Router isolation management method supporting multi-tenant virtual network

Abstract

The invention relates to the technical field of network interconnection, in particular to a router isolation management method for supporting a multi-tenant virtual network, which comprises the following steps of: collecting real-time flow data of each tenant network, comparing the flow data with a performance baseline set by each tenant to obtain a flow anomaly index, and according to the flow anomaly index, determining whether the performance baseline is abnormal or not; and identifying a fault area in the network through an algorithm, and generating fault positioning information. According to the method, the network traffic data of each tenant is collected and analyzed in real time, and the difference between the network traffic data and the established performance base line is compared, so that traffic anomaly detection is realized. Through real-time monitoring and data comparison, a flow mode deviating from a normal range can be rapidly identified, and a fault area in a network can be automatically identified through an algorithm. Besides, the applicability of different recovery strategies is analyzed by applying a decision tree technology, the recovery strategy suitable for the current network condition can be selected from multiple options, and the recovery strategy is automatically applied to router configuration.

Description

Technical Field

[0001] The present invention relates to the technical field of network interconnection, and in particular to a router isolation management method supporting a multi-tenant virtual network. Background Art

[0002] The field of network interconnection technology refers to the technologies and methods for designing and implementing network devices and systems to achieve data exchange between multiple networks. This field includes a wide range of technologies, such as routing, gateway services, bridging, and switching technologies. Router isolation management methods are a specific topic in the field of network interconnection, focusing on how to effectively manage and isolate network traffic of different tenants in a multi-tenant environment.

[0003] However, existing network interconnection technologies often lack sufficient automation and real-time data processing capabilities, and rely on manual intervention to identify and solve network problems, which is inefficient when dealing with complex multi-tenant environments. Due to the lack of real-time monitoring and automatic fault location systems, it is often impossible to quickly and accurately locate the source of the problem after a fault occurs, delaying the time to handle the fault. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a router isolation management method supporting a multi-tenant virtual network.

[0005] In order to achieve the above object, the present invention adopts the following technical solution to support a router isolation management method for a multi-tenant virtual network, comprising the following steps:

[0006] Collect the real-time traffic data of each tenant network, compare the traffic data with the performance baseline set by each tenant, obtain traffic anomaly indicators, identify the fault area in the network through algorithms based on the traffic anomaly indicators, and generate fault location information;

[0007] Based on the fault location information, use a decision tree to analyze the applicability of each recovery strategy, compare the degree of impact of each recovery strategy on the network, select a recovery strategy, generate an optimal recovery strategy, and apply the optimal recovery strategy to the router configuration;

[0008] Monitor the implementation effect of the preferred recovery strategy, collect the network performance data after recovery in real time, compare it with the data before recovery, and generate a performance recovery evaluation value;

[0009] Based on the performance recovery evaluation value, the isolation measures between tenants are refined to generate isolation and traffic management results.

[0010] Preferably, the steps of obtaining the traffic anomaly indicator are:

[0011] Collect the real-time traffic data of each tenant's network, compare it with the performance baseline set by each tenant, obtain the deviation value of each data point, and generate a deviation data set;

[0012] Based on the deviation data set, the ratio of the sum of the absolute values ​​of the deviations of each data point to the total number of data points is calculated, and with reference to the data volatility, the calculation formula is:

[0013]

[0014] Where E represents the weighted mean deviation, d i represents the flow value of the ith data point, b represents the baseline flow value, N represents the total number of data points, λ is the adjustment coefficient, and μ is the average flow value of all data points;

[0015] Based on the weighted average deviation, an abnormal threshold is set, and the deviation value exceeding the abnormal threshold is marked as abnormal, so as to obtain a flow abnormality index.

[0016] Preferably, the steps of acquiring the fault location information are:

[0017] Receiving the traffic anomaly indicator, identifying data points of traffic anomaly, and obtaining an abnormal data set;

[0018] Based on the abnormal data set, the similarity between each data point is calculated, and the calculation formula is:

[0019]

[0020] Among them, S represents the similarity score, x i and x j are two abnormal data points, σ ​​is the standard deviation, and M is the number of abnormal data points;

[0021] Based on the similarity score, the fault area in the network is determined to obtain fault location information.

[0022] Preferably, the steps for obtaining the preferred recovery strategy are:

[0023] Collecting the fault location information, applying a decision tree model to analyze the applicability of each network recovery strategy, and obtaining a recovery strategy applicability score;

[0024] Based on the applicability score of the recovery strategy, the comprehensive impact of each strategy is calculated using the following formula:

[0025]

[0026] Among them, I represents the comprehensive influence of the strategy, w k is the weight of the k-th strategy, r k Score the suitability of the k-th strategy, Tk is the theoretical optimal performance value of the kth strategy, and K is the total number of strategies;

[0027] Based on the comprehensive impact, various restoration strategies are compared and selected to obtain an optimal restoration strategy.

[0028] Preferably, the steps of applying the preferred recovery strategy to the router configuration are:

[0029] Based on the preferred recovery strategy, adjusting routing protocols, interface settings and security policies;

[0030] Update router settings and monitor response and error logs during configuration.

[0031] Preferably, the steps for obtaining the performance recovery evaluation value are:

[0032] Collecting the delay, bandwidth utilization and packet loss rate after implementing the preferred recovery strategy to obtain a performance data set after recovery;

[0033] Based on the performance data set after recovery, the difference between the performance data set after recovery and the performance data set before recovery is calculated using the following formula:

[0034]

[0035] Among them, P represents the average performance improvement value, p post,i is the value of the i-th performance indicator after recovery, p pre,i is the value of the i-th performance indicator before recovery, ZN is the total number of performance indicators;

[0036] Based on the average performance improvement value, the recovery effect of the network is evaluated to obtain a performance recovery evaluation value.

[0037] Preferably, the steps of obtaining the isolation and traffic management results are:

[0038] Based on the performance recovery evaluation value, the performance impact between tenants is analyzed, the affected tenants are identified, and the isolation measures that need to be refined are determined.

[0039] Preferably, the step of obtaining the isolation and traffic management results also includes: configuring traffic redirection and traffic restriction settings according to the detailed isolation measures to generate the isolation and traffic management results.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are:

[0041] The present invention realizes the detection of traffic anomalies by collecting and analyzing the network traffic data of each tenant in real time and comparing it with the established performance baseline. This real-time monitoring and data comparison makes it possible to quickly identify traffic patterns that deviate from the normal range and automatically identify faulty areas in the network through algorithms. In addition, by using decision tree technology to analyze the applicability of different recovery strategies, a recovery strategy suitable for the current network conditions can be selected from multiple options and automatically applied to the router configuration. The effect of the implemented recovery strategy is monitored and compared with previous data to further generate a performance recovery evaluation value. This method enhances the transparency and traceability of the network recovery process, optimizes resource allocation, and improves the overall reliability and service quality of the network. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the steps of the present invention;

[0043] Figure 2 Schematic diagram of the steps for obtaining the optimal recovery strategy in the present invention;

[0044] Figure 3 It is a schematic diagram of the steps for obtaining the isolation and traffic management results in the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0046] See also Figure 1 The present invention provides a technical solution to support a router isolation management method for a multi-tenant virtual network, comprising the following steps:

[0047] Collect the real-time traffic data of each tenant's network, compare the traffic data with the performance baseline set by each tenant, obtain the traffic anomaly index, and use the algorithm to identify the fault area in the network based on the traffic anomaly index to generate fault location information;

[0048] Based on the fault location information, the decision tree is used to analyze the applicability of each recovery strategy, the impact of each recovery strategy on the network is compared, a recovery strategy is selected, an optimal recovery strategy is generated, and the optimal recovery strategy is applied to the router configuration;

[0049] Monitor the implementation effect of the optimal recovery strategy, collect network performance data after recovery in real time, compare it with the data before recovery, and generate performance recovery evaluation values;

[0050] Based on the performance recovery evaluation value, refine the isolation measures between tenants and generate isolation and traffic management results.

[0051] The steps to obtain traffic anomaly indicators are as follows:

[0052] Collect the real-time traffic data of each tenant's network, compare it with the performance baseline set by each tenant, obtain the deviation value of each data point, and generate a deviation data set;

[0053] Based on the deviation data set, the ratio of the sum of the absolute values ​​of the deviation of each data point to the total number of data points is calculated, and with reference to the data volatility, the calculation formula is:

[0054]

[0055] Where E represents the weighted mean deviation, d i represents the flow value of the ith data point, b represents the baseline flow value, N represents the total number of data points, λ is the adjustment coefficient, and μ is the average flow value of all data points;

[0056] Based on the weighted average deviation, an abnormal threshold is set, and the deviation value exceeding the abnormal threshold is marked as abnormal to obtain the traffic abnormality index.

[0057] Specifically, the real-time traffic data of the tenant network is recorded, the traffic values ​​are extracted at fixed intervals, and each extracted value is numerically compared with a predetermined performance baseline, wherein the performance baseline is obtained by statistically analyzing the traffic monitoring data of the same tenant network for several days in the early stage, selecting a time interval with relatively stable traffic, adding all the traffic data in the interval and calculating the average value, and further performing a difference operation between the traffic value of each time period and the obtained baseline traffic value, so as to obtain the deviation value of each data point, and store these deviation values ​​item by item in the data recording file, and then perform a data cleaning process on all the deviation values ​​in the file, eliminate extreme values ​​caused by recording equipment failure or line abnormality, and eliminate records that exceed the set range after manual inspection. The set range refers to the upper and lower limits of the traffic of the same tenant in a stable business state in the past, and the upper and lower limits are obtained through statistical analysis of the early monitoring data, so as to ensure the availability and continuity of the remaining deviation values, and then mark the cleaned deviation value sequence and store it in the deviation data set file.

[0058] The benefit of the formula is that it combines the overall traffic deviation and the degree of traffic dispersion by considering both the absolute deviation and the fluctuation amplitude, so that the evaluation results can more accurately reflect the degree of deviation of network traffic from the baseline.

[0059] The steps for obtaining the N parameter are as follows: by recording the traffic of the designated tenant network every 5 minutes within 24 hours, a total of 288 data points are obtained, and N=288 is obtained after counting these 288 data points;

[0060] di The steps to obtain the parameters are as follows: record the actual flow value at each time point at an interval of 5 minutes, and obtain all d i ;

[0061] The steps for obtaining the b parameter are as follows: based on the continuous monitoring of the tenant network for several days (e.g., 7 days) in the early stage, a time period with stable services and no abnormal fluctuations is selected, and the traffic data in the time period is averaged to obtain the baseline value b. Based on the analysis of multiple days of data, the average value of the stable interval is selected as b=500 Mbit / s;

[0062] The steps to obtain the μ parameter are as follows: i Find the arithmetic mean to get μ, for example, count all d on that day i The data is summed and divided by 288 to obtain μ = 513 Mbit / s;

[0063] The steps for obtaining the λ parameter are as follows: by calculating the standard deviation of the historical data of the previous period (such as 7 days), the standard deviation is compared with the baseline value to obtain the fluctuation adjustment coefficient λ. For example, the standard deviation of the historical data of 7 days is 40 Mbit / s, and the ratio of the standard deviation to the baseline value of 500 Mbit / s is used to determine λ=0.08;

[0064] Calculation process:

[0065] Take 5 data points as an example (288 points are used in the actual calculation):

[0066] Given d1 = 510 Mbit / s, d2 = 520 Mbit / s, d3 = 495 Mbit / s, d4 = 550 Mbit / s, d5 = 490 Mbit / s;

[0067] μ=(510+520+495+550+490) / 5=2565 / 5=513Mbit / s;

[0068] Σ|d i -b|=|510-500|+|520-500|+|495-500|+|550-500|+

[0069] 490-500|=10+20+5+50+10=95Mbit / s;

[0070] ∑(d i -μ) 2 =(510-513) 2 +(520-513) 2 +(495-513) 2 +(550-513) 2+(490-513) 2 =9+49+324+1369+529=2280;

[0071]

[0072] Substitute λ = 0.08 and calculate:

[0073]

[0074] The result shows that the weighted average deviation of the current traffic relative to the baseline is about 19.764Mbit / s. When the result is compared with the set abnormal threshold, if the value is greater than the abnormal threshold, it means that the traffic deviation during this period exceeds the normal fluctuation range.

[0075] Based on the weighted average deviation value obtained previously, the deviation value sequence in the monitoring period is checked, and an abnormal threshold is selected. The threshold is obtained by analyzing the data without special business peaks and abnormal situations in the previous period, and the weighted average deviation values ​​of the same time period every day are counted and the average value is calculated. Then, an upper and lower limit range is established based on the statistical results, and the threshold is set at the average value plus a reference ratio. The reference ratio is derived from the normal volatility obtained from the historical records of similar businesses in the previous period. For example, the average deviation value calculated from the data of the same time period in the past 7 days is 20Mbit / s, and the standard deviation is 5Mbit / s. The abnormal threshold is set to the average value plus 2 times the standard deviation, that is, 20+2×5=30Mbit / s. The deviation value of each current period is compared with 30Mbit / s item by item. The deviation value exceeding 30Mbit / s is marked as abnormal in a data marking manner, so that these marked deviation values ​​are summarized in the data record file to form an abnormal deviation data list, and a statistical count is added at the end of the list. The abnormal count and the timestamp are stored in correspondence.

[0076] The steps to obtain fault location information are as follows:

[0077] Receive traffic anomaly indicators, identify data points of traffic anomaly, and obtain anomaly data sets;

[0078] Based on the abnormal data set, the similarity between each data point is calculated using the following formula:

[0079]

[0080] Among them, S represents the similarity score, x i and x j are two abnormal data points, σ ​​is the standard deviation, and M is the number of abnormal data points;

[0081] Based on the similarity score, the fault area in the network is determined and the fault location information is obtained.

[0082] Specifically, the received traffic anomaly index is compared and screened for multiple data points marked in the traffic data record file. In order to identify the degree of numerical anomaly in each data point, it is necessary to first read the traffic value corresponding to each data point in a fixed time interval, and load these traffic values ​​from the database file into the memory one by one. According to the upper and lower limits of the traffic obtained through long-term monitoring in the early stage, the records exceeding the upper and lower limits are eliminated. The upper and lower limits are obtained by analyzing the average and discrete degree of the traffic in the same period of multiple consecutive days in the past. The deviations of these values ​​are calculated one by one and the deviation values ​​are recorded in the deviation data list. Then, the deviation data list is compared, and the deviation points with values ​​greater than the specific upper and lower limit parameters determined by historical records are marked as abnormal data points. The upper and lower limit parameters are obtained by counting the deviation mean and standard deviation at the same time point in the past 7 days in a period without special business requests, and setting the upper and lower limit parameters to the mean plus or minus twice the standard deviation. Then, an abnormal data set is generated based on these marked abnormal data points.

[0083] The benefit of the formula is that it quantifies the differences between abnormal data points into similarity scores by utilizing the Gaussian kernel function form and considers the distance distribution between data points to achieve more representative feature associations extracted from the abnormal data set.

[0084] The steps for obtaining the M parameter are as follows: after monitoring the flow data point by point within a 24-hour period, 300 data points are recorded, and then 5 data points are judged as abnormal data points through the aforementioned abnormal indicators, then M=5;

[0085] x i and x j The parameter acquisition steps are as follows: select two abnormal data points from the aforementioned abnormal data set, and the flow value of each data point comes from the data file after cleaning and screening. For example, through continuous 24-hour monitoring, the flow values ​​corresponding to the abnormal points are read as x1=520Mbit / s, x2=530Mbit / s, x3=515Mbit / s, x4=540Mbit / s, x5=525Mbit / s, and x is selected. j =530Mbit / s as a reference point;

[0086] The steps for obtaining the σ parameter are as follows: calculate the standard deviation of the traffic values ​​of the acquired abnormal data points, and set the mean of the five abnormal points to (520+530+515+540+525) / 5=526Mbit / s, and the variance is ((520-526) 2 +(530-526) 2 +(515-526) 2 +(540-526) 2+(525-526) 2 ) / 5=36+16+121+196+1) / 5=370 / 5=74, then

[0087] Calculation process:

[0088] For each x i With x j =530Mbit / s, square the difference and substitute it into the standard deviation:

[0089] (x1-x j ) 2 =(520-530) 2 =100

[0090] (x2-x j ) 2 =(530-530) 2 =0

[0091] (x3-x j ) 2 =(515-530) 2 =225

[0092] (x4-x j ) 2 =(540-530) 2 =100

[0093] (x5-x j ) 2 =9525-530) 2 =25

[0094] σ 2 =74Substitute into:

[0095]

[0096] exp(-100 / 148)≈0.508, exp(0)=1, exp(-225 / 148)≈0.218, exp(-100 /

[0097] 148)≈0.508, exp(-25 / 148)≈0.845

[0098] The sum is 0.508+1+0.218+0.508+0.845=3.079

[0099]

[0100] The result shows that the similarity score between the two abnormal data points is about 0.616. When the S value is high, it means that the numerical proximity of the two points is high. When the S value is low, it means that the numerical proximity of the two points is low. By comparing the similarity scores between multiple abnormal data points, the abnormal points can be further aggregated and analyzed to assist in locating the network fault area based on the similarity results.

[0101] Based on the calculated similarity score, the data points marked as abnormal are associated. The similarity score of each abnormal point needs to be recorded in an independent file, and the similarity value is extracted from the file one by one, and compared with a threshold obtained by historical statistics. The threshold is obtained by comparing multiple groups of abnormal points in a long period of time in the previous period. By counting the mean and mean square error of the similarity of abnormal points in the same monitoring area in the past 30 days, the threshold is set as the mean minus twice the mean square error, and the point pairs with similarity lower than the threshold are marked. Then, the abnormal records with adjacent spatial positions are distinguished from the marked point pairs. The abnormal points are grouped according to these positional relationships. For multiple point pairs with similarity lower than the threshold in the same group, they are arranged in sequence to form a regional mapping table. By locating the adjacent coordinate information recorded in the mapping table, the corresponding path can be found in the network topology index. According to the path index value, the node position records pre-stored in the topology database file are consulted, and the coordinate values ​​and node identification codes of these nodes are stored in the position record list. Finally, the result information is output one by one from the record list and written into the fault location data file to obtain the fault location information.

[0102] The steps for obtaining the optimal recovery strategy are:

[0103] Collect fault location information, apply the decision tree model to analyze the applicability of each network recovery strategy, and obtain the recovery strategy applicability score;

[0104] Based on the recovery strategy applicability score, the comprehensive impact of each strategy is calculated using the following formula:

[0105]

[0106] Among them, I represents the comprehensive influence of the strategy, w k is the weight of the k-th strategy, r k Score the suitability of the k-th strategy, T k is the theoretical optimal performance value of the kth strategy, and K is the total number of strategies;

[0107] Based on the comprehensive impact, each recovery strategy is compared and selected to obtain the optimal recovery strategy.

[0108] Specifically, after collecting the fault location information, the extracted multiple groups of records are classified and compared, and the coordinate values ​​of the location of each network node, the node name identification and the corresponding time period of the fault in the record file are read, and this information is stored in independent data fields. Then a decision tree parameter configuration file prepared in advance is selected. The file contains multiple numerical thresholds and classification labels for decision tree branch judgment. Each threshold value is determined by statistically comparing various fault conditions in the past 30 days to find the key point data for distinguishing normal strategies from abnormal strategies. These data points are summarized from multiple monitoring logs, and the source and calculation method of each threshold value are recorded in the internal document. The corresponding label value is obtained from the historical classification records. Each extracted record item is matched with the decision tree branch condition. When the classification label corresponding to the node feature value reflected in the record item matches the numerical threshold given by the branch condition, the record item is recursively judged downward along the decision tree path. If the coordinate value difference in the record item exceeds the specific upper and lower limits extracted from past statistics, the corresponding upper and lower limits are obtained by adding and subtracting the average coordinate deviation value and standard deviation of several working days, then the corresponding branch processing step is entered downward, and finally a numerical score is output at the terminal node of the decision tree. The corresponding scores of each record item are integrated into a list and stored in the final result file to obtain the recovery strategy applicability score.

[0109] The benefit of the formula is that through comprehensive consideration of weights, applicability scores and theoretical optimal performance values, the final calculated comprehensive impact can better reflect the actual impact of each strategy in a real environment.

[0110] The steps of obtaining the K parameter are: obtaining the parameter by summarizing the number of all candidate strategies enabled in the network recovery strategy library during a monitoring period, for example, a total of K=4 recovery strategies are counted during the monitoring period;

[0111] w k The steps for obtaining parameters are as follows: before each strategy is put into use, the key indicator response files in multiple actual operations are recorded. By analyzing the frequency of use of the strategy in the past month, the degree of impact on the multi-route service quality, and the coverage of the hardware nodes involved in the strategy, the importance of the strategy is quantified in a numerical way. For these non-numerical data contents, they need to be converted through pre-defined quantitative standards. For example, after recording the frequency of strategy use, it is divided by the maximum number of strategy uses in the period to obtain the relative frequency ratio, the ratio of the number of hardware node coverage to the total number of nodes is obtained, and the degree of service quality impact is quantified as a decimal value between 0 and 1 through the correlation analysis method. Finally, these quantitative parameters are weighted and averaged to synthesize the strategy importance value w. k , and confirmed w1=0.7, w2=0.5, w3=0.6, w4=0.4 based on multi-day monitoring data;

[0112] r k The steps for obtaining parameters are as follows: numerically extract the applicability score of each strategy in recent practical applications. The score quantifies the effectiveness of the strategy in various fault scenarios through a decision tree model, ranging from 0 to 100. The data is obtained through continuous monitoring records and statistical calculations. Here, r1=85, r2=70, r3=90, and r4=65 can be read from the existing records;

[0113] T k The steps of obtaining parameters are as follows: by summarizing the performance data of the strategy under the optimal deployment conditions, the best performance value of the strategy under the ideal conditions of bandwidth, delay, packet loss rate, etc. is selected from a large amount of experimental data, and the value is defined as the theoretical optimal performance value of the strategy. For example, according to the experimental data, T1 = 100, T2 = 95, T3 = 100, and T4 = 90 are obtained;

[0114] Calculation process:

[0115] Substituting the above parameters into:

[0116]

[0117] First calculate the exponential part of each term:

[0118]

[0119] Multiply it by the weight:

[0120] w1×exp(-0.85)=0.7×0.427=0.2989

[0121] w2×exp(-0.7368)=0.5×0.479=0.2395

[0122] w3×exp(-0.9)=0.6×0.407=0.2442

[0123] w4×exp(-0.7222)=0.4×0.486=0.1944

[0124] Sum and divide by 4:

[0125]

[0126] The results show that the comprehensive influence of the strategy is about 0.244. When the value is higher, it means that the overall characteristics of the strategy have a more significant impact on the network status. When the value is lower, it means that the strategy has a smaller impact on the network status. Values ​​in different ranges can guide screening and comparison in the strategy optimization step. If the value is higher than 0.5, it can be identified as a more prominent strategy, and if it is lower than 0.2, it is a relatively weak strategy.

[0127] When comparing the recovery strategies, first read the strategy comprehensive impact data list just obtained, identify the comprehensive impact value of each strategy in the list, and then consult the set judgment criteria from the attached documents. The judgment criteria are extracted from the strategy operation records of the past few days. The comprehensive impact of all strategies is statistically analyzed to obtain the mean and fluctuation range. The interval of the mean plus or minus two times the standard deviation is used as the judgment standard interval. The comprehensive impact values ​​of all strategies are numerically compared with the standard interval. When the comprehensive impact of a strategy is greater than the upper limit of the standard interval, the strategy is marked as a high-impact strategy in the marking file. If the comprehensive impact of a strategy is lower than the lower limit of the standard interval, the strategy is marked as a low-impact strategy in the marking file. The remaining strategies are classified as medium-impact ranges. Then, the high-impact strategies are listed in the order of the strategy numbers in the marking record file, and the corresponding strategy numbers are extracted. The number is compared with the information entries of the strategy name and the corresponding operation instructions in the index file. Finally, the number and the corresponding strategy name are listed in the output file to form a list of optimal recovery strategy result data.

[0128] See also Figure 2 ,The steps to apply the optimal recovery strategy to the router configuration are:

[0129] Adjust routing protocols, interface settings, and security policies based on the optimal recovery strategy;

[0130] Update router settings and monitor response and error logs during configuration.

[0131] Specifically, according to the specific guidance of the optimal recovery strategy, the routing protocol is first adjusted, such as modifying the cost weight parameter of Open Shortest Path First (OSPF) to ensure that data flows preferentially pass through the path with the lowest cost. At the same time, the priority of the Border Gateway Protocol (BGP) is adjusted, and the preference value of the route announcement is set. The adjustment is based on the real-time analysis results of the current network traffic and data transmission. Further, the interface settings are updated, including adjusting the rate limit of the port. The rate is dynamically adjusted according to the traffic load of the port. At the same time, the access control list (ACL) is configured to define which IP addresses have access rights and which are denied, thereby enhancing network security.

[0132] After completing the router configuration adjustment, update the router settings. First, upload the latest configuration file, which contains instructions for previously adjusted protocols, interfaces, and security settings. Then execute the configuration file and monitor error logs that may appear during the process. Use network diagnostic tools such as traceroute and ping to test network connectivity, verify data paths and network delays, ensure the correct implementation of all configurations, and finally confirm the network operation status to ensure normal operation of the router.

[0133] The steps for obtaining the performance recovery evaluation value are as follows:

[0134] Collect the delay, bandwidth utilization and packet loss rate after the optimal recovery strategy is implemented to obtain a performance data set after recovery;

[0135] Based on the performance data set after recovery, the difference between the performance data after recovery and the performance data before recovery is calculated using the following formula:

[0136]

[0137] Among them, P represents the average performance improvement value, p post,i is the value of the i-th performance indicator after recovery, p pre,i is the value of the i-th performance indicator before recovery, ZN is the total number of performance indicators;

[0138] Based on the average performance improvement value, the recovery effect of the network is evaluated to obtain a performance recovery evaluation value.

[0139] Specifically, for the data acquisition process after the implementation of relevant strategies, by continuously monitoring the data link status in the actual operating network environment, the recording device continuously collects the delay value, bandwidth utilization ratio and data packet loss ratio within the specified time interval, starting from every morning, the delay of the network node is measured at a fixed interval, the time series record of a single data point is stored in a pre-configured recording file, each data record is marked with a timestamp, the network transmission status in each time slice is scanned in turn, the delay value in each time slice is accumulated and its average delay value is calculated, and then the recorded bandwidth utilization value is scanned in turn, and the ratio of the actual throughput to the maximum available bandwidth is counted in the target time interval, and the ratio is converted into bandwidth utilization. When processing the data packet loss rate, the total number of data packets sent and received in the same time interval is counted, and the ratio of the number of lost packets to the total number of sent packets is calculated. The obtained delay mean, bandwidth utilization and packet loss rate data are stored in a data table file in a serialized format, and this step is repeated according to the predetermined recording cycle, so as to form a complete post-recovery performance data set after continuously accumulating a certain number of data points;

[0140] The benefit of the formula is that it can quantitatively demonstrate the degree of improvement of network optimization measures by differentially measuring the performance indicators after restoration and before restoration;

[0141] P represents the average performance improvement value, and its acquisition steps are as follows: by numerically recording the performance indicators in the same monitoring period, the performance indicator set before recovery {p pre,1 ,p pre,2 ,p pre,3} and the performance indicator set after recovery {p post,1 ,p post,2 ,ppost,3 These performance indicators include normalized latency, bandwidth utilization, and packet loss rate recorded by network monitoring equipment. Data is collected over a long period of time during operation, and p is obtained by averaging hundreds of records. pre,i and p post,i The actual value of ZN is the total number of performance indicators, which is set by clearly defining the number of indicators and retrieving the commonly used performance parameter setting range in the industry;

[0142] Calculation process:

[0143] Set ZN = 3 performance indicators, take p pre,1 =70.0, p post,1 =80.0, p pre,2 =55.0, p post,2 =60.0, p pre,3 =85.0, p post,3 =90.0,

[0144] Then (p post,1 -p pre,1 )=80.0-70.0=10.0

[0145] (p post,2 -p pre,2 )=60.0-55.0=5.0

[0146] (p post,3 -p pre,3 )=90.0-85.0=5.0

[0147] Add the difference: 10.0 + 5.0 + 5.0 = 20.0

[0148] Substitute into the formula:

[0149]

[0150] The results show that considering the improvements in the three performance indicators, the average performance improvement value of the network is about 6.6667. The larger the value, the more obvious the overall performance improvement. When the value exceeds 10, it means a significant overall improvement, while less than 5 means limited improvement.

[0151] See also Figure 3 ,The steps to obtain the isolation and traffic management results are:

[0152] Based on the performance recovery evaluation value, analyze the performance impact between tenants, identify the affected tenants, and determine the isolation measures that need to be refined;

[0153] Based on the detailed isolation measures, configure traffic redirection and traffic restriction settings to generate isolation and traffic management results.

[0154] Specifically, based on the performance recovery evaluation value, the performance data of each tenant network is analyzed and compared with the established performance baseline. The tenants with abnormal performance are identified through difference analysis. During the identification process, key indicators such as response time and data transmission rate are first extracted from the performance data, and then specific performance deviation thresholds are set. For example, a response time delay of more than 10% or a data transmission rate drop of more than 15% are used as criteria for identifying affected tenants. Tenants that need attention are screened out through these criteria. Furthermore, it is analyzed whether the tenant's performance degradation is associated with the recent network recovery activity. This analysis is achieved by comparing the performance records before and after recovery. Finally, the detailed isolation measures that need to be implemented are determined based on the analysis results. These measures specifically address the identified performance problems and ensure that each tenant can operate within the safety threshold, thereby obtaining detailed isolation measures.

[0155] Based on the detailed isolation measures obtained, specific network configuration adjustments are implemented, involving traffic redirection and traffic restriction settings. First, the business priority and network usage pattern of each tenant are determined, and then the corresponding traffic control rules are set, such as setting high-priority channels for critical businesses and setting bandwidth caps for non-critical businesses. Next, based on the current network traffic and expected business growth, the traffic restriction parameters are dynamically adjusted. For example, when the total network traffic exceeds 80% of the maximum capacity, the bandwidth cap for non-critical businesses is automatically reduced to maintain stable operation and efficient use of the network. The implementation of these measures is completed through detailed configuration of the router. During the configuration process, the traffic status is continuously monitored, and the traffic management strategy is adjusted according to real-time data. The isolation and traffic management results finally generated ensure the effective management of the network and the smooth operation of each tenant's business.

[0156] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A router isolation management method supporting a multi-tenant virtual network, characterized in that: The following steps are involved: Collect the real-time traffic data of each tenant network, compare the traffic data with the performance baseline set by each tenant, obtain traffic anomaly indicators, identify the fault area in the network through algorithms based on the traffic anomaly indicators, and generate fault location information; Based on the fault location information, use a decision tree to analyze the applicability of each recovery strategy, compare the degree of impact of each recovery strategy on the network, select a recovery strategy, generate an optimal recovery strategy, and apply the optimal recovery strategy to the router configuration; Monitor the implementation effect of the preferred recovery strategy, collect the network performance data after recovery in real time, compare it with the data before recovery, and generate a performance recovery evaluation value; Based on the performance recovery evaluation value, the isolation measures between tenants are refined to generate isolation and traffic management results.

2. The router isolation management method supporting multi-tenant virtual networks according to claim 1, characterized in that: The steps for obtaining the traffic anomaly indicator are as follows: Collect the real-time traffic data of each tenant's network, compare it with the performance baseline set by each tenant, obtain the deviation value of each data point, and generate a deviation data set; Based on the deviation data set, the ratio of the sum of the absolute values ​​of the deviations of each data point to the total number of data points is calculated, and with reference to the data volatility, the calculation formula is: Where E represents the weighted mean deviation, d i represents the flow value of the ith data point, b represents the baseline flow value, N represents the total number of data points, λ is the adjustment coefficient, and μ is the average flow value of all data points; Based on the weighted average deviation, an abnormal threshold is set, and the deviation value exceeding the abnormal threshold is marked as abnormal, so as to obtain a flow abnormality index.

3. The router isolation management method supporting multi-tenant virtual networks according to claim 1, characterized in that: The steps for obtaining the fault location information are as follows: Receiving the traffic anomaly indicator, identifying data points of traffic anomaly, and obtaining an abnormal data set; Based on the abnormal data set, the similarity between each data point is calculated, and the calculation formula is: Among them, S represents the similarity score, x i and x j are two abnormal data points, σ ​​is the standard deviation, and M is the number of abnormal data points; Based on the similarity score, the fault area in the network is determined to obtain fault location information.

4. The router isolation management method supporting multi-tenant virtual networks according to claim 1, characterized in that: The steps for obtaining the preferred recovery strategy are: Collecting the fault location information, applying a decision tree model to analyze the applicability of each network recovery strategy, and obtaining a recovery strategy applicability score; Based on the applicability score of the recovery strategy, the comprehensive impact of each strategy is calculated using the following formula: Among them, I represents the comprehensive influence of the strategy, w k is the weight of the k-th strategy, r k Score the suitability of the k-th strategy, T k is the theoretical optimal performance value of the kth strategy, and K is the total number of strategies; Based on the comprehensive impact, various restoration strategies are compared and selected to obtain an optimal restoration strategy.

5. The router isolation management method supporting multi-tenant virtual networks according to claim 1, characterized in that: The steps of applying the preferred recovery strategy to the router configuration are: Based on the preferred recovery strategy, adjusting routing protocols, interface settings and security policies; Update router settings and monitor response and error logs during configuration.

6. The router isolation management method supporting multi-tenant virtual networks according to claim 1, characterized in that: The steps for obtaining the performance recovery evaluation value are as follows: Collecting the delay, bandwidth utilization and packet loss rate after implementing the preferred recovery strategy to obtain a performance data set after recovery; Based on the performance data set after recovery, the difference between the performance data set after recovery and the performance data set before recovery is calculated using the following formula: Among them, P represents the average performance improvement value, p post,i is the value of the i-th performance indicator after recovery, p pre,i is the value of the i-th performance indicator before recovery, ZN is the total number of performance indicators; Based on the average performance improvement value, the recovery effect of the network is evaluated to obtain a performance recovery evaluation value.

7. The router isolation management method supporting multi-tenant virtual networks according to claim 1, characterized in that: The steps for obtaining the isolation and traffic management results are as follows: Based on the performance recovery evaluation value, the performance impact between tenants is analyzed, the affected tenants are identified, and the isolation measures that need to be refined are determined.

8. The router isolation management method supporting multi-tenant virtual networks according to claim 7, characterized in that: The step of obtaining the isolation and traffic management results also includes: configuring traffic redirection and traffic restriction settings according to the detailed isolation measures to generate isolation and traffic management results.

Images