Hub performance detection method and device, equipment and storage medium

Abstract

The invention relates to the technical field of concentrators, and discloses a hub performance detection method and device, equipment and a storage medium. The method comprises the following steps: constructing a port performance data matrix based on a plurality of physical ports of the hub, and performing correlation calculation and port group division to obtain an initial port group division result; calculating the occupation condition of each port group on the bandwidth of the backboard, and carrying out bandwidth competition analysis to obtain a port group bandwidth allocation scheme; carrying out dynamic detection threshold calculation, and constructing a port group performance detection rule; performing detection time slice distribution on each port group to obtain a multi-port group parallel monitoring scheme; and performing real-time detection on the multi-port group parallel monitoring scheme, identifying the performance abnormal propagation link between the port groups, and generating hierarchical alarm information. The method effectively improves the accuracy of the performance detection of the hub, and realizes the accurate identification of the performance abnormal propagation link between the port groups.

Description

Technical Field

[0001] The present invention relates to the technical field of hubs, and in particular to a performance detection method, device, equipment and storage medium for a hub. Background Art

[0002] Traditional hub performance detection methods mainly rely on static monitoring of a single port, which cannot effectively identify the data forwarding relationship between ports, and it is difficult to accurately reflect the actual operation status of the hub in a dynamic network environment. This fragmented detection method leads to inaccurate positioning of performance anomalies and difficulty in tracing the fault propagation path.

[0003] The current hub backplane bandwidth resources often face the problem of multi-port competition, and the traditional detection method lacks a dynamic optimization mechanism for bandwidth resource allocation, which easily causes bandwidth starvation in some port groups, while the bandwidth resources of other port groups are not fully utilized. This unbalanced bandwidth resource allocation seriously affects the overall performance of the hub and increases the probability of network failure. Summary of the invention

[0004] The present invention provides a hub performance detection method, device, equipment and storage medium, which effectively improves the accuracy of hub performance detection and realizes accurate identification of performance abnormality propagation links between port groups.

[0005] In a first aspect, the present invention provides a performance detection method for a hub, the performance detection method for the hub comprising: A port performance data matrix is ​​constructed based on multiple physical ports of the hub, and correlation calculation and port group division are performed to obtain an initial port group division result; Based on the initial port group division result, the occupation of the backplane bandwidth by each port group is calculated, and bandwidth competition analysis is performed to obtain a port group bandwidth allocation plan; Performing dynamic detection threshold calculation on the port group bandwidth allocation scheme to construct port group performance detection rules; Allocate detection time slices to each port group according to the port group performance detection rule to obtain a multi-port group parallel monitoring solution; The multi-port group parallel monitoring scheme is subjected to real-time detection to identify performance anomaly propagation links between port groups and generate graded alarm information.

[0006] In a second aspect, the present invention provides a performance detection device for a hub, the performance detection device for the hub comprising: A construction module is used to construct a port performance data matrix based on multiple physical ports of the hub, and perform correlation calculation and port group division to obtain an initial port group division result; An analysis module, used to calculate the occupancy of the backplane bandwidth by each port group based on the initial port group division result, and perform bandwidth competition analysis to obtain a port group bandwidth allocation plan; A calculation module, used to calculate a dynamic detection threshold for the port group bandwidth allocation scheme and construct a port group performance detection rule; An allocation module, configured to allocate detection time slices to each port group according to the port group performance detection rule, and obtain a multi-port group parallel monitoring solution; A generation module is used to perform real-time detection on the multi-port group parallel monitoring solution, identify performance anomaly propagation links between port groups, and generate graded alarm information.

[0007] The third aspect of the present invention provides a hub performance detection device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory to enable the hub performance detection device to execute the above-mentioned hub performance detection method.

[0008] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned hub performance detection method.

[0009] In the technical solution provided by the present invention, by constructing a port performance data matrix and performing correlation calculation, the intelligent division of port groups is realized, the port set with data forwarding dependency is accurately identified, and the accuracy of performance detection is effectively improved; bandwidth competition analysis and dynamic resource allocation mechanism are adopted to reasonably solve the problem of competition for backplane bandwidth among multiple port groups, and the performance bottleneck caused by uneven bandwidth resource allocation is avoided; a dynamic detection threshold calculation method is introduced, and the detection parameters are adaptively adjusted according to the performance characteristics of different port groups, eliminating the detection deviation caused by the traditional fixed threshold; a multi-port group parallel monitoring mechanism based on time slices is established, and the timeliness and resource utilization of performance detection are significantly improved by optimizing the scheduling strategy of detection tasks; accurate identification of performance abnormality propagation links between port groups is realized, and a clear decision-making basis is provided for fault location and processing through a hierarchical alarm mechanism; an incremental data processing method is adopted, and there is no need to store the full amount of historical data, which reduces the system storage overhead and improves the operation efficiency of the detection system. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0011] Figure 1 A schematic diagram of a flow chart of a method for detecting the performance of a hub provided in an embodiment of the present application; Figure 2 A schematic block diagram of the structure of a performance detection device for a hub provided in an embodiment of the present application; Figure 3 A schematic block diagram of the structure of a performance detection device for a hub provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change based on actual conditions.

[0014] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0015] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0016] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0017] See also Figure 1 , Figure 1 A flow chart of a performance detection method for a hub provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the performance detection method of the hub provided in the embodiment of the present application includes steps S100 to S500.

[0018] Step S100: constructing a port performance data matrix based on multiple physical ports of the hub, and performing correlation calculation and port group division to obtain an initial port group division result; It is understandable that the execution subject of the present invention may be a performance detection device of a hub, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0019] Specifically, a data collection mechanism is established, which records the data packet reception of each physical port in real time. A data collection matrix is ​​created based on multiple physical ports of the hub to store the data packet information received by each port within a specific time, including the size, timestamp, source port, destination port, and protocol type of the data packet, forming an original data packet sequence. The original data packet sequence is filtered by signal-to-noise threshold to filter out data packets with a signal-to-noise ratio higher than the preset threshold, thereby removing abnormal data caused by channel interference, hardware failure or other external factors to obtain network data packets. In order to analyze the data interaction mode between ports, the filtered network data packets are segmented according to the timestamp, the entire data stream is divided into multiple time slices, and the source port and destination port information of the data packet are recorded in each time period. In this way, the data forwarding delay between ports is calculated, that is, the time delay experienced by the statistical data packet from the source port to the destination port. The change of data forwarding delay can reflect the network load, port congestion level, and possible fault hazards. At the same time, in order to comprehensively evaluate the performance status of the port, the data cache usage of each physical port is counted, and the cache queue length of each port is calculated based on the arrival and forwarding status of the network data packet, that is, the number of data packets that are still waiting for forwarding in a certain period of time is counted, thereby reflecting the cache pressure and flow control of the port. To evaluate the port's occupancy of backplane bandwidth resources, the data transmission volume of each port per unit time is calculated based on the network data packet, and it is compared with the total backplane bandwidth of the hub to obtain the port backplane bandwidth occupancy rate, which reflects the consumption of total bandwidth resources by each port and is used to determine whether there is bandwidth competition or resource waste on the port. A port performance data matrix is ​​constructed based on the data forwarding delay between ports, the port cache queue length, and the port backplane bandwidth occupancy rate. The matrix uses physical ports as rows and different performance indicators as columns to comprehensively reflect the overall performance of each port in the data forwarding process. Based on the port performance data matrix, the data forwarding mode of each physical port is analyzed, and the correlation between ports is measured using the correlation calculation method. For example, the Pearson correlation coefficient, mutual information or dynamic time warping (DTW) method is used to evaluate the similarity or dependency between the data forwarding volumes of different ports, so as to determine which ports belong to the same data flow trend or have a strong interactive relationship. Based on the results of the correlation calculation, clustering algorithms such as K-means clustering, hierarchical clustering or DBSCAN density clustering are applied to divide the physical ports into several port groups, so as to conduct more targeted performance monitoring and optimization in the future, and finally obtain the initial port group division results.

[0020] The data forwarding volume of each physical port in the port performance data matrix is ​​accumulated in segments, and the forwarding data volume of each physical port is aggregated within a certain time window to obtain the time series data sequence of the physical port, so as to observe the data flow of the port in the time dimension, so that the subsequent correlation analysis can be calculated based on the change trend of the time series. The time series data sequence of the port is normalized, and the data forwarding volume of all ports is within the same scale range using methods such as minimum-maximum normalization, Z-score standardization or logarithmic transformation to obtain the standardized data forwarding volume. Based on the standardized data forwarding volume, a data forwarding correlation calculation matrix between each physical port is constructed. This matrix is ​​used to measure the similarity of data flow between different ports. The correlation between ports is calculated using methods such as Pearson correlation coefficient, Euclidean distance, dynamic time warping (DTW) or mutual information, and the calculation results are filled into the correlation matrix to quantify the correlation of forwarding data between different ports. In order to screen out the truly relevant port combinations, the data forwarding correlation is compared with the preset threshold. If the correlation between two ports is greater than the threshold, they are considered to belong to a strongly correlated port pair. All port combinations that meet this condition are screened out to obtain a set of associated port pairs. When obtaining the set of associated port pairs, the transitive correlation between ports is considered. For example, if the correlation between port A and port B is high, the correlation between port B and port C is also high. Even if the direct correlation between port A and port C is low, they may still belong to the same port group. Therefore, a transitive closure operation is performed on the set of associated port pairs, that is, through recursive calculation, ports with transitive correlation are merged to construct a more stable initial partition set of port groups. This step can avoid the port group division being too scattered and ensure that the ports within the port group have a strong correlation. All physical ports are grouped and clustered according to the initial partition set of the port group to ensure that the division of the port group can not only reflect the real data interaction relationship between the ports, but also avoid overfitting or over-aggregation, and finally obtain the initial port group division result.

[0021] Step S200: Calculate the occupancy of the backplane bandwidth by each port group based on the initial port group division result, and perform bandwidth competition analysis to obtain a port group bandwidth allocation plan; Specifically, the sum of the data forwarding volume of all ports in each port group is extracted according to the initial port group division result, and the port group bandwidth demand matrix is ​​constructed, which reflects the total amount of bandwidth resources required by each port group per unit time. The port group bandwidth demand matrix is ​​analyzed to determine their actual occupation of backplane bandwidth resources. Based on the matrix, the bandwidth demand of each port group is mapped to the backplane resources of the hub to form a backplane bandwidth resource competition relationship diagram, which reflects the bandwidth competition relationship between different port groups and their demand for shared resources. Based on the backplane bandwidth resource competition relationship diagram, the bandwidth competition coefficient of each port group is calculated, that is, the bandwidth competition intensity of a port group relative to other port groups is measured by calculating the proportion of the port group's demand to the total available bandwidth, or based on the number of ports that share the same backplane resources of each port group. The port group competition intensity matrix is ​​obtained, which reflects the relative advantages or disadvantages of different port groups in resource competition. Based on the port group competition intensity matrix, a backplane bandwidth allocation priority sequence is constructed, that is, the port groups are sorted according to the competition intensity, demand urgency and overall network load of each port group, their priorities in bandwidth scheduling are determined, and the port group bandwidth scheduling order is generated. This order determines the priority of different port groups when competing for bandwidth resources. Bandwidth resource allocation calculation is performed based on the port group bandwidth scheduling order, and strategies such as weighted allocation, dynamic adjustment or fair scheduling are adopted to ensure the reasonable allocation of bandwidth. At the same time, the actual demand, priority and competition of the port group are considered, and the bandwidth allocation ratio of each port group is calculated. Finally, the backplane bandwidth resources are allocated to each port group according to the allocation ratio to ensure that all port groups achieve the best bandwidth utilization efficiency under the premise of meeting the minimum demand, and finally the port group bandwidth allocation plan is obtained.

[0022] Step S300: dynamically calculate the detection threshold of the port group bandwidth allocation scheme and construct a port group performance detection rule; Specifically, the data sequence of the port group bandwidth allocation scheme is divided, that is, the bandwidth allocation data is split according to the time dimension to obtain the bandwidth allocation data sequence of each port group in different time periods, and the dynamic change trend of the port group bandwidth usage is captured. Since bandwidth allocation has time-varying characteristics, it is impossible to effectively monitor the abnormal situation of the port group by relying solely on static thresholds. Therefore, based on the bandwidth allocation data sequence, the data is dynamically detected. The threshold is calculated using sliding window statistics, moving mean, standard deviation calculation and other methods to generate the bandwidth dynamic threshold of the port group. The threshold is adaptively adjusted so that the detection rule adapts to the bandwidth fluctuation under different load conditions and effectively distinguishes normal fluctuations from abnormal conditions. Based on the dynamic threshold of the port group bandwidth, a port group performance parameter calculation function is constructed. The function is used to quantify the performance status of the port group and generate performance parameter calculation rules. For example, according to the indicators such as the instantaneous bandwidth occupancy rate, average bandwidth utilization rate, bandwidth jitter and packet loss rate of the port group, combined with the dynamic threshold calculation formula, the performance evaluation function of the port group is defined, thereby ensuring the consistency and reliability of performance measurement under different load conditions. In order to refine the performance monitoring system of the port group, the performance parameter calculation rules are graded and quantified, that is, the performance parameters are divided into different levels, such as normal, mild abnormality, moderate abnormality and severe abnormality, and the specific thresholds of each level are calculated based on key parameters such as bandwidth utilization, packet loss rate, and delay change to form graded performance detection parameters. All graded performance detection parameters are combined to generate a port group performance parameter set to reflect the different performance states of the port group. According to the port group performance parameter set, the graded threshold value of performance detection is set for each port group to ensure accurate identification of abnormal situations at different levels, and a mapping relationship between each level of performance threshold value and performance degradation strategy is established. When the performance parameter of a port group exceeds a specific threshold value, the corresponding degradation measures are triggered, such as only recording logs in the case of mild abnormality, and limiting bandwidth or switching to backup ports in the case of severe abnormality, so as to ensure the stable operation of the network. After completing the mapping of threshold values ​​and degradation strategies, the performance detection parameter quantization function is updated to adapt to the latest monitoring requirements, and combined with the dynamic adjustment mechanism, it is ensured that the detection rules can be adaptively optimized as the network status changes, and finally the port group performance detection rules are formed.

[0023] Step S400: Allocate detection time slices to each port group according to the port group performance detection rule to obtain a multi-port group parallel monitoring solution; Specifically, the performance level score of each port group is calculated based on the port group performance detection rule, that is, according to the performance of each port group in key performance indicators such as bandwidth occupancy, data packet forwarding delay, cache queue length, packet loss rate, etc., combined with a preset weighted calculation formula or a normalized scoring method based on historical data, the comprehensive performance score of each port group is calculated, and the port groups are prioritized based on the scoring results to form a port group priority sequence, which reflects the importance of different port groups in the overall network environment and the urgency of detection. Due to the different network states and detection requirements of different port groups, relying solely on static allocation strategies will lead to uneven allocation of detection resources or delayed monitoring. Therefore, the detection time slice length of each port group is calculated according to the port group priority sequence. In this process, a dynamic adjustment strategy is adopted, that is, for port groups with higher priorities, a shorter detection time interval is allocated to ensure more frequent monitoring, and for port groups with lower priorities, the detection interval is appropriately increased, thereby optimizing the utilization of the overall monitoring resources and forming a detection time slice allocation result. The detection time slice allocation result is optimized to avoid time conflicts of detection tasks. The execution time overlap check is to analyze whether there is excessive overlap in the detection time slices of different port groups, especially between port groups with fierce bandwidth competition. Reasonable arrangement of detection time can reduce the additional load interference caused by the detection process, thereby ensuring the accuracy of monitoring. By adjusting the distribution of time slices, adjacent time slices can achieve the best parallelism while avoiding competition as much as possible, and generate a port group detection task schedule, which records the specific detection time range of each port group. Based on the port group detection task schedule, a parallel detection task sequence is constructed to maximize the number of tasks executed in parallel while ensuring that the detection tasks between different port groups do not interfere with each other, thereby improving the efficiency of the entire monitoring system, and finally generating a detection scheduling instruction, which contains the detection execution time, detection content and resource usage requirements of each port group. The detection scheduling instructions are sorted and organized according to the port group priority, and a scheduling matrix is ​​constructed to reflect the monitoring order of each port group in different time windows. At the same time, it is ensured that the scheduling is carried out according to the established priority during the execution process, thereby avoiding the decrease in detection accuracy caused by resource competition. Based on the scheduling matrix, the execution plan of parallel detection is optimized, that is, by analyzing the task dependencies of the matrix, it is ensured that subsequent tasks are executed after all necessary detection steps are completed, and finally the parallel detection execution sequence of each port group is generated to determine the specific implementation steps of the detection. After completing the planning of all detection tasks, a multi-port group parallel monitoring plan is generated.

[0024] Step S500: Perform real-time detection on the multi-port group parallel monitoring solution, identify the performance abnormality propagation links between port groups, and generate graded alarm information.

[0025] Specifically, the multi-port group parallel monitoring scheme is converted into a real-time detection execution instruction sequence, and based on the monitoring task scheduling matrix, the detection task of the port group is converted into a specific instruction set to obtain a port group performance data collection sequence, which is used to control each port group to collect data in a predetermined order during the monitoring process, and ensure that all port groups can complete data acquisition within the set time slice, while minimizing the interference between monitoring tasks, thereby ensuring the timeliness and accuracy of the data. Since parameters such as data flow, bandwidth occupancy, and cache status of different port groups are dynamically changing, when performing real-time detection, the performance parameters of each port group are analyzed based on the port group performance data collection sequence, and their deviation values ​​are calculated, that is, the deviation between the current detection data and the expected value of the historical data or the prediction model is compared, so as to identify abnormal situations, and record all performance parameters that exceed the normal range, forming a port group performance abnormality record, reflecting the abnormal trend of the port group within a specific time. Perform time series correlation analysis on the performance anomaly records of the port group. By analyzing the time series of abnormal events, find the causal relationship between the port groups. In particular, when the anomalies of multiple port groups occur in temporal continuity, it indicates that the anomaly of a port group will affect other port groups. Therefore, based on the topology structure, traffic forwarding path and data correlation analysis method, build the performance anomaly propagation link between port groups to determine how the anomaly spreads inside the hub and find the possible source of the anomaly. Since some anomalies will have a chain reaction on multiple port groups, the impact range of the anomaly is evaluated after identifying the anomaly propagation link. Based on the expansion of the anomaly propagation path, build a fault impact range assessment table. The assessment table can integrate multiple indicators, such as the number of ports affected by the anomaly, the duration of the anomaly, the type of anomaly and the degree of impact on the overall network performance, to obtain performance anomaly level data. Determine the alarm trigger condition based on the performance anomaly level data, that is, define how different levels of anomalies should trigger an alarm. Combined with the severity of abnormal propagation, the scope of impact, and the actual impact on network performance, hierarchical alarm triggering rules are constructed. For example, for mild anomalies, only logs are recorded and monitored, while for moderate anomalies, alarms are issued to notify administrators, and for severe anomalies, automated fault recovery measures are taken, such as traffic rerouting or port isolation. After the hierarchical alarm triggering rules are formulated, these rules are combined with the performance anomaly propagation link between port groups to generate the final hierarchical alarm information, which includes the level and scope of the anomaly, as well as the propagation path and possible root causes of the anomaly.

[0026] In the embodiment of the present invention, by constructing a port performance data matrix and performing correlation calculation, the intelligent division of port groups is realized, the port set with data forwarding dependency is accurately identified, and the accuracy of performance detection is effectively improved; bandwidth competition analysis and dynamic resource allocation mechanism are adopted to reasonably solve the problem of competition for backplane bandwidth among multiple port groups, and the performance bottleneck caused by uneven bandwidth resource allocation is avoided; a dynamic detection threshold calculation method is introduced, and the detection parameters are adaptively adjusted according to the performance characteristics of different port groups, eliminating the detection deviation caused by the traditional fixed threshold; a multi-port group parallel monitoring mechanism based on time slices is established, and the timeliness and resource utilization of performance detection are significantly improved by optimizing the scheduling strategy of detection tasks; accurate identification of performance abnormality propagation links between port groups is realized, and a clear decision-making basis is provided for fault location and processing through a hierarchical alarm mechanism; an incremental data processing method is adopted, and there is no need to store the full amount of historical data, which reduces the system storage overhead and improves the operation efficiency of the detection system.

[0027] In a specific embodiment, the process of executing step S100 may specifically include the following steps: Create a data acquisition matrix based on multiple physical ports of the hub, record the real-time data packet reception status of each physical port, and obtain the original data packet sequence; Perform signal-to-noise threshold filtering on the original data packet sequence, filter out data packets with a signal-to-noise ratio higher than a preset threshold, and obtain network data packets; Segment the network data packets according to the data packet timestamp, record the source port and destination port information of the data packet in each time period, and obtain the data forwarding delay between ports; Based on the network data packets, the data buffer usage status of each physical port is counted, the number of data packets waiting to be forwarded is calculated, and the port buffer queue length is obtained; Perform bandwidth resource occupancy statistics on network data packets, calculate the ratio of data transmission volume of each port per unit time to the total backplane bandwidth, and obtain the port backplane bandwidth occupancy rate; Build a port performance data matrix based on the data forwarding delay between ports, the port buffer queue length, and the port backplane bandwidth occupancy rate; The data forwarding volume of each physical port in the port performance data matrix is ​​subjected to correlation calculation and port group division to obtain an initial port group division result.

[0028] Specifically, a data collection mechanism is established for all physical ports in the hub, where each port Record all received data packets, including the arrival time of the data packets , Packet size , Source Port and the destination port , forming a port data acquisition matrix, recorded as:

[0029] in, Indicates the total number of data packets collected per unit time, and Record the key information of each data packet. Since there are noisy data packets in the network environment, the original data packet sequence is filtered by signal-to-noise threshold, that is, the signal-to-noise ratio of each data packet is calculated. And filter out those above the preset threshold Data packets, get a valid set of network data packets:

[0030] in, Defined as:

[0031] in, Indicates data packet The signal power, Represents the noise power. Packets with high signal-to-noise ratio are sorted by timestamp Segmentation, fixed time window Calculate the data forwarding delay between ports, that is, record the source port of the same data flow and the destination port The arrival time difference of data packets defines the data forwarding delay between ports. for:

[0032] in, Indicates the number of packets from the source port within the unit time window. To the target port The number of packets, and Data packets Timestamps at the source and destination ports. To evaluate the port cache usage, count the cache queue length of each physical port, that is, calculate the number of packets still waiting to be forwarded within a unit time window. Define ports In time The length of the cache queue at for:

[0033] in, Represents time window In-port arrival The number of packets, Indicates the number of packets successfully forwarded within the time window. The length of the cache queue is used to reflect the traffic load of the port and help detect congestion. In order to evaluate the bandwidth resource usage, the bandwidth usage of each port is calculated, that is, the data transmission volume of the port per unit time and the total backplane bandwidth. Define the port In the time window Data transfer volume within for:

[0034] Then calculate the backplane bandwidth utilization of the port for:

[0035] in, Representative port exist The amount of data transferred in a given period of time, is the total backplane bandwidth of the hub. If If the value is too high, it indicates that the port is in a bandwidth competition state. Build a port performance data matrix based on the data forwarding delay between ports, the length of the port buffer queue, and the port backplane bandwidth occupancy rate. , where each row of the matrix corresponds to a physical port and the columns contain these key performance indicators:

[0036] in, represents the total number of physical ports, and , and Corresponding ports Data forwarding delay, cache queue length and bandwidth occupancy rate. In order to analyze the correlation between ports, the port performance data matrix The data forwarding volume of each physical port is correlated and the port groups are divided accordingly. Compute Port and Port The correlation between:

[0037] in, and Respectively represent ports and In time The amount of data forwarding at and is their respective means. Greater than a preset threshold , then the port is considered and Port The ports with strong correlation are classified into the same port group. Based on the calculated correlation matrix, a clustering algorithm (such as K-means or hierarchical clustering) is used to divide the ports into several port groups to obtain the initial port group division results. For example, if the following port groups are obtained in a calculation:

[0038] The port They are similar in data transmission characteristics and belong to the same group. Form another group, The port groups are used to optimize bandwidth allocation, fault detection, and traffic management, thereby improving the overall performance of the hub.

[0039] Among them, constructing a port performance data matrix according to the data forwarding delay between ports, the port cache queue length and the port backplane bandwidth occupancy rate also includes: segmenting the port performance data matrix according to the time dimension to obtain a time-divided performance data set; performing probability statistical analysis on the time-divided performance data set, establishing a port performance state migration probability graph, and obtaining a port state transition model; constructing a causal reasoning network based on the port state transition model, analyzing the change law of the port performance parameters, and obtaining a performance change causal chain; inputting the performance change causal chain into the reverse reachability analysis model, tracing the propagation path of the performance anomaly, and obtaining the port performance anomaly tracing result; performing performance degradation risk assessment on each physical port according to the port performance anomaly tracing result, and obtaining a port performance risk level sequence; establishing a performance warning threshold system based on the port performance risk level sequence, and obtaining a multi-level performance warning rule; performing parameter mapping on the multi-level performance warning rule, constructing a port performance prediction model, and obtaining a port performance trend prediction result; performing deviation analysis on the port performance trend prediction result and the real-time performance data, updating the port performance data matrix, and completing the accurate evaluation of the port performance.

[0040] In a specific embodiment, the execution step calculates the correlation degree and divides the data forwarding volume of each physical port in the port performance data matrix into port groups, and the process of obtaining the initial port group division result may specifically include the following steps: Accumulate the data forwarding amount of each physical port in the port performance data matrix in sections to obtain a physical port timing data sequence; Normalizing the data forwarding volume of each port in the physical port timing data sequence to obtain a standardized data forwarding volume; Based on the standardized data forwarding volume, a data forwarding correlation calculation matrix between physical ports is constructed to obtain the data forwarding correlation; Compare the data forwarding correlation with a preset threshold, filter out the port combination whose data forwarding correlation is greater than the preset threshold, and obtain a set of associated port pairs; A transitive closure operation is performed on the associated port pair set, ports with a transitive association relationship are merged to obtain an initial port group partition set, and all physical ports are grouped and clustered according to the initial port group partition set to obtain an initial port group partition result.

[0041] Specifically, from the port performance data matrix Extract the data forwarding volume of each physical port and calculate the data forwarding volume in a fixed time window. The data are accumulated in , to form the timing data sequence of the physical port. Indicates the port In time The instantaneous data forwarding volume at the time window The cumulative calculation within is expressed as:

[0042] in, Representative port In time period The cumulative data forwarding volume within is the segment start time, and is the length of the time window. This accumulation method can reduce the impact of instantaneous fluctuations, so that the data forwarding volume of the port shows a more stable timing characteristic and can be better used for subsequent correlation analysis. The timing data sequence of the physical port is normalized to ensure that the data forwarding volume of all ports has the same numerical scale, to avoid the influence of some ports being amplified or ignored in subsequent calculations, and to obtain the standardized data forwarding volume. The normalization method includes minimum-maximum normalization and Z-score normalization, where the formula for Min-Max normalization is:

[0043] in, Yes Port In time The standardized data forwarding volume at and Respectively represent the minimum and maximum data forwarding volume of all ports in the time period. Based on the standardized data forwarding volume, a data forwarding correlation calculation matrix between each physical port is constructed to measure the similarity of data transmission between different ports. The Pearson correlation coefficient is used to calculate the port and Port The correlation between them is as follows:

[0044] in, and Port and The average normalized data forwarding volume over all time periods. The value range is between [-1,1]. When the value is close to 1, it means that the data forwarding behaviors of the two ports are strongly correlated. When the value is close to -1, it means that the behaviors between the two are negatively correlated. When it is close to 0, it means that there is no obvious correlation between the two. For comparison, if , then the port is considered and There is a strong association between them, so they are included in the associated port pair set:

[0045] After obtaining the set of associated port pairs, in order to identify the indirectly associated port group, a transitive closure operation is performed, that is, if the port With port associated, and the port And port If related, then it is considered and There is also some indirect association, so these ports are merged into the same port group. The transitive closure is calculated using the adjacency matrix method. Let is the adjacency matrix of the associated port pairs, where Indicates the port and Port If we associate Calculated by the following recursive formula:

[0046] In getting the transitive closure matrix Finally, based on the concept of connected components, all ports with transitive associations are merged to form an initial partition set of port groups. Clustering algorithms are used to optimize the port groups, such as K-means or hierarchical clustering algorithms, to obtain a more stable initial port group partition result.

[0047] Among them, before obtaining the initial port group division result and calculating the occupancy of each port group on the backplane bandwidth based on the initial port group division result, it also includes: constructing a port group communication topology diagram based on the initial port group division result, quantifying the data transmission delay between each port group as the weight of the connection edge, and obtaining a port group weighted topology diagram; performing connectivity detection on the port group nodes in the port group weighted topology diagram, calculating the shortest path set between the port groups, and obtaining the port group connectivity matrix; hierarchically dividing the port group connectivity matrix according to the size of the communication delay, using the minimum spanning tree algorithm to calculate the optimal communication link, and obtaining the initial topology optimization result; and performing distributed collaborative optimization on the initial topology optimization result according to the data flow. Division, build a multi-center distributed topology optimization model, and obtain a topology division scheme; classify each port group in layers according to the topology division scheme, establish a master-slave relationship mapping table between port groups, and obtain a hierarchical port group structure; calculate the data flow coordination coefficient between port groups based on the hierarchical port group structure, build a distributed collaborative control compensation model, and obtain a collaborative control compensation matrix; combine the collaborative control compensation matrix with the hierarchical port group structure, establish an adaptive topology optimization rule base, and obtain a port group topology optimization scheme; update the data forwarding path of the port group according to the port group topology optimization scheme, optimize and reconstruct the communication link between the port groups, and obtain the optimized port group division result.

[0048] In a specific embodiment, the process of executing step S200 may specifically include the following steps: According to the initial port group division result, the sum of data forwarding volume in each port group is extracted to obtain the port group bandwidth requirement matrix; Perform backplane resource mapping on the bandwidth requirements of each port group in the port group bandwidth requirement matrix to obtain a backplane bandwidth resource competition relationship diagram; Based on the backplane bandwidth resource competition relationship diagram, the bandwidth competition coefficient of each port group is calculated to obtain the port group competition intensity matrix; Construct a backplane bandwidth allocation priority sequence based on the port group competition intensity matrix to obtain the port group bandwidth scheduling order; Based on the bandwidth scheduling order of the port group, bandwidth resource allocation calculation is performed to obtain the bandwidth allocation ratio of each port group, and backplane bandwidth resources are allocated to each port group according to the bandwidth allocation ratio to obtain a port group bandwidth allocation plan.

[0049] Specifically, from the port performance data matrix Extract the data forwarding volume of each port and add them up according to the port group to calculate the total data forwarding volume of each port group. Assume that the total number of ports in the hub is , the total number of port groups is , then each port In unit time The data forwarding volume in is expressed as:

[0050] in, Representative port In the time window The total amount of data forwarded within Is the port at time The instantaneous data transmission volume at a port group. , its total bandwidth requirement is calculated by accumulating the bandwidth requirements of all ports belonging to the port group, that is:

[0051] Constructing a port group bandwidth requirement matrix , each row of the matrix corresponds to a port group, containing the total bandwidth requirements of the port group in different time windows:

[0052] Map the bandwidth requirements of each port group in the port group bandwidth requirement matrix to the backplane resources to construct a backplane bandwidth resource competition relationship diagram. Assume that the total backplane bandwidth of the hub is The bandwidth competition of each port group on the backplane is represented by an adjacency matrix Indicates that Represents a port group and Whether to share the same backplane resources, and is defined as follows:

[0053] If two port groups and Between , it means that there is bandwidth competition between the two port groups, and if , it means that they use independent resources and will not affect each other. After establishing the bandwidth resource competition relationship diagram, calculate the bandwidth competition coefficient of each port group to measure the relative pressure of each port group in resource competition. Bandwidth competition coefficient Define as port group The ratio of the bandwidth demand of the port to the total demand of its competing port group, that is:

[0054] in, Indicates the port group The collection of all port groups that have bandwidth competition. If a port group has a high bandwidth competition coefficient, it is at a disadvantage in resource competition and needs a higher priority to ensure its bandwidth needs. Based on the port group bandwidth competition coefficient, the port group competition intensity matrix is ​​constructed. , where each element Reflects the port group Relative to port group Competitive pressure:

[0055] The backplane bandwidth allocation priority sequence is constructed based on the port group competition intensity matrix, that is, all port groups are sorted by competition intensity, and the port groups with higher priorities will obtain more resources when bandwidth is allocated. Priority sequence Arranged in descending order of competitive intensity, they are defined as:

[0056] Among them, argsort means according to The index values ​​are sorted from large to small to obtain the bandwidth scheduling order of the port group. After determining the bandwidth scheduling order of the port group, specific bandwidth resource allocation calculations are performed to ensure that the needs of each port group are reasonably met. Define port groups Bandwidth allocation ratio for:

[0057] in, Indicates a port group The allocated backplane bandwidth is affected by the contention intensity and ensures that the total allocated bandwidth of all port groups does not exceed the backplane bandwidth limit, that is, it meets the following constraints:

[0058] If the demand of a port group exceeds its allocated bandwidth, the scheduling strategy is further adjusted, such as introducing a bandwidth reservation mechanism to reserve the minimum bandwidth for the key business port group. Based on the calculated bandwidth allocation ratio, the backplane bandwidth resources are allocated to each port group to form a port group bandwidth allocation plan.

[0059] In this embodiment, a two-layer deep reinforcement learning agent model is used for the generation process of the port group bandwidth allocation scheme, including: taking the port group bandwidth scheduling order as input, respectively constructing a high-level policy network and a low-level execution network, wherein the high-level policy network adopts a multi-head attention structure based on Transformer, and the low-level execution network adopts a long short-term memory network structure; the high-level policy network includes three encoder blocks, each encoder block includes a multi-head self-attention layer and a feedforward neural network layer, the number of heads of the multi-head self-attention layer is 8, the hidden layer dimension is 512, and the feedforward neural network layer includes two fully connected layers; the low-level execution network includes two long short-term memory layers, each layer includes 128 neurons, and a fully connected layer with a dimension of 256 and an output dimension is added after the long short-term memory layer. The prediction layer with the degree being the number of port groups; the bandwidth scheduling sequence of the port groups is input into the high-level policy network, and the bandwidth allocation strategy is generated according to the historical bandwidth usage data of the port groups to obtain the global bandwidth allocation guidance plan; the global bandwidth allocation guidance plan is input into the low-level execution network, and the specific bandwidth allocation action is generated according to the real-time bandwidth demand of the port groups to obtain the fine-grained bandwidth allocation sequence; based on the fine-grained bandwidth allocation sequence, the port group performance indicators are calculated, including bandwidth utilization, port service quality and load balancing, to obtain the bandwidth allocation optimization evaluation result; according to the bandwidth allocation optimization evaluation result, the global bandwidth allocation guidance plan of the high-level policy network is adjusted to generate a new port group bandwidth allocation strategy; the new port group bandwidth allocation strategy is combined with the fine-grained bandwidth allocation sequence to obtain the final port group bandwidth allocation plan.

[0060] In a specific embodiment, the process of executing step S300 may specifically include the following steps: The bandwidth allocation scheme of the port group is divided into data sequences to obtain bandwidth allocation data sequences of each port group in different time periods; A dynamic detection threshold is calculated based on the bandwidth allocation data sequence to obtain a dynamic threshold of the port group bandwidth; Construct a port group performance parameter calculation function according to the dynamic threshold of the port group bandwidth to obtain a performance parameter calculation rule; Performing hierarchical quantization operations on the performance parameter calculation rules to obtain hierarchical performance detection parameters, and combining the hierarchical performance detection parameters to generate a port group performance parameter set; According to the port group performance parameter set, a performance detection grading threshold is set for each port group, a mapping relationship between each level of performance threshold and a performance degradation strategy is established, and a performance detection parameter quantization function is updated to obtain a port group performance detection rule.

[0061] Specifically, the allocated bandwidth of each port group is extracted from the bandwidth allocation scheme and allocated according to the fixed time window. The bandwidth allocation timing data of the port group is formed by dividing. Assuming a port group In time The bandwidth allocation at , then in the time window The bandwidth allocation data sequence in is expressed as:

[0062] in, Represents a port group In the time interval The cumulative bandwidth allocation within is the start time of the time window. This segmentation method effectively captures the time variation trend of the bandwidth allocation of the port group and provides a data basis for dynamic detection threshold calculation. Dynamic detection threshold calculation is performed on the bandwidth allocation data sequence of the port group to distinguish normal bandwidth fluctuations from abnormal bandwidth usage. Dynamic threshold calculation is based on statistical methods, such as the sliding mean and standard deviation method to set reasonable upper and lower limits. Define port groups The sliding mean bandwidth and standard deviation for:

[0063]

[0064] in, Indicates the past The number of data points in a time window. These statistics are used to define the bandwidth dynamic threshold of the port group.

[0065]

[0066] in, is the threshold adjustment factor, set to To cover 95% of the normal fluctuation range. If the bandwidth allocation value of a port group exceeds this range, it is considered abnormal. Construct a port group performance parameter calculation function based on the port group bandwidth dynamic threshold to form a performance parameter calculation rule. Define the performance deviation of the port group for:

[0067] when When it is smaller, it means that the bandwidth allocation is close to the historical average level, while a larger value means an abnormal situation. Based on this calculation rule, a hierarchical quantization operation is performed to divide the bandwidth deviation value into multiple levels. The hierarchical performance detection parameter combination forms a port group performance parameter set, where each port group has a specific anomaly detection rule and is used for subsequent anomaly detection and alarm. According to the port group performance parameter set, a performance detection hierarchical threshold value is set for each port group, and a mapping relationship between each level of performance threshold value and performance degradation strategy is established. For example, when When it is in the mild abnormal range, only log warnings are recorded. When it enters the moderate abnormal range, management notifications are triggered, and bandwidth adjustment policies are implemented in serious abnormalities, such as reducing the bandwidth priority of the port group or switching to a backup link. After completing the mapping of threshold values ​​and degradation policies, the performance detection parameter quantization function is updated to ensure that the detection rules can adapt to network changes. The exponentially weighted moving average method is used to update the historical mean and standard deviation to dynamically adjust the threshold.

[0068]

[0069] in, is the smoothing coefficient, usually 0.1-0.3 to adapt to short-term fluctuations. This dynamic update method can adapt the port group performance detection rules to different network environments and provide more accurate alarms when anomalies occur.

[0070] In a specific embodiment, the process of executing step S400 may specifically include the following steps: Calculate the performance level score of each port group based on the port group performance detection rule to obtain the port group priority sequence; Calculate the detection time slice length of each port group according to the port group priority sequence to obtain the detection time slice allocation result; Performing a time overlap check on the detection time slice allocation result to obtain a port group detection task schedule, and constructing a parallel detection task sequence based on the port group detection task schedule to obtain a detection scheduling instruction; The detection scheduling instructions are sorted and arranged according to the port group priority to obtain a scheduling matrix, and the parallel detection execution timing of each port group is generated according to the scheduling matrix to obtain a multi-port group parallel monitoring solution.

[0071] Specifically, a performance scoring model for a port group is defined. The model should integrate multiple key performance indicators, including bandwidth utilization, data forwarding delay, cache queue length, packet loss rate, etc. Assume that the port group In time The performance indicators of the (Bandwidth occupancy), (data forwarding delay), (buffer queue length) and (packet loss rate), then the performance level score of the port group Defined as:

[0072] in, is the weight of different indicators, which is set according to the network management strategy. For example, if the bandwidth utilization rate has a greater impact on network performance, then After calculating the performance scores of all port groups, sort them according to the scores to form a port group priority sequence:

[0073] in, An array of performance scores for all port groups. The argsort() operation is used to sort the scores from high to low to ensure that port groups with higher performance scores receive higher detection priority. The detection time slice length of each port group is calculated based on the port group priority sequence to ensure that key port groups receive more intensive monitoring. Define port groups Detection time slice Inversely proportional to its performance score, namely:

[0074] in, and are the maximum and minimum detection time slice lengths, respectively, and and are the maximum and minimum performance scores of all port groups respectively. The port group with a higher performance score has a shorter detection time slice. The shorter the time, the higher the detection frequency, while the port group with a lower score is detected less frequently to optimize the utilization efficiency of detection resources. Perform a time overlap check on the detection time slice allocation results to ensure that there are not too many tasks executed concurrently, resulting in insufficient detection resources. Build a time window collection , set the detection time slice of all port groups Map it to the timeline. And check whether there is too much overlap, that is, whether the number of detection tasks in the same time period exceeds the set maximum parallel detection capacity If a certain period of time Number of tasks at satisfy:

[0075] in, Represents a port group The detection time window, If is an indicator function, the time slices are adjusted to make the tasks more evenly distributed on the time axis. For example, a greedy adjustment algorithm is used to preferentially move the port group detection time slices with lower priority to the idle time period, thereby optimizing the overall detection scheduling. After completing the time overlap check, a parallel detection task sequence is constructed based on the port group detection task schedule to form a detection scheduling instruction. The detection scheduling instruction uses a matrix Indicates that Represents a port group In time slice Whether the detection task within is executed, that is:

[0076] Sort the detection scheduling instructions according to the port group priority to build a scheduling matrix :

[0077] in, Indicates the total number of time slices, Indicates the port groups sorted by priority. represents the length of its detection time slice, and Record the execution status of the port group in different time slices. Based on the scheduling matrix , and finally generate the parallel detection execution sequence of each port group, that is, arrange parallel execution detection tasks at different time points. Definition Represents at time The port group set executed at , then:

[0078] Through reasonable task arrangement, we can ensure that the detection tasks are evenly distributed on the timeline, reduce task conflicts as much as possible, and improve detection efficiency. Generate the optimal multi-port group parallel monitoring solution.

[0079] In a specific embodiment, the process of executing step S500 may specifically include the following steps: Convert the multi-port group parallel monitoring scheme into a real-time detection execution instruction sequence to obtain a port group performance data collection sequence; Detect the performance parameter deviation value of each port group according to the port group performance data collection sequence, and obtain the port group performance abnormality record; Perform time-series correlation analysis on the abnormal performance records of the port groups to obtain the performance abnormality propagation links between the port groups; A fault impact range assessment table is constructed based on the performance anomaly propagation link between port groups to obtain performance anomaly level data; Determine the alarm triggering conditions based on the performance anomaly level data and obtain the graded alarm triggering rules; The hierarchical alarm triggering rules are combined with the performance anomaly propagation links between port groups to obtain hierarchical alarm information.

[0080] Specifically, based on the scheduling matrix Parse the detection tasks of each port group and convert them into specific execution instructions. Assume that the total number of port groups is , the total number of detection time slices is , the scheduling matrix Recorded port group In different time slices The detection execution sequence is defined according to the detection execution status within for:

[0081] Based on this, the port group performance data collection sequence It is expressed as:

[0082] in, Indicates at time The set of port groups to be tested. This sequence is used to drive the real-time detection system to collect performance parameters for the port groups according to the preset scheduling plan. After obtaining the port group performance data collection sequence, the performance parameter deviation value of each port group is detected to determine whether there is an abnormality. Define port groups In time The performance parameter set at , including bandwidth utilization , Data forwarding delay , Cache queue length and packet loss rate The deviation value is The calculation is as follows:

[0083] in, and Respectively represent port groups The mean and standard deviation in the historical time window. Exceeding the threshold , it is recorded as a performance exception:

[0084] Generate port group performance abnormality records, including Represents a port group In time Whether anomalies occur. Perform time-series correlation analysis on performance anomaly records to identify performance anomaly propagation links between port groups. Define anomaly propagation matrix ,in Represents a port group and The degree of abnormal correlation between: ; in, represents the time interval for the anomaly to propagate, Representative from arrive A data flow path exists. If ,illustrate The exception is propagated to Therefore, the performance anomaly propagation link between port groups is constructed. After obtaining the anomaly propagation link, a fault impact range assessment table is constructed to determine the anomaly level data. Define port groups In time The scope of influence The number of port groups that it directly affects:

[0085] At the same time, define the severity of the abnormality As the anomaly spreads, the cumulative impact is:

[0086] in, Indicates the maximum time window for backtracking. If the value is higher than the preset threshold, the abnormal impact is considered to be large and the alarm level is increased. After obtaining the performance abnormality level data, determine the alarm trigger conditions to generate the graded alarm trigger rules. Define the alarm level :

[0087] in, and The thresholds for medium and high-level alarms are respectively. The hierarchical alarm triggering rules are combined with the performance anomaly propagation link between port groups to generate the final hierarchical alarm information. Define alarm information :

[0088] in, Contains port groups The alarm level, the scope of abnormal propagation, and the set of affected port groups. Different fault recovery measures are taken according to the alarm information. For example, only logs are recorded in the case of low-level alarms, and operation and maintenance personnel are notified in the case of medium-level alarms. In the case of high-level alarms, automatic flow control or rerouting is required to prevent the abnormal impact from expanding.

[0089] See also Figure 2 , Figure 2A schematic block diagram of the structure of the hub performance detection device 200 provided in the embodiment of the present application is shown as follows: Figure 2 As shown, the performance detection device 200 of the hub includes: A construction module 210 is used to construct a port performance data matrix based on multiple physical ports of the hub, and perform correlation calculation and port group division to obtain an initial port group division result; The analysis module 220 is used to calculate the occupancy of the backplane bandwidth by each port group based on the initial port group division result, and perform bandwidth competition analysis to obtain a port group bandwidth allocation plan; A calculation module 230 is used to calculate a dynamic detection threshold for a port group bandwidth allocation scheme and construct a port group performance detection rule; An allocation module 240 is used to allocate detection time slices to each port group according to the port group performance detection rule to obtain a multi-port group parallel monitoring solution; The generating module 250 is used to perform real-time detection on the multi-port group parallel monitoring solution, identify the performance abnormality propagation link between the port groups, and generate graded alarm information.

[0090] Through the collaborative cooperation of the above-mentioned components, by constructing a port performance data matrix and calculating the correlation, the intelligent division of port groups is realized, the port set with data forwarding dependency is accurately identified, and the accuracy of performance detection is effectively improved; the bandwidth competition analysis and dynamic resource allocation mechanism are adopted to reasonably solve the problem of competition for backplane bandwidth among multiple port groups, and avoid the performance bottleneck caused by uneven allocation of bandwidth resources; the dynamic detection threshold calculation method is introduced to adaptively adjust the detection parameters according to the performance characteristics of different port groups, eliminating the detection deviation caused by the traditional fixed threshold; a multi-port group parallel monitoring mechanism based on time slices is established, and the timeliness and resource utilization of performance detection are significantly improved by optimizing the scheduling strategy of detection tasks; the accurate identification of performance anomaly propagation links between port groups is realized, and a clear decision-making basis is provided for fault location and processing through a hierarchical alarm mechanism; the incremental data processing method is adopted, and there is no need to store the full amount of historical data, which reduces the system storage overhead and improves the operation efficiency of the detection system.

[0091] See also Figure 3 , Figure 3 The performance detection device 300 of the hub provided in the embodiment of the present application is a schematic block diagram of the structure, wherein the performance detection device 300 of the hub includes a processor 301 and a memory 302, and the processor 301 and the memory 302 are connected via a device bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0092] The non-volatile storage medium can store a computer program. The computer program includes program instructions. When the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned hub performance detection methods.

[0093] The processor 301 is used to provide computing and control capabilities to support the operation of the performance detection device 300 of the entire hub.

[0094] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned hub performance detection methods.

[0095] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the performance detection device 300 of the hub involved in the scheme of the present application. The specific performance detection device 300 of the hub may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0096] It should be understood that the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0097] It should be noted that technicians in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working process of the hub performance detection device 300 described above can refer to the corresponding process of the aforementioned hub performance detection method, and will not be repeated here.

[0098] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement the performance detection method of the hub provided in the embodiment of the present application.

[0099] The computer-readable storage medium may be an internal storage unit of the performance detection device 300 of the hub in the aforementioned embodiment, such as a hard disk or memory of the performance detection device 300 of the hub. The computer-readable storage medium may also be an external storage device of the performance detection device 300 of the hub, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped with the performance detection device 300 of the hub.

[0100] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0101] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0102] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A hub performance detection method, characterized in that: include: A port performance data matrix is ​​constructed based on multiple physical ports of the hub, and correlation calculation and port group division are performed to obtain an initial port group division result; Based on the initial port group division result, the occupation of the backplane bandwidth by each port group is calculated, and bandwidth competition analysis is performed to obtain a port group bandwidth allocation plan; Performing dynamic detection threshold calculation on the port group bandwidth allocation scheme to construct port group performance detection rules; Allocate detection time slices to each port group according to the port group performance detection rule to obtain a multi-port group parallel monitoring solution; The multi-port group parallel monitoring scheme is subjected to real-time detection to identify performance anomaly propagation links between port groups and generate graded alarm information.

2. The hub performance detection method according to claim 1, characterized in that: The port performance data matrix is ​​constructed based on multiple physical ports of the hub, and correlation calculation and port group division are performed to obtain an initial port group division result, including: Create a data acquisition matrix based on multiple physical ports of the hub, record the real-time data packet reception status of each physical port, and obtain the original data packet sequence; Performing signal-to-noise threshold filtering on the original data packet sequence, screening out data packets with a signal-to-noise ratio higher than a preset threshold, and obtaining network data packets; Segmenting the network data packets according to the data packet timestamps, recording the source port and destination port information of the data packets in each time period, and obtaining the data forwarding delay between ports; Based on the network data packets, the data buffer usage status of each physical port is counted, the number of data packets waiting to be forwarded is calculated, and the port buffer queue length is obtained; Perform bandwidth resource occupancy statistics on the network data packets, calculate the ratio of the data transmission volume of each port per unit time to the total backplane bandwidth, and obtain the port backplane bandwidth occupancy rate; Constructing a port performance data matrix according to the inter-port data forwarding delay, the port buffer queue length and the port backplane bandwidth occupancy rate; The data forwarding volume of each physical port in the port performance data matrix is ​​subjected to correlation calculation and port group division to obtain an initial port group division result.

3. The hub performance detection method according to claim 2, characterized in that: The step of calculating the correlation and dividing the data forwarding volume of each physical port in the port performance data matrix into port groups to obtain an initial port group division result includes: Accumulate the data forwarding amount of each physical port in the port performance data matrix in sections to obtain a physical port timing data sequence; Normalizing the data forwarding amount of each port in the physical port timing data sequence to obtain a standardized data forwarding amount; Based on the standardized data forwarding amount, a data forwarding correlation calculation matrix between the physical ports is constructed to obtain a data forwarding correlation; Compare the data forwarding correlation with a preset threshold, filter out the port combination whose data forwarding correlation is greater than the preset threshold, and obtain a set of associated port pairs; A transitive closure operation is performed on the associated port pair set, ports with a transitive association relationship are merged to obtain an initial port group partition set, and all physical ports are grouped and clustered according to the initial port group partition set to obtain an initial port group partition result.

4. The hub performance detection method according to claim 1, characterized in that: The calculating the occupancy of the backplane bandwidth by each port group based on the initial port group division result, and performing bandwidth competition analysis to obtain a port group bandwidth allocation scheme includes: Extracting the sum of data forwarding volume in each port group according to the initial port group division result to obtain a port group bandwidth requirement matrix; Performing backplane resource mapping on the bandwidth requirements of each port group in the port group bandwidth requirement matrix to obtain a backplane bandwidth resource competition relationship diagram; Calculate the bandwidth competition coefficient of each port group based on the backplane bandwidth resource competition relationship diagram to obtain a port group competition intensity matrix; Constructing a backplane bandwidth allocation priority sequence according to the port group competition intensity matrix to obtain a port group bandwidth scheduling order; Based on the port group bandwidth scheduling sequence, bandwidth resource allocation calculation is performed to obtain the bandwidth allocation ratio of each port group, and backplane bandwidth resources are allocated to each port group according to the bandwidth allocation ratio to obtain a port group bandwidth allocation scheme.

5. The hub performance detection method according to claim 1, characterized in that: The dynamically calculating the detection threshold value of the port group bandwidth allocation scheme and constructing the port group performance detection rule includes: Divide the port group bandwidth allocation scheme into data sequences to obtain bandwidth allocation data sequences of each port group in different time periods; Perform dynamic detection threshold calculation based on the bandwidth allocation data sequence to obtain a dynamic threshold of the port group bandwidth; Constructing a port group performance parameter calculation function according to the port group bandwidth dynamic threshold to obtain a performance parameter calculation rule; Performing hierarchical quantization operations on the performance parameter calculation rules to obtain hierarchical performance detection parameters, and combining the hierarchical performance detection parameters to generate a port group performance parameter set; According to the port group performance parameter set, a performance detection grading threshold is set for each port group, a mapping relationship between each level of performance threshold and a performance degradation strategy is established, and a performance detection parameter quantization function is updated to obtain a port group performance detection rule.

6. The hub performance detection method according to claim 1, characterized in that: The method of allocating detection time slices to each port group according to the port group performance detection rule to obtain a multi-port group parallel monitoring solution includes: Calculate the performance level score of each port group based on the port group performance detection rule to obtain a port group priority sequence; Calculate the detection time slice length of each port group according to the port group priority sequence to obtain a detection time slice allocation result; Performing a time overlap check on the detection time slice allocation result to obtain a port group detection task schedule, and constructing a parallel detection task sequence based on the port group detection task schedule to obtain a detection scheduling instruction; The detection scheduling instructions are sorted and arranged according to the port group priorities to obtain a scheduling matrix, and a parallel detection execution sequence of each port group is generated according to the scheduling matrix to obtain a multi-port group parallel monitoring solution.

7. The hub performance detection method according to claim 1, characterized in that: The performing real-time detection on the multi-port group parallel monitoring scheme, identifying the performance abnormality propagation link between the port groups, and generating graded alarm information includes: Converting the multi-port group parallel monitoring scheme into a real-time detection execution instruction sequence to obtain a port group performance data collection sequence; Detecting the performance parameter deviation value of each port group according to the port group performance data collection sequence to obtain the port group performance abnormality record; Performing time series correlation analysis on the abnormal performance records of the port groups to obtain the abnormal performance propagation links between the port groups; Building a fault impact range assessment table based on the inter-port group performance anomaly propagation link to obtain performance anomaly level data; Determine an alarm trigger condition according to the performance abnormality level data to obtain a graded alarm trigger rule; The hierarchical alarm triggering rule is combined with the inter-port group performance abnormality propagation link to obtain hierarchical alarm information.

8. A performance detection device for a hub, characterized in that: A method for performing a performance detection of a hub according to any one of claims 1 to 7, comprising: A construction module is used to construct a port performance data matrix based on multiple physical ports of the hub, and perform correlation calculation and port group division to obtain an initial port group division result; An analysis module, used to calculate the occupancy of the backplane bandwidth by each port group based on the initial port group division result, and perform bandwidth competition analysis to obtain a port group bandwidth allocation plan; A calculation module, used to calculate a dynamic detection threshold for the port group bandwidth allocation scheme and construct a port group performance detection rule; An allocation module, configured to allocate detection time slices to each port group according to the port group performance detection rule, and obtain a multi-port group parallel monitoring solution; A generation module is used to perform real-time detection on the multi-port group parallel monitoring solution, identify performance anomaly propagation links between port groups, and generate graded alarm information.

9. A performance detection device for a hub, characterized in that: The performance detection device of the hub includes: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the performance detection device of the hub to execute the performance detection method of the hub according to any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the performance detection method of the hub as described in any one of claims 1 to 7 is implemented.

Images