Network data management system based on intelligent flow control

Abstract

The invention relates to the technical field of network optimization, in particular to a network data management system based on intelligent flow control, which comprises a flow partitioning and sample extracting module, a flow partitioning and sample extracting module, a flow partitioning and sample extracting module, a flow partitioning and sample extracting module and a flow partitioning and sample extracting module, and generating a load feature sample set according to the screening result. In the invention, through extraction of the time sequence information and division of the sub-traffic segments, the network traffic data is subjected to packet management, and the loads of the source node and the target node are accurately marked and screened, so that more accurate load characteristic data is generated, and unnecessary data redundancy is avoided. In the calculation of the load weighted value, the prediction precision of the data flow load is further improved by combining the dynamic priority weight and the multi-dimensional weight mechanism of the flow variable quantity, so that the trend change in the subsequent time slice can be estimated more scientifically.

Description

Technical Field

[0001] The present invention relates to the technical field of network optimization, and in particular to a network data management system based on intelligent flow control. Background Art

[0002] The field of network optimization technology includes the dynamic allocation and management of network resources, and improves the network's operating efficiency and service quality through reasonable scheduling and control mechanisms. The core content of this technology field includes network traffic analysis and control, bandwidth resource allocation, data packet routing and forwarding, network load balancing, and delay management.

[0003] Among them, the network data management system uses data collectors to comprehensively monitor the traffic information in the network, subdivides and marks the collected data through traffic classification rules, and then combines the scheduling algorithm to dynamically adjust the priority and allocate resources for the data flow, thereby ensuring the orderliness of network operation and the stability of data transmission.

[0004] Existing technologies mainly rely on traffic classification rules to segment and mark collected data. Although they can provide basic traffic management functions, they lack the ability to accurately predict real-time changing load trends, making it difficult to effectively deal with large traffic fluctuations in a short period of time. In addition, in the dynamic priority adjustment, existing scheduling algorithms are unable to comprehensively consider the load distribution of source nodes and target nodes and the global resource occupancy of transmission paths, which can easily lead to resource overload or inefficient operation of some paths. In terms of abnormal traffic management, existing technologies lack accurate tracking and isolation strategies for abnormal traffic propagation paths, which may cause abnormal traffic to affect the transmission performance of normal traffic. For example, in high-concurrency scenarios, abnormal traffic occupying key path resources will lead to a serious decline in overall network performance, and data delays and packet loss rates will increase significantly, limiting the efficiency and stability of network data management systems in complex network environments. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a network data management system based on intelligent flow control.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: A network data management system based on intelligent flow control includes: The traffic partitioning and sample extraction module obtains the time series information of network traffic and divides it into sub-traffic segments, filters the data packets in the sub-traffic segments, and generates a load feature sample set based on the screening results; The distributed trend prediction and synchronization module extracts the load records of the sub-flow segments in the load feature sample set, calculates the load weighted value corresponding to each sub-flow segment based on the load records, predicts the load change trend of the sub-flow segment in the subsequent time segment with reference to the load weighted value, and generates a global load change trend list; The path optimization allocation and dynamic adjustment module calculates the load deviation value between the sub-flow segments according to the load trend information of the sub-flow segments in the global load change trend list, establishes the load distribution path of the sub-flow segment according to the load deviation value, and dynamically adjusts the load distribution path to obtain the dynamic path flow distribution result; The abnormal behavior identification and isolation module extracts the real-time traffic information of each path in the dynamic path traffic allocation result, identifies the abnormal traffic, reallocates the abnormal traffic to the isolated path, and generates a network traffic data optimization management result.

[0007] As a further solution of the present invention, the step of dividing the sub-flow segments is specifically: Capture data packets in the network based on network traffic monitoring equipment, segment and archive the traffic data at fixed time intervals, and generate a collection of time series segments; Based on the time series segment set, by parsing the source IP address and the destination IP address of each data packet, the traffic data is grouped and the information is recorded according to the node pairing rule to generate a sub-traffic segment set.

[0008] As a further solution of the present invention, the steps of acquiring the load characteristic sample set are specifically as follows: Based on the sub-flow segment set, extracting the packet size, transmission frequency, and load occupancy ratio information of the source node and the target node from the packet, screening the packet in combination with the load occupancy ratio of the source node and the target node, and integrating the screening results into a load feature to generate load feature data; The load characteristics in the load characteristic data are compared with preset load thresholds one by one, the load characteristics exceeding the thresholds are marked, and the marking results are sorted to generate a load characteristic sample set.

[0009] As a further solution of the present invention, the step of calculating the load weighted value corresponding to each sub-flow segment is specifically as follows: Based on the load feature sample set, retrieve the information of each sub-flow segment, divide each group of data into corresponding time segments according to the time segment, and extract the load change value and flow change amount of each sub-flow segment to generate the load record set of the sub-flow segment; Based on the load record set of the sub-flow segment, the formula is adopted: ; Calculate sub-flow segments Load weighted value ; in, It is a sub-flow segment The total number of time segments, It is a sub-flow segment In time segment The load value in It is a sub-flow segment In time segment The traffic weight in It is a sub-flow segment In time segment Dynamic priority weights.

[0010] As a further solution of the present invention, the steps of obtaining the global load change trend list are specifically as follows: According to the load weighted value, the formula is used: ; Calculate sub-flow segments In the next time segment The predicted load value ; in, It is a time segment inner sub-traffic segment The load growth factor, is the adjustment factor for load change trend, It is a time segment The load increment, is the time segment length; Based on the predicted load value, the load change trends of all sub-flow segments are sorted out in time dimension and space dimension, and a global load change trend list is generated by classifying and integrating the load data of each sub-flow segment.

[0011] As a further solution of the present invention, the step of calculating the load deviation value between the sub-flow segments is specifically: Based on the global load change trend list, extract the load forecast value of each sub-flow segment, sort the load values ​​in combination with the time dimension, and generate a load trend data set according to the sorting result; Based on the load trend data set, the formula is used: ; Calculate sub-flow segments and In time segment Load deviation value ; in, It is a sub-flow segment In time segment The predicted load value.

[0012] As a further solution of the present invention, the step of obtaining the dynamic path flow distribution result is specifically as follows: Based on the load deviation value, the load deviation values ​​between the sub-flow segments are integrated in a matrix form, paths of various load conditions are identified, and corresponding load paths are adjusted to generate sub-flow segment path load distribution results; Based on the sub-flow segment path load distribution results, the formula is adopted: ; Calculate sub-flow segments and Path flow distribution ratio , and according to the sub-flow segment and Path flow distribution ratio Calculate the adjusted path flow , dynamically adjust the load distribution path according to the adjusted path flow, and obtain the dynamic path flow distribution result; in, It is a sub-flow segment The load weighted value of is the initially allocated path flow value.

[0013] As a further solution of the present invention, the steps for obtaining the network traffic data optimization management result are specifically as follows: Based on the dynamic path traffic allocation result, by monitoring the data packet transmission rate in the path and the traffic value in the time interval, comparing the traffic fluctuation with the preset fluctuation threshold range, identifying the source node, target node and propagation path of the traffic anomaly, and generating the abnormal traffic identification result; Based on the abnormal traffic identification result, combined with the real-time load status and bandwidth utilization of the currently available paths, the abnormal traffic is reallocated to the isolated paths, the operating status of the isolated paths is monitored, and the network traffic data optimization management results are generated.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, network traffic data is grouped and managed by extracting time series information and dividing it into sub-traffic segments, and the loads of its source nodes and target nodes are accurately marked and screened, generating more accurate load feature data and avoiding unnecessary data redundancy. In the calculation of load weighted values, the multi-dimensional weight mechanism combining dynamic priority weights and traffic changes is combined to further improve the prediction accuracy of data flow loads, so that trend changes in subsequent time segments can be more scientifically predicted. In the path optimization and dynamic adjustment stage, the problem of uneven path load distribution is effectively solved by identifying the load deviation values ​​between sub-traffic segments and calculating the path allocation ratio, ensuring efficient use of resources. In the process of abnormal traffic identification and isolation, the source of traffic anomalies is detected in time by dynamically comparing traffic fluctuations with the threshold range, and combined with the redistribution strategy of the isolated path, the stable transmission of normal traffic is ensured, while the interference of abnormal traffic on network performance is isolated. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a system flow chart of the present invention; Figure 2 A flow chart for dividing sub-flow segments of the present invention; Figure 3 A flow chart of obtaining a load characteristic sample set for the present invention; Figure 4 A flow chart for calculating the load weighted value corresponding to each sub-flow segment of the present invention; Figure 5 A flowchart for obtaining a global load change trend list for the present invention; Figure 6 A flow chart for calculating the load deviation value between sub-flow segments according to the present invention; Figure 7 A flow chart of obtaining dynamic path flow distribution results of the present invention; Figure 8 The present invention is a flow chart for obtaining network traffic data optimization management results. DETAILED DESCRIPTION

[0016] 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.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 , the network data management system based on intelligent flow control includes: The traffic partitioning and sample extraction module obtains the time series information of network traffic and divides it into sub-traffic segments, filters the data packets in the sub-traffic segments, and generates a load feature sample set based on the screening results; The distributed trend prediction and synchronization module extracts the load records of the sub-flow segments in the load feature sample set, calculates the load weighted value corresponding to each sub-flow segment based on the load records, predicts the load change trend of the sub-flow segment in the subsequent time segment with reference to the load weighted value, and generates a global load change trend list; The path optimization allocation and dynamic adjustment module calculates the load deviation value between the sub-flow segments according to the load trend information of the sub-flow segments in the global load change trend list, establishes the load distribution path of the sub-flow segment according to the load deviation value, and dynamically adjusts the load distribution path to obtain the dynamic path flow distribution result; The abnormal behavior identification and isolation module extracts the real-time traffic information of each path in the dynamic path traffic allocation results, identifies abnormal traffic, reallocates abnormal traffic to the isolated path, and generates network traffic data optimization management results; The load feature sample set specifically includes data packet size, data packet transmission frequency, source node load occupancy ratio, and target node load occupancy ratio. The global load change trend list includes sub-flow segment load weighted values, time dimension distribution of sub-flow segment load change trends, and spatial dimension distribution of sub-flow segment load change trends. The dynamic path traffic allocation results specifically include sub-flow segment path load allocation paths and dynamic adjustment records of load paths. The network traffic data optimization management results include abnormal traffic identification results, isolation path allocation results, and abnormal traffic propagation paths.

[0019] See also Figure 2 , the steps of dividing the sub-flow segments are as follows: Capture data packets in the network based on network traffic monitoring equipment, segment and archive the traffic data at fixed time intervals, and generate a collection of time series segments; First, use network traffic monitoring equipment to capture data packets in the network in real time, record the timestamp information of each data packet, sort the traffic data according to the timestamp information to ensure the continuity and accuracy of the time series, and use the time interval segmentation method to segment the traffic data at fixed intervals (such as 5 seconds or 10 seconds). Pack the traffic data in continuous time segments in the time series into independent time segment sets, and record the number of data packets in each time segment and their transmission interval. Use data storage formats (such as CSV or JSON file formats) to archive and store the time segment set, ensure that each time segment contains a complete set of data packets associated with the timestamp, and generate a time series segment set.

[0020] Based on the set of time series segments, the source IP address and destination IP address of each data packet are parsed, the traffic data is grouped and the information is recorded according to the node pairing rules to generate a set of sub-traffic segments.

[0021] The network traffic collection system parses the IP address information of each data packet, extracts the source IP address and destination IP address of the data packet from the header information of the data packet, forms a node association table of the data packet using the source IP address and the destination IP address, and groups the data packets in each time segment according to the pairing rules in the association table. After grouping, the traffic data of the source node and the destination node are stored as an independent sub-traffic segment, and the specific information of the source node and the destination node in each sub-traffic segment is recorded, including the total data volume of the traffic transmission, the transmission frequency, and the average delay of the data packet transmission. Each time segment is marked using the grouped sub-traffic segment set to generate a sub-traffic segment set.

[0022] See also Figure 3 , the steps for obtaining the load characteristic sample set are as follows: Based on the sub-flow segment set, the packet size, transmission frequency, and the load occupancy ratio of the source node and the target node are extracted from the data packet. The data packet is screened in combination with the load occupancy ratio of the source node and the target node, and the screening results are integrated into the load characteristics to generate load characteristic data. By calling the data packet transmission log recorded in the network monitoring device, the header information of each data packet is parsed one by one to obtain its transmission frequency and size. The load occupancy ratio of the source node and the target node is defined as the ratio of the data packet size processed by the source node or the target node per unit time to its theoretical maximum carrying capacity. The actual load occupancy ratio is obtained by counting the cumulative amount of data processed by the source node and the target node in a specified time period and comparing it with its rated processing capacity. The load occupancy ratio threshold range of the source node and the target node is set through sampling rules, and the load data packets exceeding the threshold are marked with priority, and their corresponding data packet size, transmission frequency, and load occupancy ratio of the source node and the target node are integrated into load characteristic data.

[0023] Compare the load characteristics in the load characteristic data with the preset load thresholds one by one, mark the load characteristics exceeding the thresholds, and sort out the marking results to generate a load characteristic sample set; By comparing each characteristic value in the load characteristic data with the preset threshold value one by one, the threshold value includes the maximum allowable range of the source node load occupancy ratio, the maximum allowable range of the target node load occupancy ratio, the maximum allowable value of the data packet size and the lower limit of the data packet transmission frequency. For the source node and target node load occupancy ratio, the ratio rule is used to judge the node load that exceeds the set threshold. For the data packet size and transmission frequency, it is directly judged whether it meets the screening conditions according to the numerical range. All data packets exceeding the threshold are marked as abnormal load characteristics, and the source nodes and target nodes are classified separately according to the classification rules to generate the final load characteristic sample set. The sample set includes the source nodes, target nodes, data packet sizes and transmission frequencies that exceed the threshold.

[0024] See also Figure 4 , the steps for calculating the load weighted value corresponding to each sub-flow segment are as follows: Based on the load feature sample set, retrieve the information of each sub-flow segment, divide each group of data into corresponding time segments according to the time segment, and extract the load change value and flow change amount of each sub-flow segment to generate the load record set of the sub-flow segment; First, the associated information of each sub-traffic segment is retrieved one by one from the load feature sample set, including the source node, target node, data packet size and transmission frequency. The data in the sample set is grouped using the timestamp, and each group of data is mapped to its own time segment according to the time segment division. The total data packet size, transmission frequency and load change value of each sub-traffic segment in each time segment are extracted, the load occupancy ratio of the source node and the target node in the time segment is recorded, the total traffic change of the source node and the target node in the time segment is calculated, the sub-traffic segment load records in each time segment are summarized, and stored in chronological order as a load time series set.

[0025] Based on the load record collection of sub-flow segments, the formula is: ; Calculate sub-flow segments Load weighted value , which indicates the comprehensive load level of the sub-traffic segment in all time segments, in Mbps.

[0026] in, It is a sub-flow segment The total number of time segments included indicates the number of time segments for network traffic data. The entire monitoring period is divided by setting the time interval, for example, a time segment is divided every 5 seconds or 10 seconds. Taking the total monitoring time of 100 seconds and the time interval of 5 seconds as an example, the following is obtained: , It is a sub-flow segment In time segment The load value in indicates the traffic size in the time segment, in Mbps. The total size of the data packet is directly extracted through the network traffic monitoring device. For example, in the time segment The total number of packets monitored within is Mbps, then , It is a sub-flow segment In time segment The traffic weight in is used to measure the importance of traffic. It is calculated based on the transmission frequency of the sub-traffic segment in the time segment. The higher the transmission frequency, the greater the weight. It is a sub-flow segment In time segment The dynamic priority weight reflects the real-time load fluctuation and congestion in the time segment. It is obtained by analyzing the load fluctuation amplitude and node congestion rate in the time segment. The fluctuation amplitude is determined by the difference between the maximum and minimum load values. The node congestion rate is determined by monitoring the congestion of the source node or the target node. For example, if the time segment The maximum internal load is 150Mbps, the minimum is 100Mbps, the fluctuation range is 50Mbps, and the node congestion rate is 80%. Combining the above parameters, we can determine .

[0027] If the parameter : The number of time segments is 3. :The load values ​​are Mbps, Mbps, Mbps. :The traffic weights are , , . : The dynamic priority weights are , , , put it into the formula: ; Calculate the numerator: ; Calculate the denominator: ; Calculate the weighted load value: ; This result shows that the sub-flow segment The weighted load value 126.87Mbps, by introducing dynamic priority weight , making the calculation result more in line with the actual dynamic change characteristics of the sub-traffic segment. This value is slightly higher than the benchmark load value (such as the basic value of 125Mbps without dynamic adjustment), which can more accurately reflect the comprehensive load level of the sub-traffic segment and provide reliable basic data for global trend prediction.

[0028] See also Figure 5 , the specific steps for obtaining the global load change trend list are: According to the load weighted value, the formula is used: ; Calculate sub-flow segments In the next time segment The predicted load value ; in, It is a time segment inner sub-traffic segment The load growth coefficient is a dimensionless value that represents the relative ratio of the load value of the current time segment to the overall maximum load. It is used to measure the load proportion of the current time segment and is calculated based on the historical load value. The formula is as follows: , It is the adjustment coefficient of the load change trend. It is a dimensionless value used to balance the weight of the load change trend on the prediction result. It is set according to the dynamic characteristics of the network. For example, a larger adjustment coefficient is set when the network fluctuation is high. The specific value is verified by historical monitoring data experiments. For example, a higher value is set during congestion to enhance the response to the change trend. It is a time segment The load increment is in Mbps, which represents the difference between the load value of the current time segment and the load value of the previous time segment. The calculation formula is: , It is a sub-flow segment In the previous time segment The load value in It is the length of the time segment in seconds, indicating the time interval between adjacent time segments. It is directly obtained according to the set time segment interval. For example, if the length of each time segment is set to 10 seconds, then .

[0029] Set parameter sub-flow segment The weighted load value of: Mbps. Load growth factor: The current load value is Mbps, the historical maximum load value is 200Mbps, then , load increment: load of the previous time segment Mbps, the current load is Mbps, then , time segment length: Seconds, adjustment factor: , enter the formula to calculate: ; ; This result shows that the sub-flow segment In time segment The predicted load value is 96.65Mbps, which is slightly higher than the load value of the previous time segment (95Mbps, as the baseline value), indicating that the load growth trend of this traffic segment in the network is gradually increasing. This result can be used to adjust the network resource allocation for the next time segment to ensure network load balancing.

[0030] Based on the predicted load value, the load change trends of all sub-flow segments are sorted out in the time dimension and space dimension, and a global load change trend list is generated by classifying and integrating the load data of each sub-flow segment.

[0031] First, the predicted load value of each sub-flow segment in all time segments is extracted, for example, In time segment The predicted load value is Mbps, the predicted load values ​​of other sub-flow segments are calculated and extracted in a similar way. The predicted load values ​​of each sub-flow segment are sorted according to the time segment number to form a load change sequence in the time dimension. In the spatial dimension, according to the topological distribution of the source nodes and target nodes of the sub-flow segments, the sub-flow segments with the same source nodes and target nodes are classified into the same spatial domain, for example, the sub-flow segment and The source node is A, the target node is B, and the predicted load values ​​are Mbps and Mbps, the total spatial domain load of the two is Mbps. For all sub-traffic segments in the spatial domain, after the total load is calculated, a global load change trend list is generated through joint analysis of the time dimension and the spatial dimension. The list includes the load change sequence of each time segment, the total load value of each spatial domain, and the correlation information between the time and spatial domains. For example, if The total load value of a certain space domain AB is significantly higher than the average value Mbps, the spatial domain is marked as a high-load area. The path allocation is optimized or the traffic scheduling strategy is adjusted through the marking information, and finally a global load change trend list is generated, including the load sequence of the time segment, the total load distribution of the spatial domain, and the identification information of the relevant high-load areas.

[0032] See also Figure 6 , the steps for calculating the load deviation value between sub-flow segments are as follows: Based on the global load change trend list, extract the load forecast value of each sub-flow segment, sort the load values ​​in combination with the time dimension, and generate a load trend data set based on the sorting results; Collect the load trend information of the sub-flow segments in the global load change trend list, extract the load forecast value of each sub-flow segment from the global load change trend list, including the load sequence in the time dimension and the load distribution data in the spatial domain, sort the forecast load values ​​of each sub-flow segment according to the time dimension, and generate the load trend curve of each time segment, such as the sub-flow segment The load sequence is Mbps, combined with the spatial domain information of the source node and the target node, extract the load distribution data of all sub-flow segments in the same spatial domain, and group the predicted values ​​of each sub-flow segment by spatial domain. For example, spatial domain AB contains sub-flow segments and The predicted load values ​​are Mbps and Mbps, the total load of the two is calculated as Mbps, and mark the load trend of each spatial domain for subsequent analysis, such as marking high-load areas and low-load areas, to form a load trend data set.

[0033] Based on the load trend data set, the formula is used: ; Calculate sub-flow segments and In time segment Load deviation value , in Mbps, is used to measure the difference in load distribution between two sub-traffic segments.

[0034] in, It is a sub-flow segment In time segment The predicted load value in Mbps.

[0035] If the sub-flow segment The predicted load value is: Mbps, traffic sub-segment The predicted load value is: Mbps, enter the parameters: ; This result shows that the sub-flow segment and In time segment The load deviation value is 9.15Mbps. The existence of the deviation value indicates that there is a difference in load distribution between the two. This value can be used to adjust the resource allocation ratio of sub-traffic segments when optimizing path allocation, such as reducing the traffic on paths with higher loads or increasing the utilization of paths with lower loads, thereby achieving the purpose of network load balancing.

[0036] See also Figure 7 ,The specific steps for obtaining the dynamic path traffic allocation results are: Based on the load deviation value, the load deviation values ​​between the sub-flow segments are integrated in a matrix form, the paths of various load conditions are identified, and the corresponding load paths are adjusted to generate the sub-flow segment path load distribution results; The load distribution path of the sub-flow segment path is established according to the load deviation value, and the load deviation value between the aforementioned sub-flow segments is extracted, for example, the sub-flow segment and The load deviation value is Mbps, the load deviation values ​​of all sub-flow segments are integrated into the load deviation matrix between sub-flow segments. For example, The load deviation matrix of the sub-flow segment set is: , high-load paths and low-load paths are identified by sorting the values ​​in the load deviation matrix, e.g., the maximum value in the deviation matrix Corresponding sub-flow segment and , indicating that the load difference between the two is the largest, part of the traffic in the high-load path is redistributed to the low-load path, such as Part of the flow is allocated to the sub-flow segment , a preliminary path load distribution path table is formed, and the mapping relationship between the source node and the target node of the sub-flow segment is recorded according to the distribution path table to generate the sub-flow segment path load distribution result.

[0037] Based on the sub-flow segment path load distribution results, the formula is used: ; Calculate sub-flow segments and Path flow distribution ratio , and according to the sub-flow segment and Path flow distribution ratio Calculate the adjusted path flow , dynamically adjust the load distribution path according to the adjusted path flow, and obtain the dynamic path flow distribution result; in, It is a sub-flow segment The load weight value of the path, in Mbps, indicates the weight of the path. is the initial allocated path traffic value in Mbps, which represents the initial traffic allocation of the path, extracted from the preliminary path allocation table, e.g. Mbps.

[0038] Set parameter sub-flow segment Weighted value of: Mbps. Sub-traffic segment Weighted value of: Mbps. Sub-traffic segment and Preliminary path flow: Mbps.

[0039] Calculate the allocation ratio : ; Calculate the adjusted path flow based on the allocation ratio ; This result shows that the adjusted path flow Mbps is based on load weighted values Mbps and Mbps is calculated by calculating the dynamic allocation result of the path allocation ratio , realizing the path and Traffic is redistributed under the current load conditions. Initial path traffic Mbps, the ratio under uniform distribution is 0.5, but the distribution ratio calculated based on the weighted value is , indicating the path Compared to the path The load pressure is higher, so more flow distribution is required. Adjusted distribution results Mbps, in accordance with the calculation result of the allocation ratio, and the remaining traffic Mbps to the less loaded path This result ensures that the path traffic distribution conforms to the weight of the load-weighted value, and dynamically adjusts to balance the load, reducing the pressure on high-load paths and improving the balance of the overall network.

[0040] See also Figure 8 ,The specific steps for obtaining the network traffic data optimization management results are: Based on the dynamic path traffic allocation results, by monitoring the packet transmission rate within the path and the traffic value within the time interval, comparing the traffic fluctuation with the preset fluctuation threshold range, identifying the source node, target node and propagation path of the traffic anomaly, and generating the abnormal traffic identification results; By real-time monitoring of the data packet transmission rate and transmission interval within the path, the traffic value of each path within a fixed time period is obtained. The obtained traffic data is compared and screened in combination with the preset traffic fluctuation threshold range (for example, 10% to 30%) to determine whether there is abnormal fluctuation. The difference between the average traffic value (for example, 100Mbps) and the real-time traffic value (for example, 120Mbps) of a certain path in the time interval is calculated, and the difference ratio (for example, 20%) is compared with the fluctuation threshold range. If it exceeds the upper limit of 30%, it is determined to be abnormal. By sorting out the information such as the traffic source node, the target node, and the timestamp, the propagation path of the abnormal traffic is further tracked. For example, the traffic fluctuation abnormality of a certain path comes from node A, the target is node B, and the timestamps are concentrated in the peak hours of each day (for example, 18:00 to 19:00). Combined with the transmission data packet size, packet loss rate and delay data on the abnormal path (for example, the packet loss rate is 5% and exceeds the normal range by 3% to 4%), the source characteristics, path and related parameter information of the abnormal traffic are recorded, and finally integrated into the abnormal traffic identification result.

[0041] Based on the abnormal traffic identification results, combined with the real-time load status and bandwidth utilization of the currently available paths, the abnormal traffic is reallocated to the isolated paths, the operating status of the isolated paths is monitored, and the network traffic data optimization management results are generated; According to the source node, target node and traffic size of the abnormal traffic, by analyzing the real-time load status and bandwidth utilization of all available paths in the current network (for example, paths with bandwidth utilization less than 50% are selected as isolation path candidates), combined with the available transmission capacity of the isolation path (for example, the transmission capacity of an isolation path is 1Gbps, the current occupancy rate is 40%, and the remaining available bandwidth is 600Mbps), it is preferentially allocated to the adapted path, and the traffic distribution rate is further adjusted to ensure that the transmission rate meets the carrying capacity of the isolation path (for example, abnormal traffic with a traffic size of 200Mbps is distributed to two isolation paths in a 150Mbps and 50Mbps split manner respectively). The operating status of the isolation path after the redistribution of traffic is monitored in real time, including key parameters such as packet loss rate (such as increased from the original 1% to 2%, which is still within the normal range) and latency (such as increased from 20ms to 25ms). The performance change data of the isolation path and the traffic redistribution strategy are combined to form the final network traffic data optimization management result.

[0042] 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 network data management system based on intelligent flow control, characterized in that: The system comprises: The traffic partitioning and sample extraction module obtains the time series information of network traffic and divides it into sub-traffic segments, filters the data packets in the sub-traffic segments, and generates a load feature sample set based on the screening results; The distributed trend prediction and synchronization module extracts the load records of the sub-flow segments in the load feature sample set, calculates the load weighted value corresponding to each sub-flow segment based on the load records, predicts the load change trend of the sub-flow segment in the subsequent time segment with reference to the load weighted value, and generates a global load change trend list; The path optimization allocation and dynamic adjustment module calculates the load deviation value between the sub-flow segments according to the load trend information of the sub-flow segments in the global load change trend list, establishes the load distribution path of the sub-flow segment according to the load deviation value, and dynamically adjusts the load distribution path to obtain the dynamic path flow distribution result; The abnormal behavior identification and isolation module extracts the real-time traffic information of each path in the dynamic path traffic allocation result, identifies the abnormal traffic, reallocates the abnormal traffic to the isolated path, and generates a network traffic data optimization management result.

2. The network data management system based on intelligent flow control according to claim 1 is characterized in that: The steps of dividing the sub-flow segments are specifically as follows: Capture data packets in the network based on network traffic monitoring equipment, segment and archive the traffic data at fixed time intervals, and generate a collection of time series segments; Based on the time series segment set, by parsing the source IP address and the destination IP address of each data packet, the traffic data is grouped and the information is recorded according to the node pairing rule to generate a sub-traffic segment set.

3. The network data management system based on intelligent flow control according to claim 2 is characterized in that: The steps of acquiring the load characteristic sample set are specifically as follows: Based on the sub-flow segment set, extracting the packet size, transmission frequency, and load occupancy ratio information of the source node and the target node from the data packet, screening the data packet in combination with the load occupancy ratio of the source node and the target node, and integrating the screening results into a load feature to generate load feature data; The load characteristics in the load characteristic data are compared with preset load thresholds one by one, the load characteristics exceeding the thresholds are marked, and the marking results are sorted to generate a load characteristic sample set.

4. The network data management system based on intelligent flow control according to claim 3 is characterized in that: The step of calculating the load weighted value corresponding to each sub-flow segment is specifically as follows: Based on the load feature sample set, retrieve the information of each sub-flow segment, divide each group of data into corresponding time segments according to the time segment, and extract the load change value and flow change amount of each sub-flow segment to generate the load record set of the sub-flow segment; Based on the load record set of the sub-flow segment, the formula is adopted: ； Calculate sub-flow segments Load weighted value ; in, It is a sub-flow segment The total number of time segments, It is a sub-flow segment In time segment The load value in It is a sub-flow segment In time segment The traffic weight in It is a sub-flow segment In time segment Dynamic priority weights.

5. The network data management system based on intelligent flow control according to claim 4 is characterized in that: The steps for obtaining the global load change trend list are specifically as follows: According to the load weighted value, the formula is used: ； Calculate sub-flow segments In the next time segment The predicted load value ; in, It is a time segment inner sub-traffic segment The load growth factor, is the adjustment factor for load change trend, It is a time segment The load increment, is the time segment length; Based on the predicted load value, the load change trends of all sub-flow segments are sorted out in time dimension and space dimension, and a global load change trend list is generated by classifying and integrating the load data of each sub-flow segment.

6. The network data management system based on intelligent flow control according to claim 5, characterized in that: The step of calculating the load deviation value between the sub-flow segments is specifically as follows: Based on the global load change trend list, extract the load forecast value of each sub-flow segment, sort the load values ​​in combination with the time dimension, and generate a load trend data set according to the sorting result; Based on the load trend data set, the formula is used: ； Calculate sub-flow segments and In time segment Load deviation value ; in, It is a sub-flow segment In time segment The predicted load value.

7. The network data management system based on intelligent flow control according to claim 6 is characterized in that: The steps for obtaining the dynamic path flow distribution result are specifically as follows: Based on the load deviation value, the load deviation values ​​between the sub-flow segments are integrated in a matrix form, paths of various load conditions are identified, and corresponding load paths are adjusted to generate sub-flow segment path load distribution results; Based on the sub-flow segment path load distribution results, the formula is adopted: ； Calculate sub-flow segments and Path flow distribution ratio , and according to the sub-flow segment and Path flow distribution ratio Calculate and adjust the path flow , dynamically adjust the load distribution path according to the adjusted path flow, and obtain the dynamic path flow distribution result; in, It is a sub-flow segment The load weighted value of is the initially allocated path flow value.

8. The network data management system based on intelligent flow control according to claim 7 is characterized in that: The steps for obtaining the network traffic data optimization management result are specifically as follows: Based on the dynamic path traffic allocation result, by monitoring the data packet transmission rate in the path and the traffic value in the time interval, comparing the traffic fluctuation with the preset fluctuation threshold range, identifying the source node, target node and propagation path of the traffic anomaly, and generating the abnormal traffic identification result; Based on the abnormal traffic identification result, combined with the real-time load status and bandwidth utilization of the currently available paths, the abnormal traffic is reallocated to the isolated paths, the operating status of the isolated paths is monitored, and the network traffic data optimization management results are generated.

Images