Data transmission system and method of multi-level Spine-Leaf IB network architecture

Abstract

The invention relates to the technical field of data transmission, and discloses a data transmission system and method of a multi-level Spine-Leaf IB network architecture, the system comprises a Spine layer, a Leaf layer and a transmission control module, the transmission control module comprises an acquisition unit, a judgment unit, an optimization unit and a storage unit; the acquisition unit is configured to determine an initial transmission path number and an initial path flow distribution value of the to-be-transmitted data based on the transmission mode and the transmission demand information; the judgment unit is configured to judge whether to optimize the initial path flow distribution value or not according to the transmission stability index; the optimization unit is configured to determine an optimization coefficient corresponding to the initial path flow distribution value according to the comparison result, and obtain an optimized path flow distribution value; the storage unit is configured to store a network transmission impact index. The method can meet the high requirement of a large data center for data transmission, effectively disperse the flow and reduce the influence of a single point of fault.

Description

Technical Field

[0001] The present invention relates to the technical field of data transmission, and in particular, to a data transmission system and method for a multi-level Spine-Leaf IB network architecture. Background Art

[0002] With the rapid development of data centers and high-performance computing applications, traditional network architectures (such as three-tier architectures) have gradually exposed problems such as bandwidth bottlenecks, high latency, and insufficient scalability. Especially in scenarios of large-scale distributed computing, artificial intelligence training, and big data analysis, the network needs to support parallel transmission of massive amounts of data, while requiring extremely low latency and high reliability. Due to the complex hierarchy and single path of traditional tree-shaped or three-tier network architectures, network congestion and single-point failures are likely to occur, making it difficult to meet the requirements of modern high-performance computing.

[0003] As an emerging flat network topology structure, the Spine-Leaf network architecture provides high-bandwidth and low-latency data transmission capabilities by fully interconnecting Spine switches and Leaf switches. However, with the further expansion of the network scale, the traditional two-tier Spine-Leaf architecture still faces problems such as insufficient scalability and complex traffic management in ultra-large-scale data centers and cross-regional network scenarios. For example, when the number of switches in the Leaf layer and the Spine layer increases, the fully interconnected design may lead to a sharp increase in hardware costs and wiring complexity; at the same time, the need for cross-regional data transmission also makes it difficult for the traditional two-tier architecture to efficiently manage global traffic.

[0004] In addition, as a high-performance interconnection technology, the InfiniBand (IB) network has the advantages of high bandwidth, low latency, and low CPU overhead, and is widely used in high-performance computing and data centers. However, existing Spine-Leaf IB network architectures still have deficiencies in traffic dispersion, fault recovery, and dynamic routing optimization. Especially in a multi-level network environment, how to efficiently manage cross-layer traffic, avoid single-point failures, and achieve flexible scalability is still a technical problem.

[0005] Therefore, it is necessary to design a data transmission system and method for a multi-level Spine-Leaf IB network architecture to solve the problems existing in the current technology. Summary of the Invention

[0006] In view of this, the present invention proposes a data transmission system and method for a multi-level Spine-Leaf IB network architecture, aiming to effectively disperse traffic and reduce the impact of single-point failures.

[0007] In one aspect, the present invention provides a data transmission system for a multi-level Spine-Leaf IB network architecture, including: a Spine layer, a Leaf layer, and a transmission control module. The Spine layer is connected to a plurality of the Leaf layers, and each Leaf layer is connected to a plurality of computing nodes; the transmission control module is communicatively connected to the Spine layer, the Leaf layer, and the computing nodes. The transmission control module includes a collection unit, a judgment unit, an optimization unit, and a storage unit; The collection unit is configured to determine the source node and the destination node of the data to be transmitted, collect the location information of the source node and the destination node, parse the location information, and determine the transmission mode of the data to be transmitted based on the parsing result; collect the transmission requirement information of the data to be transmitted, and determine the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission mode and the transmission requirement information; The judgment unit is configured to control the collection unit to collect the real-time status information of the data to be transmitted during the transmission process, calculate the transmission stability index according to the real-time status information, and determine whether to optimize the initial path traffic allocation value according to the transmission stability index; The optimization unit is configured to, when the judgment unit determines to optimize the initial path traffic allocation value, control the collection unit to collect the real-time network status information of each initial transmission path, calculate the network transmission impact index according to the real-time network status information, compare the network transmission impact index with historical data, determine the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtain the optimized path traffic allocation value; The storage unit is configured to store the network transmission impact index.

[0008] Further, when the collection unit parses the location information and determines the transmission mode of the data to be transmitted based on the parsing result, it includes: When the location information shows that the source node and the destination node are connected to the same Leaf layer, it is determined that the transmission mode of the data to be transmitted is local transmission; When the location information shows that the source node and the destination node are not connected to the same Leaf layer, it is determined that the transmission mode of the data to be transmitted is cross-Leaf layer transmission.

[0009] Further, when the collection unit determines the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission mode and the transmission requirement information, it includes: The transmission requirement information includes the amount of data to be transmitted, the current time, and the transmission deadline; When the transmission mode of the data to be transmitted is local transmission, determine that the number of initial transmission paths is one initial transmission path, and determine the initial path traffic allocation value of the initial transmission path according to the amount of transmitted data; When the amount of transmitted data is less than or equal to the preset amount of transmitted data, determine that the initial path traffic allocation value of the initial transmission path is the first path traffic allocation value; When the amount of transmitted data is greater than the preset amount of transmitted data, determine that the initial path traffic allocation value of the initial transmission path is the second path traffic allocation value; wherein, the second path traffic allocation value is greater than the first path traffic allocation value; When the transmission mode of the data to be transmitted is cross-Leaf layer transmission, determine the number of initial transmission paths of the data to be transmitted according to the amount of transmitted data, divide the data to be transmitted into data blocks equal in number to the number of initial transmission paths according to the number of initial transmission paths, each data block corresponds to one initial transmission path for transmission, and determine the initial path traffic allocation value of each initial transmission path according to the current time and the transmission deadline.

[0010] Further, when the acquisition unit determines the number of initial transmission paths of the data to be transmitted according to the amount of transmitted data, it includes: Compare the amount of transmitted data with the first amount of transmitted data and the second amount of transmitted data, and determine the number of initial transmission paths of the data to be transmitted according to the comparison result; wherein, the first amount of transmitted data is less than the second amount of transmitted data; When the amount of transmitted data is less than or equal to the first amount of transmitted data, determine that the number of initial transmission paths of the data to be transmitted is the first number; When the amount of transmitted data is greater than the first amount of transmitted data and less than or equal to the second amount of transmitted data, determine that the number of initial transmission paths of the data to be transmitted is the second number, and the second number is greater than the first number; When the amount of transmitted data is greater than the second amount of transmitted data, determine that the number of initial transmission paths of the data to be transmitted is the third number, and the third number is greater than the second number.

[0011] Further, when the acquisition unit determines the initial path traffic allocation value of each initial transmission path according to the current time and the transmission deadline, it includes: Calculate the remaining transmission time according to the transmission deadline and the current time, compare the remaining transmission time with the first remaining transmission time and the second remaining transmission time, and determine the initial path traffic allocation value of each initial transmission path according to the comparison result; wherein, the first remaining transmission time is less than the second remaining transmission time; When the remaining transmission time is greater than or equal to the second remaining transmission time, determine that the initial path traffic allocation value of each of the initial transmission paths is the third path traffic allocation value; When the remaining transmission time is less than the second remaining transmission time and greater than or equal to the first remaining transmission time, determine that the initial path traffic allocation value of each of the initial transmission paths is the fourth path traffic allocation value, and the fourth path traffic allocation value is greater than the third path traffic allocation value; When the remaining transmission time is less than the first remaining transmission time, determine that the initial path traffic allocation value of each of the initial transmission paths is the fifth path traffic allocation value, and the fifth path traffic allocation value is greater than the fourth path traffic allocation value.

[0012] Further, when the determination unit calculates the transmission stability index according to the real-time status information and determines whether to optimize the initial path traffic allocation value according to the transmission stability index, it includes: Analyze the real-time status information to obtain the packet loss rate during the transmission process, the transmission network delay eigenvalue, and the transmission rate eigenvalue; Perform weighted calculation on the packet loss rate during the transmission process, the transmission network delay eigenvalue, and the transmission rate eigenvalue to obtain the transmission stability index; Compare the transmission stability index with the transmission stability threshold, and determine whether to optimize the initial path traffic allocation value according to the comparison result; When the transmission stability index is less than the transmission stability index threshold, it is determined to optimize the initial path traffic allocation value; When the transmission stability index is greater than or equal to the transmission stability index threshold, it is determined not to optimize the initial path traffic allocation value.

[0013] Further, when the optimization unit controls the acquisition unit to acquire the real-time network status information of each initial transmission path and calculates the network transmission influence index according to the real-time network status information, it includes: Acquire the number of links and the number of nodes of each of the initial transmission paths; Analyze the real-time network status information to obtain the link packet loss rate, the link bandwidth utilization rate, the link delay, and the node load; Calculate the network transmission influence index according to the number of links, the number of nodes, the link packet loss rate, the link bandwidth utilization rate, the link delay, and the node load; The network transmission influence index is obtained by the following formula: ; Wherein, Ip represents the network transmission impact index; L represents the number of links; ILk represents the network transmission impact index of the k-th link; PLk represents the link packet loss rate of the k-th link; BWk represents the bandwidth utilization rate of the k-th link; LDk represents the delay time of the k-th link; Np represents the number of nodes; ω1, ω2, and ω3 respectively represent the first weight coefficient, the second weight coefficient, and the third weight coefficient; α1, α2, and α3 respectively represent the first adjustment coefficient, the second adjustment coefficient, and the third adjustment coefficient.

[0014] Further, when the optimization unit compares the network transmission impact index with historical data, determines the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtains the optimized path traffic allocation value, it includes: When there is a historical network transmission impact index in the historical data that is the same as the network transmission impact index, adjust the initial path traffic allocation value according to the historical optimization coefficient corresponding to the historical network transmission impact index, and use the product value of the historical optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When there is no historical network transmission impact index in the historical data that is the same as the network transmission impact index, calculate the difference between the network transmission impact index and the historical network transmission impact indices in the historical data one by one, obtain the minimum difference, determine the optimization coefficient of the initial path traffic allocation value according to the minimum difference, and obtain the optimized path traffic allocation value.

[0015] Further, when the optimization unit determines the optimization coefficient of the initial path traffic allocation value according to the minimum difference and obtains the optimized path traffic allocation value, it includes: Compare the minimum difference with a first minimum difference and a second minimum difference, and determine the optimization coefficient of the initial path traffic allocation value according to the comparison result; wherein, the first minimum difference is less than the second minimum difference; Set an optimization coefficient interval, wherein the optimization coefficient interval includes a first optimization coefficient, a second optimization coefficient, and a third optimization coefficient, and the first optimization coefficient is less than the second optimization coefficient, and the second optimization coefficient is less than the third optimization coefficient; When the minimum difference is less than or equal to the first minimum difference, determine the optimization coefficient of the initial path traffic allocation value as the first optimization coefficient, and use the product value of the first optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When the minimum difference is greater than the first minimum difference and less than or equal to the second minimum difference, determine that the optimization coefficient of the initial path traffic allocation value is the second optimization coefficient, and use the product value of the second optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When the minimum difference is greater than the second minimum difference, determine that the optimization coefficient of the initial path traffic allocation value is the third optimization coefficient, and use the product value of the third optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: The data transmission system of the multi-level Spine-Leaf IB network architecture provided by the present invention can intelligently allocate transmission paths and traffic according to the actual transmission data volume and transmission requirements, thereby improving the efficiency and stability of data transmission. First, the system flexibly determines the initial number of transmission paths by comparing the transmission data volume with the preset first transmission data volume and second transmission data volume, which ensures that data can obtain appropriate transmission resources under different load conditions. Second, the system calculates the initial path traffic allocation value of each initial transmission path according to the remaining transmission time and the current time, which further improves the flexibility and response speed of data transmission. Furthermore, the system calculates the transmission stability index by real-time monitoring the status information during the transmission process, and determines whether it is necessary to optimize the initial path traffic allocation value based on this, which enhances the reliability and stability of data transmission. Finally, when optimization is required, the system calculates the network transmission impact index according to the real-time network status information, determines the optimization coefficient by comparing with historical data, so as to obtain the optimized path traffic allocation value, which further improves the performance and efficiency of data transmission.

[0017] In summary, the data transmission system of the multi-level Spine-Leaf IB network architecture provided by the present invention has significant advantages, can meet the high requirements of large data centers for data transmission, and effectively disperses traffic and reduces the impact of single-point failures.

[0018] On the other hand, the present invention also proposes a data transmission method for a multi-level Spine-Leaf IB network architecture, including the following steps: S100: Determine the source node and target node of the data to be transmitted, collect the location information of the source node and target node, parse the location information, and determine the transmission method of the data to be transmitted based on the parsing result; collect the transmission requirement information of the data to be transmitted, and determine the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission method and transmission requirement information; S200: Collect the real-time status information of the data to be transmitted during the transmission process, calculate the transmission stability index according to the real-time status information, and determine whether to optimize the initial path traffic allocation value according to the transmission stability index; S300: When the determination unit determines to optimize the initial path traffic allocation value, collect the real-time network status information of each initial transmission path, calculate the network transmission impact index according to the real-time network status information, compare the network transmission impact index with historical data, determine the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtain the optimized path traffic allocation value; S400: Store the network transmission impact index.

[0019] It can be understood that the above-mentioned multi-level Spine-Leaf IB network architecture data transmission system and method have the same beneficial effects, which will not be elaborated here. Brief Description of the Drawings

[0020] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings: Figure 1 It is a functional block diagram of a data transmission system with a multi-level Spine-Leaf IB network architecture provided by an embodiment of the present invention; Figure 2 It is a flowchart of a data transmission method with a multi-level Spine-Leaf IB network architecture provided by an embodiment of the present invention. Detailed Embodiments

[0021] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. Hereinafter, the present invention will be described in detail with reference to the drawings and in combination with the embodiments.

[0022] Refer to Figure 1 As shown, in some embodiments of the present application, this embodiment provides a data transmission system with a multi-level Spine-Leaf IB network architecture, including: Spine layer, Leaf layer and transmission control module. A number of Leaf layers are connected to the Spine layer, and a number of computing nodes are connected to each Leaf layer; the transmission control module is communicatively connected to the Spine layer, Leaf layer and computing nodes. The transmission control module includes a collection unit, a judgment unit, an optimization unit and a storage unit; The collection unit is configured to determine the source node and target node of the data to be transmitted, collect the location information of the source node and target node, parse the location information, and determine the transmission mode of the data to be transmitted based on the parsing result; collect the transmission requirement information of the data to be transmitted, and determine the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission mode and transmission requirement information; The judgment unit is configured to control the collection unit to collect the real-time status information of the data to be transmitted during the transmission process, calculate the transmission stability index according to the real-time status information, and judge whether to optimize the initial path traffic allocation value according to the transmission stability index; The optimization unit is configured to, when the judgment unit determines to optimize the initial path traffic allocation value, control the collection unit to collect the real-time network status information of each initial transmission path, calculate the network transmission impact index according to the real-time network status information, compare the network transmission impact index with historical data, determine the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtain the optimized path traffic allocation value; The storage unit is configured to store the network transmission impact index.

[0023] It can be understood that Spine-Leaf IB (Spine-Leaf Infiniband) is the Infiniband network architecture of the Spine-Leaf structure. In this architecture, the Spine layer serves as the core switching layer for high-speed and low-latency data forwarding, while the Leaf layer serves as the edge access layer responsible for connecting each computing node. The advantage of this architecture lies in its high scalability, high reliability and high performance, which can meet the high requirements for data transmission in large data centers.

[0024] It can be understood that the data transmission system of the multi-level Spine-Leaf IB network architecture provided in this embodiment can intelligently allocate transmission paths and traffic according to the actual transmission data volume and transmission requirements, thereby improving the efficiency and stability of data transmission. First, the system flexibly determines the initial number of transmission paths by comparing the transmission data volume with the preset first and second transmission data volumes, which ensures that appropriate transmission resources can be obtained for data under different load conditions. Second, the system calculates the initial path traffic allocation value for each initial transmission path based on the remaining transmission time and the current time, which further improves the flexibility and response speed of data transmission. Furthermore, the system calculates the transmission stability index by real-time monitoring the status information during the transmission process and determines whether it is necessary to optimize the initial path traffic allocation value based on this, which enhances the reliability and stability of data transmission. Finally, when optimization is required, the system calculates the network transmission impact index based on the real-time network status information, determines the optimization coefficient by comparing with historical data, and thus obtains the optimized path traffic allocation value, which further improves the performance and efficiency of data transmission. In summary, the data transmission system of the multi-level Spine-Leaf IB network architecture provided in this embodiment has significant advantages and can meet the high requirements for data transmission in large data centers.

[0025] Specifically, when the acquisition unit analyzes the location information and determines the transmission method of the data to be transmitted based on the analysis result, it includes: When the location information shows that the source node and the target node are connected to the same Leaf layer, it is determined that the transmission method of the data to be transmitted is local transmission; When the location information shows that the source node and the target node are not connected to the same Leaf layer, it is determined that the transmission method of the data to be transmitted is cross-Leaf layer transmission.

[0026] Specifically, when the acquisition unit determines the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission method and transmission requirement information, it includes: The transmission requirement information includes the transmission data volume, the current time, and the transmission deadline; When the transmission method of the data to be transmitted is local transmission, it is determined that the initial number of transmission paths is one initial transmission path, and the initial path traffic allocation value of the initial transmission path is determined according to the transmission data volume; When the transmission data volume is less than or equal to the preset transmission data volume, it is determined that the initial path traffic allocation value of the initial transmission path is the first path traffic allocation value; When the amount of transmitted data is greater than a preset amount of transmitted data, determine that the initial path traffic allocation value of the initial transmission path is the second path traffic allocation value; wherein, the second path traffic allocation value is greater than the first path traffic allocation value. When the transmission mode of the data to be transmitted is cross-Leaf layer transmission, determine the number of initial transmission paths of the data to be transmitted according to the amount of transmitted data, divide the data to be transmitted into data blocks with the same number as the number of initial transmission paths according to the number of initial transmission paths, each data block corresponds to an initial transmission path for transmission, and determine the initial path traffic allocation value of each initial transmission path according to the current time and the transmission deadline.

[0027] It can be understood that when determining the initial path traffic allocation value of each initial transmission path, the system comprehensively considers the urgency of the data and the size of the data to be transmitted. If the current time is close to the transmission deadline, the system will tend to allocate more network resources to the data to be transmitted to ensure that the data can be transmitted on time. This dynamic resource allocation strategy enables the system to make flexible adjustments according to the actual transmission requirements and network conditions, further improving the efficiency and reliability of data transmission.

[0028] Specifically, when the acquisition unit determines the number of initial transmission paths of the data to be transmitted according to the amount of transmitted data, it includes: Compare the amount of transmitted data with a first amount of transmitted data and a second amount of transmitted data, and determine the number of initial transmission paths of the data to be transmitted according to the comparison result; wherein, the first amount of transmitted data is less than the second amount of transmitted data. When the amount of transmitted data is less than or equal to the first amount of transmitted data, determine that the number of initial transmission paths of the data to be transmitted is the first number. When the amount of transmitted data is greater than the first amount of transmitted data and less than or equal to the second amount of transmitted data, determine that the number of initial transmission paths of the data to be transmitted is the second number, and the second number is greater than the first number. When the amount of transmitted data is greater than the second amount of transmitted data, determine that the number of initial transmission paths of the data to be transmitted is the third number, and the third number is greater than the second number.

[0029] It can be understood that through this flexible path quantity determination method, the system can ensure that appropriate transmission resources can be provided under various data volume conditions, avoiding waste or insufficiency of resources. When the data volume is small, using fewer transmission paths can reduce the network overhead and complexity; while when the data volume is large, increasing the number of transmission paths can disperse the transmission load and improve the transmission efficiency. In addition, the system also optimizes by real-time monitoring the network status and according to the network transmission impact index, further enhancing the stability and performance of data transmission. This intelligent data transmission strategy enables the multi-level Spine-Leaf IB network architecture to better adapt to the complex and changeable transmission requirements of large data centers, providing strong support for data transmission.

[0030] Specifically, when the acquisition unit determines the initial path traffic allocation value of each of the initial transmission paths according to the current time and the transmission deadline, it includes: Calculating the remaining transmission time according to the transmission deadline and the current time, comparing the remaining transmission time with a first remaining transmission time and a second remaining transmission time, and determining the initial path traffic allocation value of each of the initial transmission paths according to the comparison result; wherein, the first remaining transmission time is less than the second remaining transmission time; When the remaining transmission time is greater than or equal to the second remaining transmission time, determining that the initial path traffic allocation value of each of the initial transmission paths is a third path traffic allocation value; When the remaining transmission time is less than the second remaining transmission time and greater than or equal to the first remaining transmission time, determining that the initial path traffic allocation value of each of the initial transmission paths is a fourth path traffic allocation value, and the fourth path traffic allocation value is greater than the third path traffic allocation value; When the remaining transmission time is less than the first remaining transmission time, determining that the initial path traffic allocation value of each of the initial transmission paths is a fifth path traffic allocation value, and the fifth path traffic allocation value is greater than the fourth path traffic allocation value.

[0031] It can be understood that when determining the initial path traffic allocation value of each initial transmission path, the system fully considers the urgency of the data. If the remaining transmission time is long, the system can allocate fewer network resources to avoid waste of resources; while when the remaining transmission time is short, the system will appropriately increase the proportion of resource allocation to ensure that the data can be transmitted on time. This strategy of dynamically adjusting resource allocation according to the urgency of the data not only improves the flexibility of data transmission, but also further enhances the reliability and stability of the system.

[0032] Specifically, when the determination unit calculates the transmission stability index according to the real-time status information and determines whether to optimize the initial path traffic allocation value based on the transmission stability index, it includes: Analyze the real-time status information to obtain the packet loss rate during the transmission process, the transmission network delay eigenvalue, and the transmission rate eigenvalue; Perform weighted calculation on the packet loss rate during the transmission process, the transmission network delay eigenvalue, and the transmission rate eigenvalue to obtain the transmission stability index; Compare the transmission stability index with the transmission stability threshold, and determine whether to optimize the initial path traffic allocation value according to the comparison result; When the transmission stability index is less than the transmission stability index threshold, it is determined to optimize the initial path traffic allocation value; When the transmission stability index is greater than or equal to the transmission stability index threshold, it is determined not to optimize the initial path traffic allocation value.

[0033] It can be understood that by deeply analyzing and calculating the real-time status information during the transmission process, the system can accurately evaluate the current transmission stability and make a decision on whether to optimize accordingly. This dynamic optimization strategy based on real-time data enables the system to promptly detect and solve potential transmission problems, thereby ensuring the continuous stability and high efficiency of data transmission. When the transmission stability index is lower than the set threshold, the system will immediately initiate the optimization process and adjust the path traffic allocation value to enhance the transmission stability and efficiency. This intelligent optimization mechanism not only improves the reliability of data transmission but also further reduces the risk of service interruption caused by transmission problems.

[0034] It can be understood that the packet loss rate during the transmission process, the transmission network delay eigenvalue, and the transmission rate eigenvalue respectively refer to the proportion of lost data packets during data transmission, the degree of delay generated when data is transmitted in the network, and the number of bits or bytes transmitted per unit time of data. These parameters are key indicators for evaluating data transmission performance. Specifically, the packet loss rate during the transmission process reflects the integrity of data transmission, and an excessively high packet loss rate may lead to data loss or transmission failure; the transmission network delay eigenvalue reflects the real-time nature of data transmission, and excessive delay may affect the timely arrival and processing of data; the transmission rate eigenvalue is directly related to the efficiency of data transmission, and a higher transmission rate means that data can be transmitted in a shorter time.

[0035] Specifically, when the optimization unit controls the acquisition unit to acquire the real-time network status information of each initial transmission path and calculates the network transmission impact index according to the real-time network status information, it includes: Acquire the number of links and the number of nodes of each of the initial transmission paths; Parse the real-time network status information to obtain the link packet loss rate, link bandwidth utilization rate, link latency, and node load; Calculate the network transmission impact index based on the number of links, number of nodes, link packet loss rate, link bandwidth utilization rate, link latency, and node load; The network transmission impact index is obtained through the following formula: ; where, Ip represents the network transmission impact index; L represents the number of links; ILk represents the network transmission impact index of the k-th link; PLk represents the link packet loss rate of the k-th link; BWk represents the bandwidth utilization rate of the k-th link; LDk represents the latency time of the k-th link; Np represents the number of nodes; ω1, ω2, and ω3 respectively represent the first weight coefficient, the second weight coefficient, and the third weight coefficient; α1, α2, and α3 respectively represent the first adjustment coefficient, the second adjustment coefficient, and the third adjustment coefficient.

[0036] It can be understood that a transmission path includes multiple links and multiple nodes, and each link and node may affect the transmission performance. The link packet loss rate reflects the reliability of link transmission. An excessively high packet loss rate may lead to data retransmission or transmission failure; the link bandwidth utilization rate reflects the transmission capacity of the link. An excessively high bandwidth utilization rate may lead to a transmission bottleneck; the link latency affects the real-time nature of data transmission. An excessively large latency may cause data to not arrive in time; the node load reflects the processing capacity of the node. An excessively high load may cause the node to become a transmission bottleneck. By comprehensively considering these parameters, the system can more comprehensively evaluate the performance of the transmission path, thereby providing a more accurate basis for optimizing the path traffic allocation value.

[0037] It can be understood that the calculation of the network transmission impact index Ip comprehensively considers multiple factors such as the link packet loss rate PLk, link bandwidth utilization rate BWk, link latency LDk, and node load. The link packet loss rate reflects the reliability of the link during data transmission. An excessively high packet loss rate may lead to data loss or transmission failure; the link bandwidth utilization rate reflects the resource utilization of the link. An excessively high bandwidth utilization rate may lead to a transmission bottleneck and affect the transmission efficiency; the link latency is related to the real-time nature of data transmission. An excessively large latency may affect the timely arrival and processing of data; and the node load reflects the processing capacity of the nodes in the network. An excessively high node load may lead to processing latency or transmission interruption. By comprehensively considering these factors and introducing corresponding weight coefficients and adjustment coefficients, the system can accurately evaluate the impact of network transmission and provide strong support for subsequent optimization decisions. This refined network status monitoring and analysis mechanism enables the multi-level Spine-Leaf IB network architecture to better adapt to complex and changing transmission environments and ensure the efficiency and stability of data transmission.

[0038] In this embodiment, the values of ω1, ω2 and ω3 are preferably 0.4, 0.3 and 0.3, and the values of α1, α2 and α3 are preferably 0.5, 0.6 and 0.7. The values of these weight coefficients and adjustment coefficients are set according to the actual network transmission characteristics and transmission requirements, aiming to more accurately reflect the influence degree of each factor on the transmission performance and provide a more accurate basis for subsequent optimization decisions. By finely adjusting these parameters, the system can further optimize the allocation of network resources and improve the stability and efficiency of data transmission.

[0039] Specifically, when the optimization unit compares the network transmission influence index with historical data, determines the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtains the optimized path traffic allocation value, it includes: When there is a historical network transmission influence index identical to the network transmission influence index in the historical data, adjust the initial path traffic allocation value according to the historical optimization coefficient corresponding to the historical network transmission influence index, and use the product value of the historical optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When there is no historical network transmission influence index identical to the network transmission influence index in the historical data, calculate the difference between the network transmission influence index and the historical network transmission influence indices in the historical data one by one, obtain the minimum difference, determine the optimization coefficient of the initial path traffic allocation value according to the minimum difference, and obtain the optimized path traffic allocation value.

[0040] It can be understood that the minimum difference refers to the minimum value among the differences between the network transmission influence index and each historical network transmission influence index in the historical data.

[0041] It can be understood that through this way of comparing with historical data, the system can quickly find a historical situation similar to the current network transmission status and determine a suitable optimization coefficient accordingly. When there is a situation in the historical data that matches the current network transmission influence index, the system can directly use the historical optimization coefficient for adjustment, which greatly improves the efficiency and accuracy of the optimization decision. When there is no completely matching situation in the historical data, the system calculates the difference to find the historical network transmission influence index that is closest, and determines the optimization coefficient accordingly. This approximation method can also provide strong support for the optimization decision. This intelligent optimization strategy based on historical data not only improves the stability and efficiency of data transmission, but also reduces the risk of service interruption caused by network transmission problems, providing a more reliable guarantee for data transmission in large data centers.

[0042] Specifically, when the optimization unit determines the optimization coefficient of the initial path traffic allocation value based on the minimum difference and obtains the optimized path traffic allocation value, it includes: Comparing the minimum difference with a first minimum difference and a second minimum difference, and determining the optimization coefficient of the initial path traffic allocation value according to the comparison result; wherein, the first minimum difference is less than the second minimum difference; Setting an optimization coefficient interval, where the optimization coefficient interval includes a first optimization coefficient, a second optimization coefficient, and a third optimization coefficient, and the first optimization coefficient is less than the second optimization coefficient, and the second optimization coefficient is less than the third optimization coefficient; When the minimum difference is less than or equal to the first minimum difference, determining the optimization coefficient of the initial path traffic allocation value as the first optimization coefficient, and taking the product value of the first optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When the minimum difference is greater than the first minimum difference and less than or equal to the second minimum difference, determining the optimization coefficient of the initial path traffic allocation value as the second optimization coefficient, and taking the product value of the second optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When the minimum difference is greater than the second minimum difference, determining the optimization coefficient of the initial path traffic allocation value as the third optimization coefficient, and taking the product value of the third optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value.

[0043] It can be understood that the first minimum difference and the second minimum difference are two system - preset thresholds used to judge the range where the minimum difference is located, so as to determine the corresponding optimization coefficient. By setting these two thresholds, the system can divide the minimum difference into different intervals, and each interval corresponds to an optimization coefficient.

[0044] It can be understood that when determining the optimization coefficient, the system adopts a more refined grading strategy. By setting different minimum difference intervals and assigning corresponding optimization coefficients to each interval, the system can more accurately adjust the initial path traffic allocation value according to the current network transmission status. When the minimum difference is small, it indicates that the current network transmission status is very close to some situations in the historical data. At this time, the system can use a smaller optimization coefficient for fine - tuning to avoid unstable factors caused by excessive adjustment. When the minimum difference is large, it indicates that there are significant differences between the current network transmission status and the historical data. At this time, the system needs to use a larger optimization coefficient for larger adjustments to ensure the stability and efficiency of data transmission. This grading optimization strategy not only improves the fineness of the optimization decision but also enhances the adaptability and robustness of the system.

[0045] Refer to Figure 2 As shown, in some embodiments of the present application, this embodiment provides a data transmission method for a multi-level Spine-Leaf IB network architecture, including the following steps: S100: Determine the source node and the destination node of the data to be transmitted, collect the location information of the source node and the destination node, parse the location information, and determine the transmission mode of the data to be transmitted based on the parsing result; collect the transmission requirement information of the data to be transmitted, and determine the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission mode and the transmission requirement information; S200: Collect the real-time status information of the data to be transmitted during the transmission process, calculate the transmission stability index according to the real-time status information, and determine whether to optimize the initial path traffic allocation value according to the transmission stability index; S300: When the determination unit determines to optimize the initial path traffic allocation value, collect the real-time network status information of each initial transmission path, calculate the network transmission impact index according to the real-time network status information, compare the network transmission impact index with historical data, determine the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtain the optimized path traffic allocation value; S400: Store the network transmission impact index.

[0046] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0047] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0048] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction means that implements the functions specified in one or more of the processes and / or blocks Figure 1 of the process or processes and / or boxes Figure 1 or boxes.

[0049] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the processes and / or blocks Figure 1 of the process or processes and / or boxes Figure 1 or boxes.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent substitutions can still be made to the specific embodiments of the present invention. Any modification or equivalent substitution that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A data transmission system for a multi-level Spine-Leaf IB network architecture, characterized in that, Including: A Spine layer, a Leaf layer, and a transmission control module. The Spine layer is connected to a plurality of the Leaf layers, and each Leaf layer is connected to a plurality of computing nodes; the transmission control module is communicatively connected to the Spine layer, the Leaf layer, and the computing nodes. The transmission control module includes a collection unit, a judgment unit, an optimization unit, and a storage unit; The collection unit is configured to determine the source node and the destination node of the data to be transmitted, collect the location information of the source node and the destination node, parse the location information, and determine the transmission mode of the data to be transmitted based on the parsing result; Collect the transmission requirement information of the data to be transmitted, and determine the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission mode and the transmission requirement information; The judgment unit is configured to control the collection unit to collect the real-time status information of the data to be transmitted during the transmission process, calculate the transmission stability index according to the real-time status information, and judge whether to optimize the initial path traffic allocation value according to the transmission stability index; The optimization unit is configured to, when the judgment unit determines to optimize the initial path traffic allocation value, control the collection unit to collect the real-time network status information of each initial transmission path, calculate the network transmission impact index according to the real-time network status information, compare the network transmission impact index with historical data, determine the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtain the optimized path traffic allocation value; The storage unit is configured to store the network transmission impact index.

2. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 1, wherein When the collection unit parses the location information and determines the transmission mode of the data to be transmitted based on the parsing result, it includes: When the location information shows that the source node and the destination node are connected to the same Leaf layer, determine that the transmission mode of the data to be transmitted is local transmission; When the location information shows that the source node and the destination node are not connected to the same Leaf layer, determine that the transmission mode of the data to be transmitted is cross-Leaf layer transmission.

3. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 2, wherein When the collection unit determines the initial number of transmission paths and the initial path traffic allocation value of the data to be transmitted based on the transmission mode and the transmission requirement information, it includes: The transmission requirement information includes the amount of data to be transmitted, the current time, and the transmission deadline; When the transmission mode of the data to be transmitted is local transmission, determine that the initial number of transmission paths is one initial transmission path, and determine the initial path traffic allocation value of the initial transmission path according to the amount of data to be transmitted; When the amount of data to be transmitted is less than or equal to the preset amount of data to be transmitted, determine that the initial path traffic allocation value of the initial transmission path is the first path traffic allocation value; When the amount of data to be transmitted is greater than the preset amount of data to be transmitted, determine that the initial path traffic allocation value of the initial transmission path is the second path traffic allocation value; wherein, the second path traffic allocation value is greater than the first path traffic allocation value. When the transmission mode of the data to be transmitted is cross-Leaf layer transmission, determine the initial number of transmission paths for the data to be transmitted according to the amount of transmitted data, divide the data to be transmitted into the same number of data blocks as the initial number of transmission paths according to the initial number of transmission paths, each data block is transmitted corresponding to one initial transmission path, and determine the initial path traffic allocation value for each initial transmission path according to the current time and the transmission deadline.

4. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 3, wherein When the acquisition unit determines the initial number of transmission paths for the data to be transmitted according to the amount of transmitted data, it includes: Compare the amount of transmitted data with a first amount of transmitted data and a second amount of transmitted data, and determine the initial number of transmission paths for the data to be transmitted according to the comparison result; wherein, the first amount of transmitted data is less than the second amount of transmitted data; When the amount of transmitted data is less than or equal to the first amount of transmitted data, determine that the initial number of transmission paths for the data to be transmitted is a first number; When the amount of transmitted data is greater than the first amount of transmitted data and less than or equal to the second amount of transmitted data, determine that the initial number of transmission paths for the data to be transmitted is a second number, and the second number is greater than the first number; When the amount of transmitted data is greater than the second amount of transmitted data, determine that the initial number of transmission paths for the data to be transmitted is a third number, and the third number is greater than the second number.

5. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 4, wherein When the acquisition unit determines the initial path traffic allocation value for each initial transmission path according to the current time and the transmission deadline, it includes: Calculate the remaining transmission time according to the transmission deadline and the current time, compare the remaining transmission time with a first remaining transmission time and a second remaining transmission time, and determine the initial path traffic allocation value for each initial transmission path according to the comparison result; wherein, the first remaining transmission time is less than the second remaining transmission time; When the remaining transmission time is greater than or equal to the second remaining transmission time, determine that the initial path traffic allocation value for each initial transmission path is a third path traffic allocation value; When the remaining transmission time is less than the second remaining transmission time and greater than or equal to the first remaining transmission time, determine that the initial path traffic allocation value for each initial transmission path is a fourth path traffic allocation value, and the fourth path traffic allocation value is greater than the third path traffic allocation value; When the remaining transmission time is less than the first remaining transmission time, determine that the initial path traffic allocation value for each initial transmission path is a fifth path traffic allocation value, and the fifth path traffic allocation value is greater than the fourth path traffic allocation value.

6. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 5, characterized in that, When the judgment unit calculates the transmission stability index according to the real-time status information and determines whether to optimize the initial path traffic allocation value according to the transmission stability index, it includes: Analyze the real-time status information to obtain the packet loss rate during the transmission process, the transmission network delay characteristic value, and the transmission rate characteristic value; Perform weighted calculations on the packet loss rate, transmission network delay eigenvalue, and transmission rate eigenvalue of the transmission process to obtain the transmission stability index; Compare the transmission stability index with the transmission stability threshold, and determine whether to optimize the initial path traffic allocation value according to the comparison result; When the transmission stability index is less than the transmission stability index threshold, it is determined to optimize the initial path traffic allocation value; When the transmission stability index is greater than or equal to the transmission stability index threshold, it is determined not to optimize the initial path traffic allocation value.

7. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 6, wherein When the optimization unit controls the acquisition unit to collect the real-time network status information of each initial transmission path and calculates the network transmission impact index according to the real-time network status information, it includes: Collect the number of links and the number of nodes of each of the initial transmission paths; Parse the real-time network status information to obtain the link packet loss rate, link bandwidth utilization rate, link delay, and node load; Calculate the network transmission impact index according to the number of links, the number of nodes, the link packet loss rate, the link bandwidth utilization rate, the link delay, and the node load; The network transmission impact index is obtained by the following formula: ; Where, Ip represents the network transmission impact index; L represents the number of links; ILk represents the network transmission impact index of the kth link; PLk represents the link packet loss rate of the kth link; BWk represents the bandwidth utilization rate of the kth link; LDk represents the delay time of the kth link; Np represents the number of nodes; ω1, ω2, and ω3 respectively represent the first weight coefficient, the second weight coefficient, and the third weight coefficient; α1, α2, and α3 respectively represent the first adjustment coefficient, the second adjustment coefficient, and the third adjustment coefficient.

8. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 7, wherein When the optimization unit compares the network transmission impact index with historical data, determines the optimization coefficient corresponding to the initial path traffic allocation value of each initial transmission path according to the comparison result, and obtains the optimized path traffic allocation value, it includes: When there is a historical network transmission impact index in the historical data that is the same as the network transmission impact index, adjust the initial path traffic allocation value according to the historical optimization coefficient corresponding to the historical network transmission impact index, and use the product value of the historical optimization coefficient and the initial path traffic allocation value as the optimized path traffic allocation value; When there is no historical network transmission impact index in the historical data that is the same as the network transmission impact index, calculate the difference between the network transmission impact index and the historical network transmission impact indices in the historical data one by one, and obtain the minimum difference. Determine the optimization coefficient of the initial path traffic allocation value according to the minimum difference, and obtain the optimized path traffic allocation value.

9. The data transmission system of the multi-level Spine-Leaf IB network architecture according to claim 8, wherein, When the optimization unit determines the optimization coefficient of the initial path traffic allocation value according to the minimum difference and obtains the optimized path traffic allocation value, it includes: Compare the minimum difference with the first minimum difference and the second minimum difference, and determine the optimization coefficient of the initial path traffic allocation value according to the comparison result; where, the first minimum difference is less than the second minimum difference; Set an optimization coefficient interval, where the optimization coefficient interval includes a first optimization coefficient, a second optimization coefficient, and a third optimization coefficient, and the first optimization coefficient is less than the second optimization coefficient, and the second optimization coefficient is less than the third optimization coefficient; When the minimum difference is less than or equal to the first minimum difference, determine the optimization coefficient of the initial path flow allocation value as the first optimization coefficient, and use the product value of the first optimization coefficient and the initial path flow allocation value as the optimized path flow allocation value; When the minimum difference is greater than the first minimum difference and less than or equal to the second minimum difference, determine the optimization coefficient of the initial path flow allocation value as the second optimization coefficient, and use the product value of the second optimization coefficient and the initial path flow allocation value as the optimized path flow allocation value; When the minimum difference is greater than the second minimum difference, determine the optimization coefficient of the initial path flow allocation value as the third optimization coefficient, and use the product value of the third optimization coefficient and the initial path flow allocation value as the optimized path flow allocation value.

10. A data transmission method for a multi-level Spine-Leaf IB network architecture, which is applied to a data transmission system of the multi-level Spine-Leaf IB network architecture according to any one of claims 1-9, characterized in that, Comprising: Determine the source node and the destination node of the data to be transmitted, collect the location information of the source node and the destination node, parse the location information, and determine the transmission mode of the data to be transmitted based on the parsing result; Collect the transmission requirement information of the data to be transmitted, and determine the initial number of transmission paths and the initial path flow allocation value of the data to be transmitted based on the transmission mode and the transmission requirement information; Collect the real-time status information of the data to be transmitted during the transmission process, calculate the transmission stability index according to the real-time status information, and determine whether to optimize the initial path flow allocation value according to the transmission stability index; When the judgment unit determines to optimize the initial path flow allocation value, collect the real-time network status information of each initial transmission path, calculate the network transmission impact index according to the real-time network status information, compare the network transmission impact index with historical data, and determine the optimization coefficient corresponding to the initial path flow allocation value of each initial transmission path according to the comparison result, and obtain the optimized path flow allocation value; Store the network transmission impact index.

Images