Management method and device for switch port queue, and medium

Abstract

The invention discloses a switch port queue management method and device and a medium, and relates to the technical field of computers, and the method comprises the steps: collecting queue flow data of a switch port in real time, generating a latest multi-dimensional feature based on the latest queue flow data, and monitoring the state of the latest multi-dimensional feature; the newest multi-dimensional features comprise a load state quantity, a burst state quantity and a data packet distribution state quantity; when it is monitored that state transition exists in the newest multi-dimensional features, a queue adjustment event is triggered, a corresponding queue adjustment strategy is matched according to the newest multi-dimensional features, and target strategy parameters are output; adjusting operation parameters of the switch port based on the target strategy parameters; obtaining current operation parameters of a preset time interval, and monitoring whether parameter deviation exists or not; by monitoring the multi-dimensional characteristics in real time and dynamically triggering strategy adjustment, the switch can automatically match the most adaptive management strategy according to the load, burst and distribution characteristics of the current flow.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device, and medium for managing a switch port queue. Background Art

[0002] Data center networks, as a core component of modern information technology architecture, host a wide variety of complex applications. In modern network architectures, switches, as key nodes for data transmission, support increasingly diverse business scenarios. With the continuous emergence of new network applications, particularly those driven by big data mining and machine learning, network traffic is experiencing dynamic changes and widely varying traffic characteristics. The mixed transmission of different types of traffic further exacerbates the complexity of these traffic characteristics.

[0003] Traditional queue management mechanisms, such as random early drop and weighted random early drop, often operate based on preset parameters or static rules. These algorithms require manual configuration of packet parameters for different queues, making it difficult to respond to dynamic changes in real-time traffic perception. This can easily lead to congestion and spread during traffic surges, while potentially causing resource idleness during periods of low traffic levels. The optimal set of parameters for service throughput often yields unsatisfactory results in packet loss rate and queuing delay. A single set of parameters is also unsuitable for complex and changing network traffic environments, lacking the ability to dynamically adjust based on real-time traffic conditions. This makes it difficult to strike a balance between ensuring quality of service and achieving efficient use of network resources. Summary of the Invention

[0004] In order to solve the above problems, this application proposes a switch port queue management method, including:

[0005] Collect queue flow data of the switch port in real time, generate the latest multi-dimensional features based on the latest queue flow data, and monitor the status of the latest multi-dimensional features; the latest multi-dimensional features include load status, burst status, and data packet distribution status;

[0006] When the latest multi-dimensional feature is detected to have a state transition, a queue adjustment event is triggered, and a corresponding queue adjustment policy is matched according to the latest multi-dimensional feature, and target policy parameters are output;

[0007] adjusting an operating parameter of the switch port based on the target policy parameter;

[0008] The current operating parameters of the switch port are obtained at preset time intervals, and whether there is a parameter deviation is monitored.

[0009] On the other hand, the present application also proposes a switch port queue management device, comprising:

[0010] at least one processor; and,

[0011] a memory communicatively connected to the at least one processor; wherein,

[0012] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform a method for managing a switch port queue as described in the above example.

[0013] On the other hand, the present application also proposes a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured as: a method for managing a switch port queue as described in the above example.

[0014] The present application proposes a switch port queue management method that can bring the following beneficial effects:

[0015] By capturing traffic characteristics in real time and dynamically adjusting queue management policies, this approach effectively overcomes the limitations of traditional methods. By monitoring multi-dimensional characteristics in real time and dynamically triggering policy adjustments, the switch automatically matches the most appropriate management policy based on the current traffic load, burst, and distribution characteristics. This ensures that queue management can respond promptly to changes in traffic characteristics, safeguarding the quality of service for critical services while avoiding resource waste and improving overall network resource utilization efficiency.

[0016] At the same time, the parameter offset monitoring and smooth adjustment mechanism further optimizes network stability and management efficiency. Traditionally, manual parameter configuration is not only complex but also difficult to adapt to real-time traffic changes, which can easily lead to policy failure or network jitter. Automatically monitoring parameter offsets and gradually adjusting them during a buffer period reduces the burden of manual intervention, ensuring that policies remain consistent with current traffic characteristics during implementation, avoiding network instability caused by sudden policy changes. This allows for the maintenance of efficient and stable queue management in a dynamically changing network environment, improving overall network performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 This is a flow chart of a method for managing a switch port queue in an embodiment of the present application;

[0019] Figure 2 This is a schematic diagram of the queue management framework structure in an embodiment of the present application;

[0020] Figure 3 This is a schematic diagram of the data packet size distribution analysis process in an embodiment of the present application;

[0021] Figure 4 This is a schematic diagram of a switch port queue management device in an embodiment of the present application. DETAILED DESCRIPTION

[0022] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0023] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0024] like Figure 1 As shown, an embodiment of the present application provides a method for managing a switch port queue, including:

[0025] S101: Collect queue flow data of the switch port in real time, generate the latest multidimensional features based on the latest queue flow data, and monitor the status of the latest multidimensional features; the multidimensional feature signals include load status, burst status, and data packet distribution status.

[0026] Specifically, the switch's built-in traffic monitoring chip or dedicated acquisition module enables real-time capture of port queue data. Queue traffic data for the port queue is continuously acquired, including instantaneous traffic rate and packet size. The acquisition frequency can be adjusted dynamically based on the network to balance real-time performance with resource consumption. The latest queue traffic data is acquired, and the latest multidimensional features of the latest queue traffic data are extracted. The status of the latest multidimensional features is monitored. These latest multidimensional features include load status, burst status, and packet distribution status.

[0027] It should be noted that the monitoring process is achieved by continuously comparing the multi-dimensional features generated in real time with the preset state intervals. When the feature value enters from one interval into another, it is determined that the feature state has changed.

[0028] In an embodiment of the present application, the specific process for extracting the load status quantity is: extracting the instantaneous traffic rate in the latest queue traffic data, smoothing the instantaneous traffic rate, outputting the average rate value, comparing the ratio between the average rate value and the maximum rate of the port and the preset load threshold, and outputting the load status quantity.

[0029] Specifically, the switch's port data collection function captures the queue's instantaneous traffic rate in real time. A smoothing mechanism is then employed, typically selecting an appropriate time window. Multiple instantaneous rate values ​​within the window are averaged to obtain an average rate reflecting the current load level. The ratio of this average rate to the port's maximum rate is then calculated to measure the current link load. Finally, this ratio is compared with a preset load threshold. Based on the comparison result, the corresponding load status is determined and output, indicating whether the current load is low or high.

[0030] The specific process of extracting the burst state quantity is as follows: calculate the flow rate fluctuation degree of the latest queue flow data based on the time window, compare the flow rate fluctuation degree with the preset burst threshold, and output the burst state quantity.

[0031] Specifically, an appropriate time window is determined to cover a certain amount of instantaneous traffic rate data. Within this window, the deviation of each instantaneous rate from the calculated average rate is calculated. This deviation is used to calculate an indicator reflecting the degree of data dispersion. Based on the ratio of the dispersion indicator to the average rate, a parameter is derived to measure the degree of traffic fluctuation. This parameter is compared with a preset burst threshold, and based on the comparison result, a burst status value is output to determine whether the traffic burstiness is significant.

[0032] The specific process of extracting the data packet distribution state quantity is as follows: extract the data packet size value from the latest queue traffic data, dynamically cluster and segment the data packet size value, generate the large and small packet classification results, calculate the distribution dispersion based on the centroid distance of the classification results, and output the data packet distribution state quantity.

[0033] Specifically, the system collects the size of each packet flowing through a port in real time. Using a dynamic clustering algorithm, it categorizes the packets into large and small groups. It then iterates and continuously updates the centralization trends of these two groups. Based on the updated centroids, it determines the split points and assigns newly arriving packets to the corresponding categories based on size. The distance between the two centroids is then used to calculate a parameter reflecting the distribution dispersion, which measures the balance of packet size distribution. Based on this dispersion parameter, it outputs a packet distribution status metric to determine whether the distinction between large and small flows in the current traffic is clear.

[0034] It should be noted that if Figure 2The following is a schematic diagram of the queue management framework. It is divided into three modules based on their functional implementation. The first is the traffic feature analysis module, which processes and analyzes information such as port queue traffic rate and packet size to obtain traffic characteristics such as the current average traffic rate, traffic burstiness, and packet size distribution. The second is the policy matching module. Based on the traffic characteristics provided by the traffic feature analysis module, it pre-sets multiple queue management policies or sets multiple sets of parameters within the same queue management algorithm to adapt to network traffic with different characteristics. The third is the policy dynamic switching module. When traffic characteristics change, the policy matching module performs a match. If the queue management policy changes, the policy dynamic switching module switches the queue management policy.

[0035] The traffic feature analysis module is used to obtain the switch port queue traffic rate and packet size, and calculate the three port traffic features of average traffic rate, traffic burstiness, and packet size distribution according to the corresponding formula.

[0036] For the average flow rate, the moving average method is used to calculate the average flow rate within a certain time window to smooth the data and reveal the trend. Suppose the flow rate time series data is x1, x2, ..., x n , the time window size is k and the specific average rate calculation formula is: is the average flow rate at the current moment; x t The instantaneous traffic rate of the current port queue is obtained by the switch through sampling measurement.

[0037] The coefficient of variation (CV) is used to measure the degree of traffic burstiness. The formula for calculating the coefficient of variation (CV) is: CV is the coefficient of variation. The larger the value, the greater the fluctuation of the traffic rate and the stronger the burstiness. s is the standard deviation of the traffic rate with a time window size of k. is the average flow rate at the current moment; x t The instantaneous traffic rate of the current port queue.

[0038] According to the size distribution of data packets, the size of the data packets in the current port queue is analyzed by K-means clustering method. First, the data packets are divided into two categories: B and C S , namely large packages and small packages, the centroids of these two classes are μ B and μ S , where the centroid update formula is: Among them, μ t+1 is the centroid value at time t+1, or the centroid value at the next moment; |C t| is the number of elements in set C at time t, and x is an element in set C. When a new data packet arrives, it is necessary to classify the data packet according to the clustering cutoff point. The calculation formula of the clustering cutoff point is: Among them, b is the current cluster split point. When the size of the newly arrived data packet x is less than or equal to b, the data packet x is classified into the small packet class C. S On the contrary, when x is greater than b, it is classified into large package class C B The packet size distribution coefficient D is used to represent the distribution of packet sizes in the port queue, where: When the D value is larger, the distance between the centroids of large packets and small packets is larger, which means that the size distribution of packets in the current port queue is more extreme, and the difference between large and small flows is obvious; when the D value is smaller, the centroids of large packets and small packets are closer, which means that the size distribution of packets in the current port queue is more balanced, and the difference between large and small flows is not obvious. The following figure shows the calculation process of the packet size distribution feature. Figure 3 The figure shows the flow chart of data packet size distribution analysis.

[0039] In an embodiment of the present application, before monitoring the state of the latest multidimensional feature, a process of constructing a state boundary space is also included. The state boundary space consists of a load hysteresis interval, a burst fluctuation boundary, and a balanced distribution interval. Specifically, a historical multidimensional feature sequence is cached within a dynamically scalable time window, and a distribution density analysis is performed on the historical load features within the window. The load feature values ​​are divided into low-density and high-density areas, with the starting point of the low-density area as the lower boundary and the end point of the high-density area as the upper boundary to form a load hysteresis interval; the main fluctuation mode of the burst feature within the window is extracted, and the burst fluctuation boundary is generated according to the intensity value of the main fluctuation mode; the central trend value and discrete range value of the packet distribution feature within the window are calculated, and the discrete range value is expanded with the central trend value to form a packet distribution balanced domain.

[0040] Furthermore, the state of the latest multidimensional feature is monitored. The specific process is as follows: based on the signal state boundary space, the state migration of the latest multidimensional feature is monitored. When the load state quantity exceeds the load state hysteresis interval for several consecutive times, it is determined that there is a load state migration; when the number of times the burst state quantity exceeds the burst fluctuation boundary within the time window is higher than the preset burst threshold, it is determined that there is a burst state migration; when the duration of the data packet distribution state quantity exceeds the packet balanced distribution interval, it is determined that there is a packet distribution state migration.

[0041] S102: When it is detected that the latest multi-dimensional feature has a state transition, a queue adjustment event is triggered, and a corresponding queue adjustment policy is matched according to the latest multi-dimensional feature, and target policy parameters are output.

[0042] Specifically, the interval of the multidimensional features from the previous cycle is compared with the interval of the latest features. If the interval of the latest features is inconsistent with the historical interval, a state transition is determined and an adjustment event is triggered. The combined information of the current multidimensional features is extracted, and matching entries are searched in the policy library to determine the corresponding target parameters and output them.

[0043] In an embodiment of the present application, the specific process of outputting target policy parameters based on the latest multidimensional features is: constructing a multi-level index based on the latest multidimensional features, hierarchically searching in the preset policy rule library based on the multi-level index, matching the corresponding queue adjustment strategy, and generating corresponding target parameter adjustment instructions based on the target policy parameters in the queue adjustment strategy; the target policy parameters include at least a packet loss probability parameter, a queue minimum threshold parameter, and a queue maximum threshold parameter.

[0044] Furthermore, when the load state quantity is high load or low load, or the data packet distribution state quantity is extreme distribution or balanced distribution, an instruction to adjust the threshold interval is output based on the target parameter values ​​corresponding to the queue minimum threshold parameter and the queue maximum threshold parameter; when the burst state quantity is high burst or low burst, an instruction to adjust the packet loss probability is output based on the target parameter value corresponding to the packet loss probability parameter.

[0045] It should be noted that the policy matching module presets multiple queue management strategies in advance according to the traffic characteristics provided by the traffic characteristic analysis module, or sets multiple sets of parameters in the same queue management algorithm to adapt to network traffic with different characteristics, and sets multiple sets of parameters based on the RED algorithm according to different traffic characteristics. max , minimum threshold min th , maximum threshold max th Set multiple groups of parameters.

[0046] It should be noted that for the average flow rate Quantitative analysis is performed on traffic burstiness (coefficient of variation) (CV) and packet size distribution coefficient (D). Regarding average traffic rate, in network traffic management, the ratio of the average traffic rate to the maximum port rate is a core indicator for measuring link load. Based on the principle of bandwidth reservation, engineering practices typically reserve additional bandwidth to avoid congestion caused by transient bursts. Furthermore, based on the physical link characteristics, the link load threshold can be appropriately increased for high-speed links due to their large buffer space. On low-speed links, the link load threshold should be lowered to prevent burst traffic from exhausting bandwidth.

[0047] Traffic burstiness (CV): This is performed based on the thresholds in Table 1. When CV = 1, the standard deviation of the current traffic rate is equal to the mean, indicating that the traffic fluctuation amplitude is comparable to the average level. At this point, the burstiness is significantly enhanced. Therefore, CV = 1 is used as the threshold to distinguish between obvious and subtle traffic bursts.

[0048]

[0049] Table 1

[0050] Regarding the packet size distribution coefficient D, packets are classified using the K-means clustering method, which divides them into two categories. The packet size distribution coefficient is the ratio of the centroids of these two categories. For example, if packet sizes are evenly distributed between 0 and 1, then the ranges of the two categories are (0, 0.5) and (0.5, 1), respectively. The corresponding centroid values ​​are 0.25 and 0.75, respectively. This indicates that the packet size distribution coefficient D is 3. This threshold can be used. When D is greater than 3, there is a clear distinction between packet sizes, indicating that the problem of large and small flows is more prominent. When D is less than or equal to 3, the packet size distribution is relatively good, indicating that the situation of large and small flows is within an acceptable range.

[0051] S103: Adjust the operating parameters of the switch port based on the target policy parameters.

[0052] Specifically, when adjusting switch port operating parameters based on the output target policy parameters, a smooth transition mechanism is used to avoid network jitter caused by policy switching. Specifically, a policy switching buffer period is set, during which parameter values ​​are gradually adjusted rather than directly replaced.

[0053] For example, if the current maximum probability of packet loss is P max 1. The target parameter is P max 2. The buffer period is set to 10 units of time, then the parameter change in each unit of time is (P max 2-P max 1) / 10, a smooth transition from the current parameters to the target parameters is achieved through staged fine-tuning. th and the maximum threshold max th , also following this logic, ensures the continuity of network traffic transmission and the stability of service quality during the adjustment process.

[0054] It should be noted that when the average traffic rate is larger, the congestion risk is higher, and the minimum threshold min in the RED algorithm needs to be lowered. th and the maximum threshold max th To perform early packet loss; when traffic bursts, the more severe the traffic fluctuation, the more likely it is that the maximum probability of packet loss P will be increased. maxTo enhance packet loss sensitivity; when the packet size distribution is poor, the queue length changes will be more drastic, and the minimum threshold min needs to be increased th , maximum threshold max th For example, if the preset load threshold is set to 70%, the preset burst threshold is set to 1, and the preset packet distribution coefficient threshold is set to 3, there are the following corresponding relationships: when the average traffic rate is When the port has the maximum rate, CV <= 1, and D <= 3, P max Take P max 2. min th Take min th 1. max th Take the max th 1; When the average flow rate When the port has the maximum rate, CV <= 1, and D > 3, P max Take P max 2. min th Take min th 2. max th Take the max th 1; When the average flow rate When the port has the maximum rate, CV>1, and D<=3, P max Take P max 1.min th Take min th 1. max th Take the max th 1; When the average flow rate When the port has the maximum rate, CV>1, and D>3, P max Take P max 1.min th Take min th 2. max th Take the max th 1; When the average flow rate When the port has the maximum rate, CV <= 1, and D <= 3, P max Take P max 2. min th Take min th 2. max th Take the max th 2; When the average flow rate When the port has the maximum rate, CV <= 1, and D > 3, P max Take P max 2. min th Take min th 2. max th Take the max th 1; When the average flow rate When the port has the maximum rate, CV>1, and D<=3, P max Take P max 1.min th Take min th 2. max th Take the max th 2; When the average flow rate When the port has the maximum rate, CV>1, and D>3, max Take P max 1.min th Take min th 2. max th Take the max th 1.

[0055] S104: Acquire the current operating parameters of the switch port at a preset time interval, and monitor whether there is a parameter deviation.

[0056] Specifically, the system periodically samples current operating parameters, compares the actual parameters with the target parameters, and calculates the deviation between the two. If the deviation is within an acceptable range, the current policy is maintained. If the deviation exceeds the preset range, indicating that the current parameters are no longer suitable for the network status, the system will re-trigger the traffic feature collection and analysis process and initiate a new round of parameter adjustments to ensure that the queue management policy always matches the actual network conditions.

[0057] In this embodiment, the actual operating parameters of the switch port are compared one by one with the target policy parameters matched by the queue management policy to identify the differences between the two. By calculating the difference between the current operating parameters and the target parameters, a quantitative comparison deviation value is obtained, which is used to measure the degree of deviation between the actual operating status and the expected target.

[0058] The calculated deviation is then compared to a preset tolerance threshold. This threshold is an acceptable deviation range set based on network stability requirements. If the deviation does not exceed the threshold, the current parameters are within a reasonable range and no adjustment is required. If the deviation exceeds the threshold, parameter drift is detected and the adjustment process must be initiated.

[0059] When parameter deviation is confirmed, a preset buffer period is first set to achieve a smooth parameter transition and avoid network jitter caused by sudden parameter changes. The buffer period is evenly divided into multiple equal time units. The parameter adjustment range within each time unit is determined based on the total deviation value and the number of time units. At the end of each time unit, the current execution parameters are fine-tuned according to the calculated adjustment range, gradually bringing the current operating parameters closer to the target policy parameters. At the end of the buffer period, the current operating parameters are consistent with the target policy parameters, completing the transition.

[0060] By capturing traffic characteristics in real time and dynamically adjusting queue management policies, the limitations of traditional methods are effectively overcome. Traditional queue management methods often use a single policy or static parameter combination, which is difficult to adapt to dynamically changing traffic characteristics. This often leads to problems such as parameter inapplicability and imbalance between service quality and resource utilization. However, real-time monitoring of multi-dimensional characteristics and dynamic triggering of policy adjustments enable switches to automatically match the most appropriate management policy based on the current traffic load, burst, and distribution characteristics. This ensures that queue management can respond promptly to changes in traffic characteristics, thereby protecting the service quality of important services, avoiding resource waste, and improving the overall utilization efficiency of network resources.

[0061] At the same time, the parameter offset monitoring and smooth adjustment mechanism further optimizes network stability and management efficiency. Traditionally, manual parameter configuration is not only complex but also difficult to adapt to real-time traffic changes, which can easily lead to policy failure or network jitter. Automatically monitoring parameter offsets and gradually adjusting them during a buffer period reduces the burden of manual intervention, ensuring that policies remain consistent with current traffic characteristics during implementation, avoiding network instability caused by sudden policy changes. This allows for the maintenance of efficient and stable queue management in a dynamically changing network environment, improving overall network performance.

[0062] like Figure 4 As shown, the embodiment of the present application also proposes a switch port queue management device, including:

[0063] at least one processor; and,

[0064] a memory communicatively connected to the at least one processor; wherein,

[0065] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to execute a method for managing a switch port queue as described in any one of the above embodiments.

[0066] An embodiment of the present application further provides a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to be: a method for managing a switch port queue as described in any of the above embodiments.

[0067] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the device and medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0068] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0069] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0071] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0073] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0074] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0075] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0076] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0077] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for managing a switch port queue, characterized in that: include: Collect queue flow data of the switch port in real time, generate the latest multi-dimensional features based on the latest queue flow data, and monitor the status of the latest multi-dimensional features; The latest multi-dimensional features include load state quantity, burst state quantity and data packet distribution state quantity; When the latest multi-dimensional feature is detected to have a state transition, a queue adjustment event is triggered, and a corresponding queue adjustment policy is matched according to the latest multi-dimensional feature, and target policy parameters are output; adjusting an operating parameter of the switch port based on the target policy parameter; The current operating parameters of the switch port are obtained at preset time intervals, and whether there is a parameter deviation is monitored.

2. A switch port queue management method according to claim 1, characterized in that: Generating the latest multi-dimensional features based on the latest queue traffic data specifically includes: Extracting the instantaneous flow rate from the latest queue flow data, smoothing the instantaneous flow rate, and outputting an average rate value; Compare the ratio of the average rate value to the maximum rate of the port and a preset load threshold, and output a load status value; Calculate the traffic rate fluctuation degree of the latest queue traffic data based on the time window, compare the traffic rate fluctuation degree with a preset burst threshold, and output a burst state quantity; Extracting data packet size values ​​from the latest queue traffic data, dynamically clustering and segmenting the data packet size values, and generating large and small packet classification results; The distribution dispersion is calculated based on the centroid distance of the classification result, the distribution dispersion is compared with a preset packet distribution coefficient threshold, and the data packet distribution state quantity is output.

3. A switch port queue management method according to claim 1, characterized in that: Before monitoring the status of the latest multi-dimensional feature, the method further includes: Obtain multi-dimensional feature sequences of historical traffic; Based on the multi-dimensional characteristic sequence of historical traffic, a state boundary space is constructed; the state boundary space consists of a load hysteresis interval, a sudden fluctuation boundary, and a balanced distribution interval.

4. A switch port queue management method according to claim 3, characterized in that: The monitoring of the state of the latest multi-dimensional feature specifically includes: Based on the signal state boundary space, monitoring the state transition of the latest multidimensional feature; When the load state quantity exceeds the load state hysteresis interval several times in a row, it is determined that a load state transition exists; When the number of times that the burst state quantity exceeds the burst fluctuation boundary within the time window is higher than a preset burst threshold, it is determined that a burst state transition exists; When the duration of the data packet distribution state exceeds the packet balanced distribution interval, it is determined that there is a packet distribution state migration.

5. A switch port queue management method according to claim 1, characterized in that: According to the latest multi-dimensional features, matching the corresponding queue adjustment strategy and outputting the target strategy parameter group specifically include: Building a multi-level index based on the latest multidimensional features; Based on the multi-level index, a hierarchical search is performed in a preset policy rule library to match the corresponding queue adjustment policy; Based on the target policy parameters in the queue adjustment policy, corresponding target parameter adjustment instructions are generated respectively; the target policy parameters include at least a packet loss probability parameter, a queue minimum threshold parameter, and a queue maximum threshold parameter.

6. A switch port queue management method according to claim 5, characterized in that: Generating corresponding target parameter adjustment instructions based on the target policy parameters in the queue adjustment policy specifically includes: When the load state quantity is high load or low load, or the data packet distribution state quantity is extreme distribution or balanced distribution, outputting a threshold interval adjustment instruction based on target parameter values ​​corresponding to the queue minimum threshold parameter and the queue maximum threshold parameter; When the burst state quantity is high burst or low burst, an instruction for adjusting the packet loss probability is output based on a target parameter value corresponding to the packet loss probability parameter.

7. A switch port queue management method according to claim 1, characterized in that: The monitoring of whether there is parameter deviation specifically includes: Comparing the current operating parameters of the switch port with the target policy parameters corresponding to the queue management policy, and calculating a comparison deviation value; Determine whether the comparison deviation value exceeds a preset tolerance threshold.

8. A switch port queue management method according to claim 7, characterized in that: After determining whether the comparison deviation value exceeds a preset tolerance threshold, the method further includes: When there is a parameter offset, the preset buffer period is divided into a plurality of time units, and at the end of each time unit, the current execution parameters are adjusted to gradually transition the current execution parameters to the target strategy parameters.

9. A switch port queue management device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to execute the switch port queue management method according to any one of claims 1 to 8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions are configured to execute a switch port queue management method according to any one of claims 1 to 8.

Images