Data scheduling method and system for machine room switch

Abstract

The application relates to the technical field of switch data transmission, in particular to a data scheduling method and system for a computer room switch, which specifically comprises the following steps: adding a dynamic priority field in a data frame structure, determining the weighted throughput of a switch interface based on the length, priority and dynamic priority of a data frame sent by the switch interface; determining the collision severity of the switch interface in a sliding window based on the average contention period number of a data link, combining the weighted throughput of the interface to determine the congestion characteristic value of each data link in the sliding window; optimizing and adjusting the data scheduling method of the switch based on the dynamic priority of each data frame and the congestion characteristic value of each data link; the overall data scheduling efficiency of the park network computer room switch is improved, the real-time performance of different production data exchange is enhanced, and the adaptability of the computer room switch data scheduling to different data priorities is improved.

Description

Technical Field

[0001] This application relates to the field of data transmission technology for switches, specifically to a data scheduling method and system for data center switches. Background Technology

[0002] Modern industrial production typically involves parallel production across multiple modules. Within a factory, different production modules generate massive amounts of production data, including equipment operating status data, production process data, and material management data. To address the complexity of industrial production, production modules utilize the factory's campus network to achieve data scheduling between different modules, ensuring coordinated and stable operation. However, the campus network has limited switching capacity, and switches within the computer room are prone to collisions and congestion during large-scale data exchange, hindering efficient data scheduling.

[0003] However, the data exchange priorities differ across different industrial production modules. In actual industrial production scenarios, some data, such as equipment fault alarm data and critical parameter data, require higher priority transmission, while other data, such as the operating status data of non-critical equipment, can be transmitted with a slight delay. Traditional data scheduling methods do not differentiate between data categories and are insufficiently adaptable to the data priorities of factory / industrial park networks. When network congestion occurs, it may lead to the loss of data frames that require priority transmission, causing instability in the production process and even triggering safety accidents. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a data scheduling method and system for data center switches, and the specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of this application provide a data scheduling method for data center switches, the method comprising the following steps:

[0006] Add a dynamic priority field to the data frame structure; set the dynamic priority of each data frame according to the waiting time for each data frame to be sent.

[0007] A preset sliding window is used to construct the weighted throughput of each interface of the switch within the sliding window based on the length, priority, and dynamic priority of each data frame sent by each interface of the switch.

[0008] Based on the data frame collision situation of each data link within the sliding window, the average contention period of each data link is calculated; based on the overall distribution characteristics of the average contention period of all data links of the switch interface, the collision severity of the switch interface within the sliding window is determined.

[0009] For all interfaces traversed by each data link, the congestion characteristic value of each data link within the sliding window is determined based on the weighted throughput of the interface and the collision severity.

[0010] Based on the dynamic priority of each data frame and the congestion characteristics of each data link, the data dispatching method of the switch is optimized and adjusted.

[0011] In one embodiment, the process of obtaining the dynamic priority is as follows:

[0012] The initial dynamic priority of each data frame is set to the lowest dynamic priority. When the waiting time for each data frame to be sent increases by one contention period, the corresponding dynamic priority value increases by 1. When it increases to the preset highest dynamic priority, it remains unchanged.

[0013] In one embodiment, the process of obtaining the weighted throughput of the switch interface within the sliding window is as follows:

[0014] Calculate the sum of the priority and dynamic priority of each data frame, and record it as the first sum; calculate the sum of the highest priority and the highest dynamic priority of the data frame, and record it as the second sum.

[0015] The weighted throughput of each interface of the switch within the sliding window is determined based on the length of each data frame sent by each interface of the switch within the sliding window, the first sum value, and the second sum value.

[0016] In one embodiment, the expression for the weighted throughput is:

[0017]

[0018] In the formula, This represents the weighted throughput of the current switch interface within the sliding window; Indicates the duration of the sliding window; This indicates the number of data frames sent by the current switch interface within the sliding window; This indicates the number of times the current switch interface sends data within the sliding window. The length of each data frame; and These represent the number of times the current switch interface sends data within the sliding window. The priority of each data frame and its dynamic priority; and These represent the highest priority and the highest dynamic priority of the data frame, respectively. For a preset, extremely small positive number; where, The first sum, It is the second sum.

[0019] In one embodiment, the process of obtaining the average contention period is as follows:

[0020] Within the sliding window, the ratio of the number of contention periods during which data frames are sent on each data link of the switch interface are calculated to the number of data frames sent, and this ratio is recorded as the average number of contention periods for each data link.

[0021] In one embodiment, the expression for the collision severity is:

[0022]

[0023] In the formula, This represents the collision severity of the current switch interface within the sliding window. This indicates the number of data links on the current switch interface; This indicates the number of times the current switch interface passes through the sliding window. The average number of contention periods per data link; This represents the maximum average number of contention periods for all data links passing through the current switch interface within the sliding window. It is a preset minimum positive number.

[0024] In one embodiment, the process of obtaining the congestion characteristic value of each data link within the sliding window is as follows:

[0025] Within the sliding window, obtain each interface through which each data link of the switch passes, and calculate the ratio of the weighted throughput of each interface through which each data link passes to the maximum weighted throughput of all interfaces through which it passes, denoted as the first ratio.

[0026] Based on the collision severity of each interface traversed by each data link and the first ratio, the congestion characteristic value of each data link within the sliding window is determined.

[0027] In one embodiment, the congestion characteristic is the sum of the products of the collision severity of all interfaces traversed by the data link within the sliding window and the first ratio.

[0028] In one embodiment, the process of optimizing and adjusting the data mobilization method of the switch is as follows:

[0029] After receiving a data frame, the switch compares the transmission times of the data frames and prioritizes processing the data frames that were sent earlier.

[0030] If the sending times are the same, the priority fields of the data frames are compared, and the higher priority data frames are transmitted first.

[0031] When priorities are the same, compare the dynamic priority field, and the data frame with the higher dynamic priority is transmitted first.

[0032] Based on the remaining capacity of the buffer and the size of the data frame, determine whether the data frame can enter the buffer; if the remaining capacity of the buffer is insufficient to store the data frame, packet loss processing is required according to the packet loss policy.

[0033] Based on the value of the dynamic priority field and the existing link congestion status, data frames are allocated to links: the highest dynamic priority data frames are allocated to the link with the smallest congestion characteristic value in the previous sliding window for transmission.

[0034] Data frames in the buffer are sorted in descending order of dynamic priority to ensure that high dynamic priority data frames are transmitted first.

[0035] Secondly, embodiments of this application also provide a data scheduling system for data center switches, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0036] The embodiments of this application have at least the following beneficial effects:

[0037] This application adds a dynamic priority field to the existing switch data frame structure to characterize the dynamic changes in priority caused by data frame queuing, reduce frame waiting latency, and ensure real-time data exchange between different production modules. Based on the throughput of the switch interface and the average contention period within the sliding window, the congestion characteristic value of each data link is determined, and further, the data transmission links of the switch are dynamically adjusted according to the priority of data frames in the link. This reduces the degree of link congestion during subsequent data scheduling, reduces frame loss due to collisions, and improves data scheduling efficiency.

[0038] By comparing the results of different data frame priorities and dynamic priorities, data scheduling is dynamically adjusted to achieve dynamic data scheduling for the computer room switches in the factory campus network. This reduces frame loss caused by exceeding the buffer area and avoids data frame retransmission. While improving the overall data scheduling efficiency of the computer room switches in the campus network, it also enhances the real-time performance of different production data exchanges, thereby improving the adaptability of the computer room switches' data scheduling to different data priorities. Attached Figure Description

[0039] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A flowchart illustrating the steps of a data scheduling method for a data center switch provided in one embodiment of this application;

[0041] Figure 2 This is a schematic diagram illustrating the process of obtaining weighted throughput. Detailed Implementation

[0042] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the data scheduling method and system for data center switches proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0044] The following, in conjunction with the accompanying drawings, details the specific scheme of the data scheduling method and system for data center switches provided in this application.

[0045] Please see Figure 1 The diagram illustrates a flowchart of a data scheduling method for a data center switch according to an embodiment of this application. The method includes the following steps:

[0046] Step S1: Add a dynamic priority field to the data frame structure; set the dynamic priority of each data frame according to the waiting time for each data frame to be sent.

[0047] In industrial production scenarios, different data have varying impacts on the stability and security of the production process. Traditional static priority settings cannot meet the dynamically changing network environment and high real-time data transmission requirements of factory / industrial park networks. This application introduces a dynamic priority field, which can more flexibly reflect the priority changes of data during transmission, enabling it to better adapt to the complex and ever-changing data transmission needs in industrial production.

[0048] Add a dynamic priority field to the data frame structure, set the range of the dynamic priority field to 0~7, use a field with a length of 1 byte to represent it, and set the initial value of the dynamic priority of all data frames to 0.

[0049] The revised Ethernet data frame structure in this application is as follows:

[0050] [Destination address field, source address field, type / length field, priority field, timestamp field, dynamic priority field, data field, frame check sequence field].

[0051] The MAC address field is 6 bytes long and identifies the receiver's MAC address (LAN address); the source address field is 6 bytes long and identifies the sender's MAC address; the type / length field is 2 bytes long and indicates the type or length of subsequent data; the priority field is 3 bits long and indicates the data frame's transmission priority level, ranging from 0 to 7. The IEEE 802.1Q standard directly assigns priorities to different data frames; the dynamic priority field is 3 bits long and is used to identify the dynamic priority of the data frame, with its value range denoted as... Preferably, in the embodiments of this application, Set the value to 0, and The value is set to 7, with a range of 0 to 7; the data field has a variable length, with a default range of 46-1500 bytes, which may vary depending on the specific protocol and requirements during actual use; the frame check sequence field has a length of 4 bytes and is used to detect whether errors occur during frame transmission.

[0052] The timestamp field is 4 bytes long and is used to indicate when the data frame was sent.

[0053] It should be noted that this application only provides one setting method for the value range of dynamic priority. Implementers can set the value range of dynamic priority according to the actual situation. This application does not impose any specific restrictions.

[0054] Among them, dynamic priority is adjusted in real time according to the waiting time of data frames, thereby ensuring that data with a long waiting time can be transmitted in a timely manner, reducing the risk of data frame loss, and improving the coordination and stability of the entire industrial production process.

[0055] Specifically, the initial dynamic priority of each data frame is set to 0, indicating that it has not experienced a waiting period for transmission. For each additional contention period length during the data frame's waiting period, the dynamic priority value increases by 1, remaining unchanged when it reaches 7. In this application, the contention period length for 10 Mbit / s Ethernet is 51.2. .

[0056] Step S2: Preset a sliding window. Based on the length, priority, and dynamic priority of each data frame sent by each interface of the switch within the sliding window, construct the weighted throughput of each interface of the switch within the sliding window.

[0057] In the computer room switches of the factory campus network, the usage and importance of different links are different. By measuring the throughput of the switch interface, the average number of contention periods within the data frame transmission time, as well as the priority and dynamic priority of the data frames, the degree of collision and congestion of different links can be fully reflected.

[0058] First, a sliding window is set to capture dynamic changes in the link in a timely manner, making congestion assessment more accurate and real-time. This application sets the sliding window duration to 100ms. As another embodiment of this application, implementers can set the sliding window duration according to their actual needs.

[0059] Traditional congestion assessment methods often focus only on the single metric of link throughput, failing to comprehensively consider factors such as data frame priority and collision occurrences. This application comprehensively assesses link congestion through multiple metrics, better adapting to complex data transmission scenarios in industrial production. This provides a more accurate basis for subsequent data scheduling, effectively reducing frame loss due to collisions and improving data scheduling efficiency.

[0060] Furthermore, for each interface of each switch in the computer room, the weighted throughput of that switch interface within the sliding window is calculated based on the priority, dynamic priority, and data frame length of the data frames. The expression is as follows:

[0061]

[0062] In the formula, This represents the weighted throughput of the current switch interface within the sliding window; Indicates the duration of the sliding window; This indicates the number of data frames sent by the current switch interface within the sliding window; This indicates the number of times the current switch interface sends data within the sliding window. The length of each data frame; and These represent the number of times the current switch interface sends data within the sliding window. The priority of each data frame and its dynamic priority; and These represent the highest priority and the highest dynamic priority of the data frame, respectively, both of which are set to 7 in this application; It is a very small positive number, which serves to avoid the numerator being 0. Preferably, in the embodiments of this application, it is... The value is set to 1. As another embodiment of this application, the implementer may set it according to the actual situation. The value of . Where, The first sum, It is the second sum.

[0063] By calculating weighted throughput by setting weights as a combination of data frame priority and dynamic priority, the importance of data and network resource usage can be more accurately reflected, thus providing strong support for optimizing network resource allocation, improving the flexibility and adaptability of scheduling algorithms, and avoiding network congestion.

[0064] Step S3: Calculate the average contention period of each data link based on the data frame collision situation of each data link within the sliding window; determine the collision severity of the switch interface within the sliding window based on the overall distribution characteristics of the average contention period of all data links of the switch interface.

[0065] First, within the sliding window, the ratio of the number of contention periods during which data frames are transmitted on each data link of the computer room switch interface to the total number of data frames transmitted is recorded as the average contention period for that data link. The average contention period directly reflects the collision situation of the link; a higher contention period indicates more frequent data frame collisions on that link, leading to increased transmission delays and a higher risk of data frame loss. By incorporating the average contention period into the assessment of collision severity, the severity of link collisions can be intuitively reflected.

[0066] Furthermore, each interface of a switch may have multiple links. For each data link of a switch interface, based on the data frame transmission status of the data link passing through that interface within the sliding window, the collision severity of the switch interface within the sliding window is calculated, expressed as:

[0067]

[0068] In the formula, This represents the collision severity of the current switch interface within the sliding window. This indicates the number of data links on the current switch interface; This indicates the number of times the current switch interface passes through the sliding window. The average number of contention periods per data link; This represents the maximum average number of contention periods for all data links passing through the current switch interface within the sliding window. The value is a preset, extremely small positive number to prevent the denominator from being zero. Preferably, in the embodiments of this application, the value is... The value is set to 0.001. As another embodiment of this application, the implementer can set it according to the actual situation. The value of .

[0069] Since the probability of a link collision is relatively low, usually only a few links experience severe collisions with a large average contention period. Therefore, to avoid the impact of a single link's severe collision on the use of other interfaces on the entire link, the maximum value of the average contention period of all links passing through the interface is used to normalize the collision severity, thereby improving the utilization rate of switch communication resources in subsequent dynamic data scheduling.

[0070] This reflects the collision situation and transmission efficiency of all links on the interface. A high collision severity indicates that there are many collisions on the links of that interface, which has a significant impact on data transmission. Appropriate congestion control measures need to be implemented to improve the overall transmission efficiency and reliability of the interface.

[0071] Step S4: For all interfaces traversed by each data link, determine the congestion characteristic value of each data link within the sliding window based on the weighted throughput and the collision severity of the interface.

[0072] Within a sliding window, each data link in the data center switch passes through multiple interfaces. For each data link in the data center switch within the sliding window, its congestion characteristic value is calculated based on the weighted throughput and collision severity of each interface in the data link. The expression is as follows:

[0073]

[0074] In the formula, The congestion characteristic value of the m-th data link of the switch within the sliding window is used as an assessment of the data link congestion status; S represents the number of interfaces traversed by the m-th data link of the switch within the sliding window. This indicates the collision severity of the m-th data link of the switch passing through the s-th interface within the sliding window; This represents the weighted throughput of the s-th interface traversed by the m-th data link of the switch within the sliding window; This represents the maximum weighted throughput of all interfaces traversed by the m-th data link of the switch within the sliding window. This is the first ratio.

[0075] On the one hand, weighted throughput takes into account both the link's throughput and data frame priority. High-priority data frames have a greater impact on the link's transmission efficiency, while the transmission of low-priority data frames has a relatively smaller impact on the overall link performance. By using weighted throughput, the congestion status of the link when transmitting high-priority data can be reflected more accurately.

[0076] On the other hand, throughput directly reflects the link's transmission capacity, while collision severity weight reflects the link's transmission efficiency. Combining the two allows for a comprehensive assessment of the link's congestion status. Collision severity changes over time, and the link's congestion status changes dynamically accordingly. Congestion characteristic values ​​can reflect the link's current state in real time, providing a basis for subsequent dynamic data scheduling.

[0077] Step S5: Based on the dynamic priority of each data frame and the congestion characteristic value of each data link, optimize and adjust the data dispatching method of the switch.

[0078] In industrial production processes, the real-time nature and integrity of data are crucial. By comparing multiple indicators such as transmission time, priority, and dynamic priority, it can be ensured that data is transmitted in a reasonable order, meeting the data transmission requirements of different production modules.

[0079] Meanwhile, the management of the buffer and the formulation of packet loss strategies are crucial for ensuring the stability and reliability of data scheduling. Properly utilizing the remaining buffer capacity and packet loss priority enhancement mechanisms can effectively protect the transmission of high-priority data during network congestion.

[0080] The data in the current sliding window is scheduled based on the congestion characteristic value of the previous sliding window. The scheduling process is as follows:

[0081] After receiving a data frame, the switch first compares the transmission times of the data frames and prioritizes processing the data frames that were sent earlier.

[0082] If the sending times are the same, the priority fields of the data frames are compared, and the higher priority data frames are transmitted first.

[0083] When priorities are the same, the dynamic priority field is further compared, and the data frame with the higher dynamic priority is transmitted first.

[0084] Based on the remaining capacity of the buffer and the size of the data frame, determine whether the data frame can enter the buffer; if the remaining capacity of the buffer is insufficient to store the data frame, packet loss processing is required according to the Automatic Repeat Request (ARQ) packet loss policy.

[0085] It should be noted that this application provides only one packet loss strategy. There are many existing packet loss strategies, and implementers may also use other packet loss strategies to process data frames. This application does not impose any specific restrictions.

[0086] Based on the value of the dynamic priority field and the link congestion status of the previous sliding window, the data frame is allocated to the link with the smallest congestion characteristic value in the previous sliding window for transmission, so as to avoid further deterioration of the congested link.

[0087] Data frames in the buffer are sorted in descending order of dynamic priority to ensure that high dynamic priority data frames are transmitted first.

[0088] By dynamically adjusting link allocation and queuing mechanisms, these changes can be flexibly addressed, reducing frame loss caused by exceeding the buffer area and avoiding data frame retransmission. This improves the overall data scheduling efficiency of the campus network's data center switches, enhances the real-time performance of data exchange between different production modules, and further improves the adaptability of the data center switches' data scheduling to different data priorities, ensuring the coordinated and stable operation of the entire industrial production process.

[0089] A schematic diagram of the process for obtaining weighted throughput is shown below. Figure 2 As shown.

[0090] Based on the same inventive concept as the above methods, this application also provides a data scheduling system for data center switches, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above data scheduling methods for data center switches.

[0091] In summary, the embodiments of this application provide a data scheduling method for data center switches. By adding a dynamic priority field to the existing switch data frame structure, the method characterizes the dynamic changes in priority caused by data frame queuing, reduces frame waiting latency, and ensures real-time data exchange between different production modules. Based on the throughput of the switch interface and the average contention period within the sliding window, the congestion characteristic value of each data link, and further based on the priority of data frames in the link, the method dynamically adjusts the data transmission links of the switch to reduce the degree of link congestion during subsequent data scheduling, reduce frame loss caused by collisions, and improve data scheduling efficiency.

[0092] By comparing the results of different data frame priorities and dynamic priorities, data scheduling is dynamically adjusted to achieve dynamic data scheduling for the computer room switches in the factory campus network. This reduces frame loss caused by exceeding the buffer area and avoids data frame retransmission. While improving the overall data scheduling efficiency of the computer room switches in the campus network, it also enhances the real-time performance of different production data exchanges, thereby improving the adaptability of the computer room switches' data scheduling to different data priorities.

[0093] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this application. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0094] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0095] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A data scheduling method for data center switches, characterized in that, The method includes the following steps: Add a dynamic priority field to the data frame structure; set the dynamic priority of each data frame according to the waiting time for each data frame to be sent. A preset sliding window is used to construct the weighted throughput of each interface of the switch within the sliding window based on the length, priority, and dynamic priority of each data frame sent by each interface of the switch. Based on the data frame collision situation of each data link within the sliding window, the average contention period of each data link is calculated; based on the overall distribution characteristics of the average contention period of all data links of the switch interface, the collision severity of the switch interface within the sliding window is determined. For all interfaces traversed by each data link, the congestion characteristic value of each data link within the sliding window is determined based on the weighted throughput of the interface and the collision severity. Based on the dynamic priority of each data frame and the congestion characteristics of each data link, the data scheduling method of the switch is optimized and adjusted. The process of optimizing and adjusting the data scheduling method for the switch is as follows: After receiving a data frame, the switch compares the transmission times of the data frames and prioritizes processing the data frames that were sent earlier. If the sending times are the same, the priority fields of the data frames are compared, and the higher priority data frames are transmitted first. When priorities are the same, compare the dynamic priority field, and the data frame with the higher dynamic priority is transmitted first. Based on the remaining capacity of the buffer and the size of the data frame, determine whether the data frame can enter the buffer; if the remaining capacity of the buffer is insufficient to store the data frame, packet loss processing is required according to the packet loss policy. Based on the value of the dynamic priority field and the existing link congestion status, data frames are allocated to links: the highest dynamic priority data frames are allocated to the link with the smallest congestion characteristic value in the previous sliding window for transmission. Data frames in the buffer are sorted in descending order of dynamic priority to ensure that high dynamic priority data frames are transmitted first.

2. The data scheduling method for data center switches as described in claim 1, characterized in that, The process of obtaining the dynamic priority is as follows: The initial dynamic priority of each data frame is set to the lowest dynamic priority. When the waiting time for each data frame to be sent increases by one contention period, the corresponding dynamic priority value increases by 1. When it increases to the preset highest dynamic priority, it remains unchanged.

3. The data scheduling method for data center switches as described in claim 2, characterized in that, The process of obtaining the weighted throughput of the switch interface within the sliding window is as follows: Calculate the sum of the priority and dynamic priority of each data frame, and record it as the first sum; calculate the sum of the highest priority and the highest dynamic priority of the data frame, and record it as the second sum. The weighted throughput of each interface of the switch within the sliding window is determined based on the length of each data frame sent by each interface of the switch within the sliding window, the first sum value, and the second sum value.

4. The data scheduling method for data center switches as described in claim 3, characterized in that, The expression for the weighted throughput is: ； In the formula, This represents the weighted throughput of the current switch interface within the sliding window; Indicates the duration of the sliding window; This indicates the number of data frames sent by the current switch interface within the sliding window; This indicates the number of times the current switch interface sends data within the sliding window. The length of each data frame; and These represent the number of times the current switch interface sends data within the sliding window. The priority of each data frame and its dynamic priority; and These represent the highest priority and the highest dynamic priority of the data frame, respectively. For a preset, extremely small positive number; where, The first sum, It is the second sum.

5. The data scheduling method for data center switches as described in claim 1, characterized in that, The process for obtaining the average number of contention periods is as follows: Within the sliding window, the ratio of the number of contention periods during which data frames are sent on each data link of the switch interface are calculated to the number of data frames sent, and this ratio is recorded as the average number of contention periods for each data link.

6. The data scheduling method for data center switches as described in claim 1, characterized in that, The expression for the collision severity is: ； In the formula, This represents the collision severity of the current switch interface within the sliding window. This indicates the number of data links on the current switch interface; This indicates the number of times the current switch interface passes through the sliding window. The average number of contention periods per data link; This represents the maximum average number of contention periods for all data links passing through the current switch interface within the sliding window. It is a preset minimum positive number.

7. The data scheduling method for data center switches as described in claim 1, characterized in that, The process of obtaining the congestion characteristic value of each data link within the sliding window is as follows: Within the sliding window, obtain each interface through which each data link of the switch passes, and calculate the ratio of the weighted throughput of each interface through which each data link passes to the maximum weighted throughput of all interfaces through which it passes, denoted as the first ratio. Based on the collision severity of each interface traversed by each data link and the first ratio, the congestion characteristic value of each data link within the sliding window is determined.

8. The data scheduling method for data center switches as described in claim 7, characterized in that, The congestion characteristic value is the sum of the products of the collision severity of all interfaces traversed by the data link within the sliding window and the first ratio.

9. A data scheduling system for data center switches, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-8.

Images