservation tokens are generated for retained applications and applications using alternative resources. Before the simulation is completed, the renewal period is determined based on the changing trend of the current resource scarcity coefficient relative to the initial reservation, ultimately forming the resource reservation result.

[0050] like Figure 8As shown, the retention weights of the six bandwidth resource requests M01 to M06 are 8.9, 6.2, 7.1, 5.4, 8.1, and 4.8, respectively. The conflicting edges in the graph are M01 and M02, M01 and M04, M02 and M03, M03 and M05, M04 and M05, and M05 and M06. After greedy pre-selection and local search, the graph retains M01, M03, and M06, with a total retention weight of 20.8, and there are no conflicting edges between the selected nodes. Taking the scenario of prioritizing requests with higher weights as an example, the single weight of M05 (8.1) is higher than that of M03 (7.1) and M06 (4.8), but M05 is adjacent to M03, M04, and M06. Directly retaining M05 would restrict the simultaneous execution of its adjacent requests; the retention combination shown in the graph can accommodate M03 and M06 simultaneously. This result indicates that the selection in step S40 needs to consider both the individual reservation value and its exclusionary relationship with other applications, and cannot be determined solely by the order of submission or the weight of individual applications to determine all reservation results. The independent relationship illustrated resolves the conflict constraints between nodes. The selected applications must still meet the availability constraints of each resource domain and be subject to subsequent atomic reservations. Its effect is to form candidate combinations that take into account business value within the allowable range of the conflict structure, and the resource availability of the combination is further confirmed by atomic reservations.

[0051] like Figure 9 As shown, the cross-resource domain reservation process includes a normal locking path and a rollback branch when the bandwidth domain fails. Taking an abnormal scenario where a task's identifier and queue domains have completed partial checks and acquired the corresponding locks, but the bandwidth domain reservation fails, step S40 releases all locks acquired by the task and cancels the reservation records already formed in this transaction; locks not yet acquired and incomplete reservations are not output as valid resource reservation results. The successful locking section in the diagram is used to show the normal path, while the rollback point and release arrow are used to show the handling after failure; both should be understood as different execution branches. For resource domains with dependencies, a pre-defined partial order constraint restricts their locking sequence; parallel locking is only allowed for resource domains without dependencies. This mechanism allows other tasks to re-apply for relevant resources after a failed reservation cancellation, reducing the half-reservation state where the identifier or queue is already occupied but a complete migration cannot be executed, and ensuring that the reservation results received in step S50 are consistent across resource domains.

[0052] exist Figure 8 The conflict resolution results shown are as follows: Figure 9In the business scenario corresponding to the reservation mechanism shown, tasks that have not obtained the right to apply for the original uplink can continue to filter for another available uplink from the bandwidth domain. Candidate links must simultaneously meet preset requirements for resource satisfaction, new conflict potential, and switching cost. Then, alternative resources are determined according to descending order of satisfaction, ascending order of new conflict potential, and ascending order of switching cost. In this way, tasks whose original applications were not reserved still have the opportunity to enter the reservation process with compliant alternative applications, rather than simply transferring conflicts to another congested link to increase the number of reserved resources. Both the original and alternative resources generate corresponding tokens after successful reservation, with the tokens indicating their validity period. If the shadow simulation has not yet ended, the renewal is determined based on the change in the current resource scarcity coefficient relative to the initial reservation. For situations where new tasks arrive in concentrated numbers during simulation and alternative links gradually become strained, the renewal check can limit the long-term occupation of resources by expired candidate solutions. This step connects application selection, alternative resource constraints, and reservation validity, providing a traceable and revokeable resource status for step S50.

[0053] Step S50: Based on the resource reservation results, the Monte Carlo shadow simulation mechanism is used to perform the pre-simulation verification and transactional distribution tasks, and the incremental network configuration distribution results are output.

[0054] In the resource reservation results given in step S40, each migration task's requests in the bandwidth, identifier, queue, and policy priority domains all carry a locked status, a reservation token, and an expiration date. The scheduling system first writes these reservation items into the shadow link bandwidth table, shadow identifier allocation table, shadow queue list, and shadow policy sequence table, respectively. The shadow tables are isolated from the real device configurations but maintain a consistent structure, thus allowing the incremental configurations of all concurrent tasks to be overlaid without affecting the live network forwarding. Subsequently, a set of key flows is selected from the set of service flows to be verified. During the selection, the service flow weight, the number of shared edges traversed, and the share of resources occupied in multiple resource domains are considered to ensure that subsequent simulations focus on the highest-risk traffic. Shadow simulation simultaneously overlays the shadow incremental configurations of all concurrent tasks onto the simulation topology, constructs a forwarding behavior model, and calculates the link congestion overlay utilization rate using equivalent concurrent bandwidth. At the same time, it calculates multidimensional reachability loss based on changes in packet loss rate, increase in latency, and increase in jitter. Link congestion utilization reflects the bandwidth pressure of shared links during concurrent switching, while multidimensional reachability loss converts the degree of deterioration of different service quality indicators into a unified dimensionless quantity. Together, they constitute the simulation verification index.

[0055] like Figure 10As shown, after overlaying the shadow incremental configurations of all concurrent migration tasks onto the same simulation topology, the link congestion overlay utilization samples before the secondary resolution are mostly distributed to the right of the congestion threshold of 0.86. After shifting the switching time window of the relevant tasks, the sample distribution shifts overall towards the lower utilization range. This change corresponds to the stage occupancy expression in step S10: shifting the switching time window changes the overlap between the high-occupancy stage and the synchronization and switching stages of other tasks, thereby reducing the equivalent concurrent demand of shared links in the same time slot, without requiring a reduction in the total basic request volume of the task itself. Step S50 verifies this adjustment in the shadow resource table, allowing combinations that fail the verification to be re-resolved and reserved, and then re-overlayed for simulation. The illustrated distribution may still retain tail samples exceeding the threshold; whether it actually passes should be determined according to the preset statistical caliber; for example, 800 Monte Carlo simulations can be used, and the thresholds can be compared using the 95th percentile values ​​of the congestion index and the loss index respectively. This processing can observe high-risk results under traffic disturbances and reduce the chance of missing sudden overlays by only checking a single deterministic snapshot.

[0056] like Figure 11 As shown, before the second resolution, the congestion overlay utilization rate was approximately 0.93, and the multidimensional reachability loss was approximately 0.17. After the second resolution, these two indicators decreased to approximately 0.79 and approximately 0.08, respectively. Using a congestion utilization rate threshold of 0.86 and a reachability loss threshold of 0.12 as the judgment parameters in this example, both indicators exceeded the thresholds before resolution, and both indicators fell below the thresholds after resolution. Therefore, step S503 triggers a second resolution or rollback in the previous state. Only after re-reservation and successful re-deduction is the incremental configuration difference set for formal deployment allowed to be generated. Calculated using the approximate values ​​shown in the figure, the congestion overlay utilization rate decreased by approximately 0.14, and the multidimensional reachability loss decreased by approximately 0.09, resulting in a relative decrease in loss of approximately 53%. These values ​​illustrate the improvement after resource combination adjustment in this simulation scenario. This process translates the changes shown in the figure into executable decisions, making congestion risk and business quality risk prerequisites before the configuration officially takes effect.

[0057] like Figure 11As shown, the packet loss rate decreased from approximately 0.015 to approximately 0.008, the normalized value of the latency increase decreased from approximately 0.91 to approximately 0.48, and the normalized value of the jitter increase decreased from approximately 0.72 to approximately 0.41, all three indicators showing a downward trend. Step S502 first determines the packet loss rate change, latency increase, and jitter increase under the same comparison baseline for each critical flow. Then, it normalizes the latter two items using the latency tolerance upper limit and jitter tolerance upper limit for that flow, respectively. The flow-by-flow loss is then synthesized according to preset loss weights, and finally summarized according to the service flow weights. Preset loss weights can be 0.42, 0.34, and 0.24. The service flow weights are normalized within the critical flow set. The numerator and denominator of the latency and jitter ratios use the same time unit, and the tolerance upper limit is a positive value. For high-weight services that are sensitive to latency and jitter, even if the increase in normalized latency has not exceeded 1, changes in packet loss, latency, and jitter may still collectively cause the multidimensional loss to exceed the threshold of 0.12. In the figure, the normalized latency value of 0.91 does not exceed 1; whether the overall pass is achieved still needs to be determined based on the comparison between the multidimensional reachability loss and the preset threshold, rather than solely by the improvement of a single indicator. The overall displayed value and the flow-by-flow weighted and quantile statistical results in the figure should be interpreted according to their respective definitions, and several overall readings should not be directly substituted into the loss calculation of the same flow. This multidimensional determination allows configuration combinations that can still forward but whose service quality has significantly deteriorated to be identified, and ensures that the effect of secondary mitigation simultaneously covers bandwidth pressure and critical service quality.

[0058] like Figure 12 As shown, when a single migration configuration involves core switch 01, spine switch 02, leaf switch 11, and leaf switch 18, each device may complete pre-commit, local verification, commit, and configuration confirmation at different relative times. Step S50 first organizes configuration operations according to resource dependencies, such as creating identifiers before binding ports, reserving queues before enabling rate limiting, so that the operations received by the devices meet local dependency conditions; at the same time, it forms compensation operations and configuration fingerprints for each operation. Only when all target devices successfully pre-commit does the transaction enter the commit phase. Taking leaf switch 18's pre-commit failure or timeout as an example, the system broadcasts a rollback for the transaction, causing other devices that have entered pre-commit to undo the corresponding temporary changes; if the failure occurs after commit, an idempotent rollback is performed based on the recorded compensation operations and configuration fingerprints to avoid repeatedly undoing other configurations when retrying the same failure notification. Normal commits and failure rollbacks are executed according to the corresponding transaction branches, and the time difference in the device phase is not considered as a basis for allowing partial commits. This mechanism maps the simulated resource combinations, the configuration difference sets with dependent order, and the multi-device rollback paths to each other, reducing the risk of inconsistencies in the state where some devices have adopted the new configuration while others still retain the old configuration, and preserving a basis for locating and recovering from failed operations.

[0059] Example 2: Furthermore, the present invention provides an automatic network configuration generation and dynamic conflict verification system based on service migration, employing a method for automatic network configuration generation and dynamic conflict verification based on service migration as described in the above embodiments, which can solve the technical problem of automatic network configuration generation and dynamic conflict verification based on service migration. The beneficial effects of the automatic network configuration generation and dynamic conflict verification system based on service migration provided by the present invention are the same as those of the method for automatic network configuration generation and dynamic conflict verification based on service migration provided in the above embodiments, and other technical features of the automatic network configuration generation and dynamic conflict verification system based on service migration are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0060] Example 3: This invention provides a network configuration automatic generation and dynamic conflict verification device based on service migration. The device includes at least one processor and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which are then executed to enable the at least one processor to perform the network configuration automatic generation and dynamic conflict verification method based on service migration described in Example 1. The network configuration automatic generation and dynamic conflict verification device based on service migration in this invention can include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This network configuration automatic generation and dynamic conflict verification device based on service migration is merely an example and should not limit the functionality or scope of the invention. A network configuration auto-generation and dynamic conflict checking device based on service migration may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in read-only memory or a program loaded from a storage device into random access memory. The random access memory also stores various programs and data required for the operation of the network configuration auto-generation and dynamic conflict checking device based on service migration. The processing unit, read-only memory, and random access memory are interconnected via a bus. An I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the network configuration auto-generation and dynamic conflict checking device based on service migration to communicate wirelessly or wiredly with other devices to exchange data. While a network configuration auto-generation and dynamic conflict checking device based on service migration with various systems has been described, it should be understood that implementation or possession of all the described systems is not required. It can be implemented alternatively or with more or fewer systems.

[0061] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for automatic generation and dynamic conflict verification of network configuration based on service migration. The computer program product provided by this invention can solve the technical problem of automatic generation and dynamic conflict verification of network configuration based on service migration. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the method for automatic generation and dynamic conflict verification of network configuration based on service migration provided in the above embodiments, and will not be repeated here.

[0062] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0063] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0064] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A method for automatic generation and dynamic conflict verification of network configuration based on service migration, characterized in that, The method includes: Step S10: Obtain basic data of business migration resources. Based on the basic data of business migration resources, perform a multi-dimensional resource request characterization task using a hash and Laplace feature mapping mechanism, and output a multi-dimensional resource request vector. Step S20: Based on the multi-dimensional resource request vector, a concurrent conflict relationship characterization task is performed using a hash domain index and graph model mechanism to output a multi-domain resource conflict graph; Step S30: Based on the multi-domain resource conflict graph, the conflict level determination task is performed using the Hearst exponent dynamic conflict verification mechanism, and the conflict domain level result is output. Step S40: Based on the conflict domain level results, the resource reservation task is executed using weighted independent set resolution and a two-phase locked atomic reservation mechanism, and the resource reservation results are output. Step S50: Based on the resource reservation results, the Monte Carlo shadow simulation mechanism is used to perform the pre-simulation verification and transactional distribution tasks, and the incremental network configuration distribution results are output.

2. The method for automatic generation and dynamic conflict verification of network configuration based on service migration as described in claim 1, characterized in that, Step S10, which involves obtaining basic data on business migration resources, and based on this data, performing a multi-dimensional resource request characterization task using a hash and Laplacian feature mapping mechanism to output a multi-dimensional resource request vector, specifically includes: Step S101: Obtain the set of concurrent migration tasks and the real-time network resource view within the current scheduling period, and encapsulate the set of concurrent migration tasks and the real-time network resource view into the basic data of the service migration resources. Each migration task in the set of concurrent migration tasks includes a task identifier, service type, task priority, planned execution time window, dependency path, and traffic parameters. The real-time network resource view includes the remaining bandwidth capacity of the link, the upper limit of the number of available queues on the port, the upper limit of the queue depth that the shared cache can accommodate, the idle segment of the identifier pool, and the range of policy priority values. Step S102: Based on the basic data of the business migration resources, the planned execution time window is discretized according to the migration stage, and a stage time slot matrix is ​​constructed according to the preset resource occupancy coefficient, traffic burst coefficient, basic application volume and business confidence coefficient of each migration stage to form stage resource description data. Step S103: Based on the stage resource description data, hash the sequence of device interfaces traversed by the migration task to obtain the first-level path fingerprint, perform Laplace eigenmap on the adjacency matrix of the subgraph of the migration task on the path coupling graph to obtain the second-level path fingerprint, and organize resource request data according to task identifier, resource domain, resource key, request quantity, planned execution time window, task priority, first-level path fingerprint, second-level path fingerprint, traffic burst coefficient, service confidence coefficient, and request identifier. Associate the resource request data with the current occupied quantity of the corresponding resource domain, the occupied quantity of resources that have passed pre-verification but have not yet been submitted, the available quantity, and the total quantity of resources to obtain the multi-dimensional resource request vector.

3. The method for automatic generation and dynamic conflict verification of network configuration based on service migration as described in claim 1, characterized in that, Step S20, based on the multi-dimensional resource request vector, employs a hash-based domain indexing and graph model mechanism to perform a concurrent conflict relationship characterization task and outputs a multi-domain resource conflict graph. This step specifically includes: Step S201: Perform hash domain indexing on the multidimensional resource request vector according to resource domain, resource key and time slot, and map the non-zero resource requests of each migration stage to the corresponding time slot to form domain time slot index data; Step S202: Based on the domain time slot index data, extract the task dependency path fingerprint, determine the time window overlap rate by the ratio of the number of overlapping slots corresponding to the time slots of the two task requests to the number of non-zero slots of the lesser one, and establish a conflict edge when the two task requests meet at least one of the following conditions: the resource key is the same, they belong to the same shared resource group, the dependency path has a shared edge, or the strategy matching domain has an inclusion relationship, and the time window overlap rate is greater than zero. Step S203: Based on the initial resource conflict graph, according to the preset resource domain weights of each resource domain, the product of resource coverage overlap, time window overlap rate and resource scarcity coefficient is weighted and summarized. The resource scarcity coefficient is determined according to the ratio of the current occupied amount, the occupied amount of resources that have passed pre-verification but have not yet been submitted, and the predicted new occupied amount to the total resource amount of the resource domain. The conflict edges are updated in combination with the joint occupied potential of dependent paths to obtain the multi-domain resource conflict graph.

4. The method for automatic generation and dynamic conflict verification of network configuration based on service migration as described in claim 3, characterized in that, In step S203, the calculation step of the joint occupancy potential of the dependency paths includes determining the shared edge set of the task dependency paths based on the task dependency path fingerprint, and calculating the joint occupancy potential of the dependency paths according to the following formula: ， in, For the dependent path to jointly occupy potential energy, Shared edge set for task-dependent paths For shared edge indexes, For bandwidth domain weights, For time window overlap rate, For task indexing, For the task Dependency path, For the task Traffic burst coefficient, For the task In shared edge Basic bandwidth requirements, For shared edges capacity, For shared edges The resource scarcity coefficient; the bandwidth domain weight, the time window overlap rate, the resource scarcity coefficient and the joint occupancy potential of the dependent path are all dimensionless quantities, and the basic bandwidth requirement and the capacity of the shared edge use the same bandwidth unit; the joint occupancy potential of the dependent path is merged with the multi-domain conflict potential of the corresponding conflict edge to obtain the updated conflict edge potential.

5. The method for automatic generation and dynamic conflict verification of network configuration based on service migration as described in claim 3, characterized in that, Step S30, which involves performing a conflict level determination task based on the multi-domain resource conflict graph using the Hearst exponent dynamic conflict verification mechanism and outputting the conflict domain level result, specifically includes: Step S301: Based on the multi-domain resource conflict graph, extract the remaining bandwidth capacity of the corresponding link, the upper limit of the number of available queues at the port, and the upper limit of the queue depth that the shared cache can accommodate from the multi-dimensional resource request vector. Summarize the concurrent bandwidth demand according to the request quantity and planned execution time window in the multi-dimensional resource request vector, and correct the concurrent bandwidth demand using a Hearst exponent greater than 0.5 and less than 1 and a preset self-similarity penalty coefficient to obtain the equivalent concurrent bandwidth. Calculate the queue number pressure of the number of task request queues relative to the upper limit of the number of available queues at the port and the queue depth pressure of the task request queue depth relative to the upper limit of the queue depth that the shared cache can accommodate. Determine the larger of the two values ​​as the comprehensive queue pressure, and combine it with the equivalent concurrent bandwidth to form the first conflict verification data. Step S302: Based on the first conflict verification data, determine the overlap of the two task application flag segments according to the overlapping interval of the flag segments, and determine the flag conflict potential energy by combining the time window overlap rate and the flag resource scarcity coefficient of the corresponding flag domain in the resource scarcity coefficient; obtain the preset policy priority value range, determine the rule matching domain coverage according to the ratio of the intersection of the two rule matching spaces to the smaller one's matching space, and determine the priority distance according to the difference between the two rule priority values ​​relative to the preset policy priority value range; combine the flag conflict potential energy, the rule matching domain coverage, and the priority distance to form the second conflict verification data. Step S303: Based on the second conflict verification data, determine bandwidth conflict when the equivalent concurrent bandwidth exceeds the product of the remaining bandwidth capacity of the link and the preset link security level; determine queue conflict when the comprehensive queue pressure is greater than 1; determine strong identifier conflict when the identifier conflict potential is greater than zero, the identifier segment overlap exceeds the preset identifier segment overlap threshold, and the time window overlap rate exceeds the preset time window overlap rate threshold; determine policy priority conflict when the rule matching domain coverage exceeds the preset matching domain coverage threshold, the priority distance is less than the preset priority distance threshold, and at least one of the following conditions is met: different rule actions or rules are located in the same access control list chain; summarize the conflict potential of each resource domain, and determine the conflict domain level result according to the preset conflict level threshold.

6. The method for automatic generation and dynamic conflict verification of network configuration based on service migration as described in claim 5, characterized in that, Step S40, based on the conflict domain level results, involves performing a resource reservation task using weighted independent set resolution and a two-phase lock atomic reservation mechanism, and outputting the resource reservation results. Specifically, this includes: Step S401: Based on the conflict domain level results, lock the non-conflicting resource requests in the order of the preset partial order relationship between the identifier domain, bandwidth domain, queue domain and policy priority domain, lock the resource domains that have no dependency relationship in parallel, and release all the acquired locks and roll back the reservation table when the reservation of any resource domain fails, forming the initial resource reservation data. Step S402: Based on the initial resource reservation data, construct a weighted independent set optimization problem for conflicting resource requests. Determine the retention weight of resource requests based on task priority, business confidence coefficient, and task duration. Determine the conflict penalty for retaining adjacent resource requests based on conflict edge potential energy. Under the constraint that the request amount in each resource domain does not exceed the corresponding available amount, use greedy pre-selection and local search to determine the retained resource requests, forming conflict resolution data. Step S403: Based on the conflict resolution data, for tasks that have not obtained the original resource application right, candidate alternative resources that simultaneously meet the preset satisfaction threshold, the preset new conflict potential threshold, and the preset switching cost threshold are selected from the available resources belonging to the same resource domain as the original resource application. The alternative resources are determined according to the descending order of satisfaction, the ascending order of new conflict potential, and the ascending order of switching cost. Reserved tokens are generated for the retained resource application and the resource application using the alternative resources. When the simulation is not completed, the renewal is determined according to the change ratio of the current resource scarcity coefficient relative to the resource scarcity coefficient at the initial reservation, thus forming the resource reservation result.

7. The method for automatic generation and dynamic conflict verification of network configuration based on service migration as described in claim 6, characterized in that, Step S50, based on the resource reservation results, involves using a Monte Carlo shadow simulation mechanism to perform pre-simulation verification and transactional deployment tasks, and outputting incremental network configuration deployment results. This step specifically includes: Step S501: Based on the resource reservation results, write the reserved bandwidth resources, identifier resources, queue resources, and policy order resources into the corresponding shadow resource tables. Overlay the shadow incremental configurations of all migration tasks corresponding to the resource reservation results onto the simulation topology to construct a forwarding behavior model. Obtain the set of service flows to be verified corresponding to the resource reservation results, and select a set of key flows from the set of service flows to be verified. This forms shadow projection data; Step S502: Based on the shadow projection data, determine the link congestion aggregation utilization rate according to the ratio of the equivalent concurrent bandwidth to the corresponding link capacity, and calculate the multidimensional reachability loss according to the following formula: ， in, For multidimensional accessibility loss, For the set of key flows, Index for business flow, For business flow The weight, , and To preset the loss weight, For business flow Changes in packet loss rate For business flow The increase in latency, For business flow The preset latency tolerance limit, For business flow The amount of vibration increases. For business flow The preset jitter tolerance limit; the weight of the service flow, the preset loss weight, the packet loss rate change, the multidimensional reachability loss, the ratio of the increase in latency to the preset latency tolerance limit, and the ratio of the increase in jitter to the preset jitter tolerance limit are all dimensionless quantities; the link congestion superposition utilization rate and the multidimensional reachability loss are combined to form a simulation verification index; Step S503: Based on the simulation verification index, when the multidimensional reachability loss exceeds a preset reachability loss threshold or the link congestion superposition utilization exceeds a preset congestion utilization threshold, a secondary resolution or rollback is triggered; otherwise, an incremental configuration difference set is generated. The order of distribution of each configuration operation in the incremental configuration difference set is determined according to the resource dependency relationship. A compensation operation and configuration fingerprint are generated for each configuration operation. A two-phase transaction protocol is used to perform pre-commit and commit to the target device. If any target device fails to commit or times out, a rollback is broadcast. If a device fails to execute after commit, an idempotent rollback is performed according to the compensation operation and the configuration fingerprint to obtain the incremental network configuration distribution result.

8. A network configuration automatic generation and dynamic conflict verification system based on service migration, applied to the network configuration automatic generation and dynamic conflict verification method based on service migration as described in any one of claims 1 to 7, characterized in that, The system includes: The multi-dimensional resource request characterization module is used to obtain basic data of business migration resources, and based on the basic data of business migration resources, it performs the multi-dimensional resource request characterization task using a hash and Laplace feature mapping mechanism, and outputs a multi-dimensional resource request vector. The concurrent conflict relationship characterization module is used to perform concurrent conflict relationship characterization tasks based on the multi-dimensional resource request vector, using hash domain indexing and graph model mechanism, and output a multi-domain resource conflict graph; The multi-domain dynamic conflict verification module is used to perform conflict level determination tasks based on the multi-domain resource conflict graph and using the Hearst exponent dynamic conflict verification mechanism, and output the conflict domain level results. The resource reservation and conflict resolution module is used to perform resource reservation tasks based on the conflict domain level results, using weighted independent set resolution and a two-phase locked atomic reservation mechanism, and output resource reservation results. The pre-simulation and transactional deployment module is used to perform pre-simulation verification and transactional deployment tasks based on the resource reservation results, using the Monte Carlo shadow simulation mechanism, and output incremental network configuration deployment results.

9. A network configuration automatic generation and dynamic conflict verification device based on service migration, characterized in that, The network configuration automatic generation and dynamic conflict verification device based on service migration includes: a memory, a processor, and a network configuration automatic generation and dynamic conflict verification program based on service migration stored in the memory and executable on the processor. When the network configuration automatic generation and dynamic conflict verification program based on service migration is executed by the processor, it implements a network configuration automatic generation and dynamic conflict verification method based on service migration as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a network configuration automatic generation and dynamic conflict verification program based on service migration. When the network configuration automatic generation and dynamic conflict verification program based on service migration is executed by the processor, it implements a network configuration automatic generation and dynamic conflict verification method based on service migration as described in any one of claims 1 to 7.

Images