Route dynamic generation method and system for virtual private line distribution

Abstract

The invention discloses a route dynamic generation method and system for virtual private line distribution, and particularly relates to the technical field of network optimization, and the method comprises the following steps: collecting a flow index in a set time window, calculating a flow load index, and dividing a low-flow environment, a medium-flow environment and a high-flow environment; determining whether to execute path adjustment analysis or not according to the environment type, calculating a time overhead value to evaluate path re-calculation cost, and calculating a path calculation load value to evaluate the calculation capability of the controller; preset strategy adjustment is performed according to the traffic environment, and whether a standby special line is started or not is intelligently decided by combining CNN model analysis, so that the network resource utilization rate is improved, the calculation load is reduced, and the path distribution is optimized; the method can solve the problems that a standby special line is easy to idle for a long time or unnecessary switching is easy to cause and resources are wasted in a traditional fixed path distribution mode, ensures that path adjustment is only carried out under the conditions of sufficient computing resources and acceptable overhead, and improves the network stability.

Description

Technical Field

[0001] The present invention relates to the technical field of network optimization, and more specifically, to a method and system for dynamically generating routes for virtual dedicated line allocation. Background Art

[0002] In complex network environments such as cloud computing, large-scale data centers, enterprise private networks, and 5G networks, virtual private lines (VPLs) are a common solution. They are mainly used for: cross-regional data center interconnection, enterprise-level VPN communication, high-priority service assurance, 5G network slicing, and private line scheduling. However, traditional virtual private line management methods have the following main problems:

[0003] The static path allocation mode leads to low resource utilization. Most existing virtual private lines use fixed path allocation, that is, the primary path and backup path are determined when the private line is created, and the traffic is forwarded according to a fixed strategy. Disadvantages: When a path has sufficient bandwidth, the backup private line may be idle for a long time, wasting network resources; when the primary path is congested, the backup private line may not be activated in time, resulting in service interruption.

[0004] The path switching strategy based on fixed thresholds is too simple. Traditional methods usually use fixed thresholds (such as switching dedicated lines when bandwidth utilization exceeds 80%), but network traffic is dynamic. Fixed thresholds may cause the activation of backup dedicated lines too early or too late: switching too early → long-term occupation of backup dedicated lines, increasing additional costs; switching too late → congestion of the main path affects business stability. Therefore, a method and system for dynamically generating routes for virtual dedicated line allocation are proposed in this paper to solve the above problems. Summary of the invention

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for dynamically generating routes for virtual private line allocation comprises the following steps:

[0007] Within the set time window, multiple preset traffic indicators of the current network are collected and the traffic load index is calculated, and then the traffic environment type is divided according to the traffic load index;

[0008] Determine whether to perform path adjustment analysis according to the traffic environment type obtained by the classification, and during the path adjustment analysis, respectively calculate the time overhead value used to evaluate the time required for recalculating the route and the path calculation load value used to evaluate the current network controller computing power;

[0009] Adjust the preset strategy accordingly based on the traffic environment type obtained by the division, and decide whether to directly enable the backup dedicated line based on the results of the path adjustment analysis.

[0010] In a preferred embodiment, dividing the traffic environment type according to the traffic load index means that:

[0011] Low traffic environment: FLI ≤ T1, with light network load; Medium traffic environment: T1 < FLI ≤ T2, with moderate network load; High traffic environment: FLI > T2, with congestion risk. FLI represents the traffic load index, and T1 and T2 are the preset division thresholds one and two. And in the low traffic environment, the standby dedicated line is not enabled. In the high traffic environment, the standby dedicated line is directly enabled. In the medium traffic environment, path adjustment analysis is carried out in combination with the historical virtual dedicated line allocation scheme given by the preset path adjustment model.

[0012] When the divided traffic environment type is the medium traffic environment, path adjustment analysis is carried out.

[0013] In a preferred embodiment, the acquisition logic of the traffic load index is as follows:

[0014] Obtain the average bandwidth utilization rate, average packet loss rate, active connection number corresponding to the 90th percentile, average traffic change rate in the current time window, and the delay jitter obtained by calculating the degree of change in the transmission delay between consecutive data packets through the root mean square calculation formula.

[0015] The calculation formula of the traffic load index is:

[0016] P represents the average bandwidth utilization rate in the current time window, J represents the delay jitter, R represents the average packet loss rate, A represents the active connection number corresponding to the 90th percentile, F represents the average traffic change rate, C represents a preset constant and is a positive value, w1, w2, w3, w4, and w5 are all preset proportionality coefficients, and FLI represents the traffic load index.

[0017] In a preferred embodiment, the acquisition logic of the time overhead value is as follows:

[0018] Obtain the number of nodes N in the current network topology topo , the available computing power C of the current routing calculation device proc , the convergence time T of the current network routing protocol conv , the historical routing adjustment frequency H freq , in the virtual dedicated line allocation scheme given by the preset path adjustment model in the historical situation, the average change ratio R of the routing table entries affected by the adjustment change , the reference routing table change threshold R th , the calculation formula of the time overhead value is:

[0019] PAO represents the time overhead value.

[0020] In a preferred embodiment, the logic for obtaining the path calculation load value is:

[0021] Get the current path calculation task queue length, that is, the number of path tasks waiting to be calculated M queue , the number of path calculation tasks currently being executed in parallel N exec , the average waiting time of the path calculation task is R wait , the average completion time of path calculation is T comp , the number of path calculation failures in the past preset fixed time period Q failure , the total number of path calculation requests in the past preset fixed time Q total , the number of active computing tasks W of the current computing controller active , the calculation formula for path calculation load value is:

[0022] PCL indicates the path calculation load value.

[0023] In a preferred embodiment, determining whether to directly enable the standby dedicated line according to the result of the path adjustment analysis refers to:

[0024] The results of the path adjustment analysis, namely the path calculation load value and the time overhead value, are substituted into the pre-trained convolutional neural network model. If the output result of the convolutional neural network model is 1, the backup dedicated line is directly enabled. If the output result of the convolutional neural network model is 0, the virtual dedicated line allocation plan obtained by the preset path adjustment model is executed.

[0025] In a preferred embodiment, a system for dynamically generating routes for virtual private line allocation includes:

[0026] The data collection module collects multiple preset traffic indicators of the current network within a set time window;

[0027] The data analysis module calculates the traffic load index based on the data collected by the data collection module, and then divides the traffic environment type according to the traffic load index;

[0028] A path adjustment analysis module determines whether to perform path adjustment analysis according to the traffic environment type obtained by the classification, and during the path adjustment analysis, respectively calculates a time overhead value for evaluating the time required for recalculating the route and a path calculation load value for evaluating the current network controller computing power;

[0029] The dynamic adjustment module adjusts the corresponding preset strategy according to the traffic environment type obtained by classification, and decides whether to directly enable the backup dedicated line based on the results of the path adjustment analysis.

[0030] Technical effects and advantages of the present invention:

[0031] The traditional fixed path allocation mode is prone to long-term idle or unnecessary switching of backup dedicated lines, wasting resources. The present invention divides the network traffic environment (low, medium, high) through FLI, and adopts different strategies according to different environments: low traffic environment (FLI≤T1) → continue to use the current path without adjusting the routing. Medium traffic environment (T1<FLI≤T2) → perform path optimization adjustment. High traffic environment (FLI>T2) → directly enable the backup dedicated line to reduce the pressure on the main path. Avoid bandwidth waste caused by static path management in the traditional method, and improve the utilization rate of the main path and backup dedicated lines.

[0032] Traditional methods often ignore the computational cost of path adjustment, which may lead to: too high computational load → affecting other services of the controller; waste of computational resources → unnecessary path recalculation; The present invention calculates the cost of path adjustment through PCL (path calculation load value) and PAO (time overhead value) to ensure the reasonable allocation of computational resources: PCL evaluates the computational load of the current network controller to prevent computational overload from affecting services. PAO evaluates the time overhead required for path adjustment to avoid network instability caused by frequent adjustment. Ensure that path adjustment is only carried out when there is sufficient computational resources and the overhead is acceptable, improving network stability.

[0033] The present invention uses CNN combined with PCL and PAO for intelligent decision-making, dynamically adjusting the strategy for enabling backup dedicated lines: avoiding resource waste or service interruption caused by misjudgment of fixed thresholds. Improving the utilization rate of backup dedicated lines while ensuring service stability. Traditional methods may frequently adjust paths, resulting in excessive consumption of the controller's computational resources.

[0034] In the medium traffic environment, the present invention uses a preset path adjustment model to optimize the path and only promotes load balancing when necessary: diverting part of the traffic from the main path to the low-load path to optimize resource utilization. Dynamic QoS adjustment: improving the bandwidth guarantee for high-priority services while optimizing the path for low-priority traffic. MPLS-TE routing optimization: using the CSPF (Constrained Shortest Path First) algorithm to recalculate the path. The traffic patterns of modern networks are uncertain and dynamically changing, and it is difficult for traditional methods to adaptively adjust. The present invention combines the real-time calculation of FLI to achieve adaptive classification of the network environment and carry out refined management in the medium traffic environment in combination with path adjustment analysis. Description of the Drawings

[0035] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the accompanying drawings;

[0036] Figure 1 It is a schematic diagram of the principle of a method for dynamically generating a route for virtual dedicated line allocation in the present invention.

[0037] Figure 2 A schematic diagram of a dynamic routing generation system for virtual dedicated line allocation in the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Reference Figure 1 - Figure 2 The following embodiments are obtained:

[0040] Example 1: In modern complex network environments, Virtual Private Line (VPL) is widely used in enterprises, data centers and cloud computing environments to improve the reliability, bandwidth guarantee and security of data transmission. However, existing static or semi-dynamic path allocation methods face the following challenges:

[0041] Path adjustment lag: Traditional methods usually rely on preset paths or fixed load balancing strategies. When traffic patterns change or there is sudden congestion, the response speed of adjusting the path is slow, which may lead to reduced data transmission efficiency.

[0042] Excessive consumption of computing resources: Network controllers (such as SDN controllers) perform path recalculation in high-traffic environments, which has high computing overhead and may affect other control tasks.

[0043] Low utilization of backup dedicated lines: Currently, the use of backup dedicated lines is usually based on manual configuration or simple threshold triggering. In actual applications, the backup dedicated lines may be idle for a long time or fail to be activated in time when they are really needed.

[0044] The present invention proposes a virtual dedicated line allocation method based on adaptive adjustment of traffic environment, combining flow load index (FLI), path calculation load value (PCL), and time overhead value (PAO) to dynamically evaluate network conditions to solve the following problems:

[0045] Path adjustment caused by network traffic fluctuations: The network status is monitored in real time through the traffic load index (FLI), and the traffic environment is divided into three types: low traffic, medium traffic, and high traffic. Different strategies are adopted according to different environments to avoid unnecessary computing overhead.

[0046] The high overhead problem of path adjustment calculation: The current network controller computing power is evaluated through the path calculation load value (PCL), and the time cost required for path recalculation is calculated in combination with the time overhead value (PAO), ensuring that the path adjustment is performed under the appropriate computing load and improving the adjustment efficiency.

[0047] Backup dedicated line activation policy optimization: Through the convolutional neural network (CNN) model, intelligent decisions are made on whether to enable the backup dedicated line based on the results of path adjustment analysis, rather than based on simple traffic threshold triggering, thereby improving the utilization efficiency of the backup dedicated line.

[0048] Objectives of the present invention: The present invention hopes to optimize the efficiency of dynamic adjustment of network routing through an intelligent and adaptive virtual dedicated line allocation mechanism. Specific objectives include: improving network resource utilization, optimizing the use of main paths and backup dedicated lines through traffic environment division, and avoiding long-term idleness of backup dedicated lines or waste of path computing resources. Reduce the computational cost of path adjustment, combine PCL and PAO to perform path calculation load analysis, ensure that the network controller makes adjustments under an acceptable computing load, and prevent the controller from affecting business continuity due to computing overload. Improve the accuracy of backup dedicated line activation, analyze path adjustment data through the CNN model, accurately predict whether the backup dedicated line needs to be activated, avoid unnecessary activation or delayed activation, and improve network stability and business assurance capabilities.

[0049] The present invention proposes a method for dynamically generating a route for virtual dedicated line allocation, comprising the following steps:

[0050] In the set time window, multiple preset traffic indicators of the current network are collected and the traffic load index is calculated, and then the traffic environment type is divided according to the traffic load index; traditional network routing adjustment methods usually adopt static or simple threshold strategies, which easily lead to long-term idleness or frequent switching of backup dedicated lines, reducing network resource utilization. This method divides the network traffic environment (low, medium, and high) according to FLI, and uses intelligent analysis to decide whether to adjust the path or enable the backup dedicated line, effectively optimizing resource scheduling, maximizing the use of existing paths when the network load is moderate, and avoiding unnecessary computing overhead.

[0051] Determine whether to perform path adjustment analysis according to the traffic environment type obtained by partitioning. When performing path adjustment analysis, calculate the time overhead value used to evaluate the time required for recalculating the route and the path calculation load value used to evaluate the computing power of the current network controller respectively. When performing path recalculation, if the calculation cost is not reasonably evaluated, it may cause the network controller to be overloaded, affect other services, and even cause path adjustment delays. This method evaluates the time cost of path recalculation by calculating the time overhead value (PAO), and combines the path calculation load value (PCL) to judge the computing power of the controller, ensuring that adjustments are made under acceptable computing loads to avoid affecting service stability due to overloaded computing tasks of the controller.

[0052] Perform corresponding preset policy adjustments according to the traffic environment type obtained by partitioning, and at the same time determine whether to directly enable the standby dedicated line according to the result of path adjustment analysis. The activation of traditional standby dedicated lines is usually triggered based on a fixed threshold. For example, when the network bandwidth utilization rate exceeds a certain threshold, the standby dedicated line is directly enabled, which may lead to: premature, resulting in waste of main path resources; delayed switching, leading to increased network congestion during traffic peaks. This method uses a CNN (Convolutional Neural Network) model to intelligently determine whether to enable the standby dedicated line according to the calculation results of PAO and PCL, avoiding misjudgments caused by unreasonable static threshold settings and improving the utilization efficiency of the standby dedicated line. The traffic in modern network environments is dynamic and uncertain, and burst traffic or traffic fluctuations may affect the optimal selection of routes. This method combines the real-time calculation of FLI to achieve adaptive classification of the network environment, and performs refined management by combining path adjustment analysis in medium traffic environments, ensuring that network adjustment strategies adapt to different service requirements and improving the overall network efficiency.

[0053] Dividing the traffic environment type according to the traffic load index means that:

[0054] Low traffic environment: FLI ≤ T1, the network load is light; Medium traffic environment: T1 < FLI ≤ T2, the network load is moderate; High traffic environment: FLI > T2, there is a congestion risk. FLI represents the traffic load index, and T1 and T2 are the preset partition thresholds one and two. And in the low traffic environment, the standby dedicated line is not enabled. In the high traffic environment, the standby dedicated line is directly enabled. In the medium traffic environment, path adjustment analysis is performed in combination with the historical virtual dedicated line allocation scheme given by the preset path adjustment model. For example, in the virtual dedicated line allocation scheme given by the preset path adjustment model in historical situations, the average change ratio of the route table entries affected by the adjustment can be estimated based on the statistics of historical data. By analyzing past path adjustment records, it can be calculated that in a similar traffic environment (FLI is similar), the average value of the change ratios of all route table entries can be taken with an upper and lower deviation of no more than five percent. When the traffic environment type obtained by partitioning is a medium traffic environment, path adjustment analysis is performed.

[0055] The logic for obtaining the traffic load index is:

[0056] Get the average bandwidth utilization, average packet loss rate, number of active connections corresponding to the 90% quantile, average traffic change rate, and delay jitter obtained by calculating the transmission delay change between consecutive data packets using the root mean square calculation formula in the current time window;

[0057] The flow load index calculation formula is:

[0058]

[0059] P represents the average bandwidth utilization of the current time window, J represents delay jitter, R represents the average packet loss rate, A represents the number of active connections corresponding to the 90% percentile, F represents the average traffic change rate, C represents a preset constant and is a positive value, w1, w2, w3, w4, and w5 are all preset proportional coefficients, and FLI represents the traffic load index. P is the average bandwidth utilization of the current time window, reflecting the bandwidth occupancy of the network. J is delay jitter, which measures the stability of the network by calculating the transmission delay variation between consecutive data packets through the root mean square. R is the average packet loss rate, which represents the proportion of data packets lost in the time window and reflects the reliability of the network. A is the number of active connections corresponding to the 90% percentile, reflecting the concurrent connections under high traffic conditions. F is the average traffic change rate, which is used to measure the burstiness and fluctuation of traffic. C is a preset constant value, which is a positive value and is used to adjust the baseline value of FLI to ensure its interpretability and applicability to different network environments. The average bandwidth utilization P directly affects the FLI. The linear weighting w1·P represents the direct impact of bandwidth utilization on the network load. The higher the bandwidth usage, the greater the FLI. The delay jitter J uses logarithmic transformation to reduce the impact of extreme values ​​and avoid excessive growth of FLI caused by a small amount of abnormal jitter. When the jitter is small, the impact of this item is small; when the jitter is large, the impact of this item gradually increases, reflecting the importance of network stability. The average packet loss rate R directly affects the network load. The linear weighting w3·R indicates that the higher the average packet loss rate, the more congested or serious the network is, and the larger the FLI is. The square root transformation can smooth the growth trend of the number of active connections and avoid excessive expansion of the index when the number of connections is too large. The 90% quantile is used to capture the concurrent connections at peak times, which can better reflect the network pressure during burst loads than the simple average value. The average traffic change rate F uses linear weighting w5·F to reflect the burstiness and change trend of traffic. A high traffic change rate means that the network fluctuates greatly and the load management is more difficult. The impact weight of this item can be adjusted according to different network environments.

[0060] The logic for obtaining the time cost value is:

[0061] Get the number of nodes N in the current network topology topo , which indicates the number of network devices involved in routing calculation, such as switches, routers, etc. The available computing power of the current routing calculation device C proc ,, reflects the computing resources of the controller or router. The convergence time of the current network routing protocol T conv , which indicates the time required for OSPF, BGP and other protocols to reach a stable state after a topology change. Historical route adjustment frequency H freq , represents the frequency of network adjustment in the past. The higher the frequency, the greater the dynamic change of the network. In the virtual dedicated line allocation scheme given by the preset path adjustment model under historical circumstances, the average change ratio of the routing table items affected by the adjustment is R change , used to evaluate the impact range of the adjustment. Baseline routing table change threshold R th , used to compare whether the current change exceeds the threshold. The calculation formula for the time cost value is:

[0062] PAO represents the time cost value, which indicates the time cost required for path calculation. The larger the value, the greater the calculation cost. α and β both represent the preset weight coefficients. Reflects the computing load pressure: If the topology is large (many nodes) but the computing resources are limited (low processing power), the computing pressure is high and PAO increases; if the computing power is strong enough (such as a high-performance SDN controller), the computing pressure is small and PAO decreases. The exponential weight α makes the computing overhead grow nonlinearly when the topology changes drastically, enhancing the impact of computing pressure. If the convergence time is long, the PAO increases, avoiding path adjustments when the protocol has not converged, reducing the risk of routing instability; using the logarithmic function ln(1+T conv ) Limit the scope of influence to prevent the influence on computing pressure from increasing infinitely when the convergence time is too long. If adjustments are made frequently in the past, it means that the network environment is unstable and PAO is rising, so reduce the adjustment frequency; use square root Processing to reduce the inflation effect of extreme high frequency adjustments on the calculated pressure. Use Sigmoid function To smooth the impact, ensure that the impact is small when the change ratio is small, but the PAO rises rapidly when the change ratio is large; this can avoid excessive computational pressure when adjusting small-scale paths, while ensuring that the computational overhead is not ignored when adjusting large-scale paths.

[0063] The logic for obtaining the path calculation load value is as follows:

[0064] Get the current path calculation task queue length, that is, the number of path tasks waiting to be calculated M queue , reflects the queuing status of computing tasks. The number of path computing tasks currently being executed in parallel is N exec, which indicates the real-time occupancy of computing resources. The average waiting time of the path calculation task is R wait , which represents the time from when a task enters the queue to when it is executed, and measures the degree of task backlog. The average completion time of path calculation is T comp , reflects the computation time of a single task. The number of path computation failures in the past preset fixed time period Q failure , which measures the computing stability and task execution success rate. The total number of path computing requests in the past preset fixed time is Q total , the normalized benchmark of the calculation failure rate. The number of active computing tasks of the current computing controller W active , that is, all the computing tasks being executed, reflecting the overall load of the computing controller. The calculation formula for the path calculation load value is:

[0065] PCL represents the path calculation load value, which is used to evaluate the calculation pressure of the current network controller. The larger the value, the higher the calculation pressure. The number of waiting tasks in the path calculation task queue reflects the backlog of calculation tasks: if there are many tasks waiting for calculation, but few tasks are being executed, the PCL value increases, indicating that the calculation pressure is high; if there are few calculation tasks or a large number of tasks are being executed, the calculation pressure is small. The square processing makes the impact of task backlog grow nonlinearly to prevent the queuing impact of a small number of tasks from being underestimated. Reflects the ratio of task waiting time to calculation completion time: if the waiting time is much longer than the calculation completion time, it means that the task stays in the calculation queue for too long, the calculation load is heavy, and PCL increases; if the calculation completion time is longer than or close to the waiting time, it means that the computing resources are sufficient, the task can be completed quickly, and the PCL impact is reduced. The calculation failure rate measures the stability of the system calculation. If the failure rate is high, it means that the calculation task may fail due to resource exhaustion, timeout and other problems, and PCL increases; if the failure rate is low, the impact of this item is small. The impact of using sine function to smooth failure on PCL: When the failure rate is low, this item contributes less, but when the failure rate is high, this item increases rapidly, preventing the low failure rate from causing unnecessary interference to PCL. log2(W active +1) reflects that if there are many active computing tasks, the overall controller load is high and PCL increases; if there are few active computing tasks, it means that there are still spare computing resources and PCL decreases. The logarithmic function can smooth the growth of active tasks and avoid the PCL rising too fast due to the increase of short-term tasks; it ensures that the impact is small when there are low active tasks, but the impact is significant when the task load is high.

[0066] Deciding whether to directly enable the backup dedicated line based on the result of the path adjustment analysis means: substituting the result of the path adjustment analysis, i.e., the path calculation load value and the time overhead value, into the pre-trained convolutional neural network model. If the output result of the convolutional neural network model is 1, the backup dedicated line is directly enabled. If the output result of the convolutional neural network model is 0, the virtual dedicated line allocation plan obtained by the preset path adjustment model is executed.

[0067] The input of CNN includes: Path Computation Load Value (PCL) → reflects the load of the current computing controller; Time Overhead Value (PAO) → reflects the computing overhead that may be generated by the current path adjustment; CNN output = 1 (high computing load or high time overhead) → directly enable the backup dedicated line to ensure business continuity; CNN output = 0 (acceptable computing load and time overhead) → execute the preset path adjustment model to optimize the existing path. CNN learns the data features of PCL and PAO through multiple convolutional layers, pooling layers, and fully connected layers, and finally outputs whether the backup dedicated line needs to be enabled: Feature extraction → The convolution layer extracts the timing patterns of PCL and PAO. Feature induction → The pooling layer reduces the data dimension and extracts the most critical traffic trends and computing pressure changes. Fully connected layer decision → Based on historical data and current input, a decision is made on whether to enable the backup dedicated line.

[0068] It should be noted that the preset path adjustment model refers to a path adjustment strategy that is pre-formulated based on traffic load conditions, historical adjustment experience, and path optimization algorithms when the network environment is known. It is used to determine how to allocate virtual private line resources in different situations. When CNN predicts that there is no need to directly enable the backup private line (output = 0), it is still necessary to adjust the path based on the current network situation to optimize traffic distribution and ensure network stability. The model can dynamically adjust the traffic allocation strategy of the virtual private line to achieve more flexible path management. Possible path adjustment strategies: Path adjustment based on load balancing Example: If the current traffic on a private line is large, and other private lines still have available bandwidth, some traffic is reallocated to the low-load path to reduce congestion on the main path. Path adjustment based on latency optimization: Example: If the latency of the current path is too high, adjust the traffic priority strategy to ensure that latency-sensitive services (such as video conferencing, VoIP) use low-latency paths, while low-priority services (such as file downloads) use other paths. MPLS-TE-based path optimization: Example: In an MPLS network, a new forwarding path can be selected based on the Constrained Shortest Path First (CSPF) algorithm, rather than directly.

[0069] The corresponding preset policy adjustment according to the divided traffic environment types means that the traffic environment types are divided into low traffic (FLI ≤ T1), medium traffic (T1 < FLI ≤ T2), and high traffic (FLI > T2), and different types adopt different adjustment strategies. Policy adjustment for low traffic environment (FLI ≤ T1): No standby dedicated line is required. The main path is preferentially used, and the current traffic distribution is optimized. Example: If there are multiple paths, the shortest path is preferred. If the current bandwidth utilization rate is low, QoS control can be reduced to reduce the computing load. Policy adjustment for medium traffic environment (T1 < FLI ≤ T2): Instead of directly enabling the standby dedicated line, a preset path adjustment model is used to optimize the current path. Example: Dynamic bandwidth adjustment: Based on traffic prediction, the bandwidth allocation policy is adjusted to avoid network congestion caused by upcoming peak periods. Policy adjustment for high traffic environment (FLI > T2): The standby dedicated line is directly enabled to reduce the pressure on the main path. Example: Trigger fast path switching, migrate the core service traffic to the standby dedicated line, and enable the standby VPN channel to ensure the service quality of critical services. The preset path adjustment model provides a method to optimize the current path allocation without directly enabling the standby dedicated line, such as strategies like load balancing, delay optimization, and traffic diversion. The preset policy adjustment according to the traffic environment type ensures that the optimal adjustment plan is adopted under low, medium, and high load conditions, which not only ensures the reasonable utilization of network resources but also avoids unnecessary dedicated line switching.

[0070] Embodiment 2: A routing dynamic generation system for virtual dedicated line allocation, comprising:

[0071] A data collection module that collects multiple preset traffic metrics of the current network within a set time window;

[0072] A data analysis module that calculates the traffic load index based on the data collected by the data collection module, and then divides the traffic environment type according to the traffic load index;

[0073] A path adjustment analysis module that determines whether to perform path adjustment analysis according to the divided traffic environment type, and when performing path adjustment analysis, calculates the time overhead value for evaluating the time required to recalculate the route and the path calculation load value for evaluating the computing power of the current network controller respectively;

[0074] A dynamic adjustment module that performs corresponding preset policy adjustment according to the divided traffic environment type, and at the same time determines whether to directly enable the standby dedicated line according to the result of the path adjustment analysis.

[0075] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0076] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0077] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0078] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0079] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for dynamically generating routes for virtual private line allocation, characterized in that: It includes the following steps: Within a set time window, collect multiple preset traffic metrics of the current network and calculate the traffic load index, and then divide the traffic environment type based on the traffic load index; Decide whether to perform path adjustment analysis according to the divided traffic environment type. And during path adjustment analysis, calculate the time overhead value for evaluating the time required to recalculate the route and the path calculation load value for evaluating the computing power of the current network controller respectively; Perform corresponding preset policy adjustments according to the divided traffic environment type, and at the same time decide whether to directly enable the standby dedicated line according to the result of the path adjustment analysis.

2. A method for dynamically generating routes for virtual private line allocation according to claim 1, characterized in that: Dividing the traffic environment type based on the traffic load index means: Low traffic environment: FLI ≤ T1; Medium traffic environment: T1 < FLI ≤ T2; High traffic environment: FLI > T2, where FLI represents the traffic load index, and T1 and T2 are the preset division threshold one and division threshold two. And in the low traffic environment, the standby dedicated line is not enabled. In the high traffic environment, the standby dedicated line is directly enabled. In the medium traffic environment, path adjustment analysis is carried out in combination with the historical virtual dedicated line allocation scheme given by the preset path adjustment model.

3. A method for dynamically generating routes for virtual private line allocation according to claim 2, characterized in that: When the divided traffic environment type is the medium traffic environment, perform path adjustment analysis.

4. A method for dynamically generating routes for virtual private line allocation according to claim 3, characterized in that: The acquisition logic of the traffic load index is: Obtain the average bandwidth utilization rate, average packet loss rate, the number of active connections corresponding to the 90% quantile, average traffic change rate of the current time window, and the delay jitter obtained by calculating the degree of change in the transmission delay between consecutive data packets through the root mean square calculation formula; The calculation formula of the traffic load index is: P represents the average bandwidth utilization rate of the current time window, J represents the delay jitter, R represents the average packet loss rate, A represents the number of active connections corresponding to the 90% quantile, F represents the average traffic change rate, C represents a preset constant and is a positive value, w1, w2, w3, w4, w5 are all preset proportionality coefficients, and FLI represents the traffic load index.

5. A method for dynamically generating routes for virtual private line allocation according to claim 4, characterized in that: The acquisition logic of the time overhead value is: Get the number of nodes N in the current network topology topo , the available computing power C of the current routing computing device proc , the convergence time T of the current network routing protocol conv , historical routing adjustment frequency H freq , the average change ratio R of the routing table items affected by the preset path adjustment model in the virtual dedicated line allocation scheme given in the historical situation change , the baseline routing table change threshold R th , the calculation formula of time cost value is: PAO represents the time overhead value, and α and β represent preset weight coefficients.

6. A method for dynamically generating routes for virtual private line allocation according to claim 5, characterized in that: The acquisition logic of the path calculation load value is: Get the current path calculation task queue length, that is, the number of path tasks waiting to be calculated M queue , the number of path calculation tasks currently being executed in parallel N exec , the average waiting time of the path calculation task is R wait , the average completion time of path calculation is T comp , the number of path calculation failures in the past preset fixed time period Q failure , the total number of path calculation requests in the past preset fixed time Q total , the number of active computing tasks W of the current computing controller active , the calculation formula for path calculation load value is: PCL indicates the path calculation load value.

7. A method for dynamically generating routes for virtual private line allocation according to claim 6, characterized in that: Deciding whether to directly enable the standby dedicated line according to the result of the path adjustment analysis means: Substitute the results of the path adjustment analysis, that is, the path calculation load value and the time overhead value, into the pre-trained convolutional neural network model. If the output result of the convolutional neural network model is 1, directly enable the standby dedicated line. If the output result of the convolutional neural network model is 0, execute the virtual dedicated line allocation scheme obtained by the preset path adjustment model.

8. A system for dynamically generating routes for virtual dedicated line allocation, used to implement a method for dynamically generating routes for virtual dedicated line allocation as claimed in any one of claims 1 to 7, characterized in that: It includes: A data acquisition module that collects multiple preset traffic metrics of the current network within a set time window; A data analysis module that calculates the traffic load index based on the data collected by the data acquisition module, and then divides the traffic environment type according to the traffic load index; A path adjustment analysis module that decides whether to perform path adjustment analysis according to the divided traffic environment type, and during path adjustment analysis, calculates the time overhead value for evaluating the time required to recalculate the route and the path calculation load value for evaluating the computing power of the current network controller respectively; The dynamic adjustment module adjusts the corresponding preset strategy according to the traffic environment type obtained by classification, and decides whether to directly enable the backup dedicated line based on the results of the path adjustment analysis.

Images