Flow control methods, devices, communication equipment, storage media, and computer program products
By using precise flow control technology on switches, congestion is transferred to the first-hop device. Combined with the collaborative monitoring of routers and switches, the problems of network congestion and packet loss in the distributed intelligent computing architecture are solved, achieving lossless transmission of data streams and stable network operation.
Patent Information
- Application Number
- CN202510634946.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-16
- Publication Date
- 2026-11-17
AI Technical Summary
In distributed intelligent computing architectures, network congestion and packet loss lead to a decline in communication reliability. Existing load balancing technologies cannot achieve perfect path balancing, with some links congested and others idle, affecting the reliability of communication between devices.
By using precise flow control technology in switches, congestion is transferred to the first-hop device in the network. The data flow transmission rate is adjusted based on congestion information. Combined with the collaborative monitoring and cache scheduling of routers and switches, independent monitoring and dynamic adjustment of each data flow can be achieved.
This avoids congestion, ensures lossless data transmission and stable network operation, and improves the stability and reliability of the communication network.
Smart Images

Figure CN123123263A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of wireless communication technology, and in particular to a flow control method, apparatus, communication device, storage medium, and computer program product. Background Technology
[0002] Intelligent computing centers are infrastructures that integrate high-performance computing, large-scale data processing, and artificial intelligence algorithm support. They provide efficient and intelligent computing services, driving the intelligent transformation of various industries. With the development of intelligent computing services, large-scale model training has become a key force driving the progress of artificial intelligence. As the scale of model parameters further expands, the demand for computing power also increases. Generally, multiple intelligent computing centers can be interconnected through a network to form a large-scale virtual intelligent computing cluster to meet the training needs of ultra-large-scale models.
[0003] However, in such a distributed architecture, network congestion and packet loss become major challenges hindering the improvement of intelligent computing efficiency. Even when network equipment is operating normally, abnormal scenarios may occur during actual business deployment, such as intermittent optical module outages, long-distance link bit error packet loss, and server-side congestion leading to a decrease in data reception capacity. These situations can all cause congestion problems, resulting in a decrease in the communication reliability of the communication network. Summary of the Invention
[0004] This application provides a flow control method, apparatus, communication device, storage medium, and computer program product, which can independently monitor and dynamically adjust each data stream in the network, promptly detect and handle congestion, prevent congestion from spreading, and ensure the stable operation of the overall network.
[0005] A flow control method, applied to a network device, the method comprising:
[0006] If congestion is detected, the data stream that is congested and the congestion information are determined, including one or more of the following: sender, receiver, time, data link, and delay.
[0007] The congestion information is transmitted hop-by-hop in reverse so that the first-hop device corresponding to the sending device receives the congestion information and adjusts the transmission rate based on the congestion information.
[0008] In one embodiment, the method further includes:
[0009] The network status parameters are detected to determine whether congestion has occurred. These network status parameters include one or more of link utilization and buffer occupancy.
[0010] In one embodiment, the method further includes:
[0011] The network status parameters of the transmission link are checked to determine if congestion has occurred; or,
[0012] The devices in the transmission path are inspected to determine if congestion has occurred.
[0013] In one embodiment, determining the congested data stream and congestion information includes:
[0014] The congested data stream and congestion information are then determined based on the 5-tuple information.
[0015] In one embodiment, the step of transmitting the congestion information hop-by-hop in reverse so that the first-hop device corresponding to the sending device receives the congestion information includes:
[0016] Based on the transmission path of the data stream, the congestion information is transmitted in reverse from the congested device until it reaches the first hop device of the data stream's sending device.
[0017] In one embodiment, the first hop device is the device closest to the sending device.
[0018] In one embodiment, the first hop device is a switch or a router.
[0019] A flow control method, applied to a communication device, the method comprising:
[0020] The network device receives congestion information transmitted from a network device. The congestion information includes one or more of the following: sender, receiver, time, data link, and delay. The congestion information is transmitted hop-by-hop in reverse order when the network device detects congestion.
[0021] Based on the congestion information, the transmission rate is adjusted.
[0022] In one embodiment, the congestion information includes identification information of the data stream experiencing congestion; adjusting the transmission rate based on the congestion information includes:
[0023] Based on the data table and the identification information in the congestion information, the rate limiting ratio of the data stream is determined, and the transmission rate of the data stream is adjusted based on the rate limiting ratio.
[0024] In one embodiment, the data table includes at least the identification information, rate limiting ratio, and congestion count for each data stream; determining the rate limiting ratio of the data stream based on the data table and the identification information in the congestion information includes:
[0025] If the identification information is found in the data table, and the number of congestion events corresponding to the identification information is greater than the first target value, then the rate limiting ratio corresponding to the identification information in the data table is used as the rate limiting ratio of the data stream.
[0026] In one embodiment, the data table includes at least the identification information, rate limiting ratio, and congestion count for each data stream; determining the rate limiting ratio of the data stream based on the data table and the identification information in the congestion information includes:
[0027] If the identification information is not found in the data table, or if the number of congestion events corresponding to the identification information is equal to the first target value, then the rate limiting ratio of the data stream is determined based on the real-time network status.
[0028] In one embodiment, the real-time network status includes at least port queue depth and link utilization; determining the rate-limiting ratio of the data flow based on the real-time network status includes:
[0029] The rate limiting ratio of the data stream is determined based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization.
[0030] In one embodiment, determining the rate limiting ratio of the data flow based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization includes:
[0031] Calculate the remaining space ratio of the port queue depth based on the port queue depth and the maximum value of the port queue depth;
[0032] Based on the link utilization rate and the maximum link utilization rate, calculate the remaining space ratio of the link utilization rate.
[0033] Calculate the sum of the remaining space ratio of the port queue depth and the remaining space ratio of the link utilization;
[0034] The minimum value between the second target value and the sum is determined as the rate limiting ratio of the data stream.
[0035] In one embodiment, the data table also includes the time of the most recent congestion corresponding to the identification information.
[0036] In one embodiment, the communication device is the first hop device of the sending device of the data stream.
[0037] In one embodiment, the first hop device is the device closest to the sending device, and / or the first hop device is a switch or a router.
[0038] A flow control device, applied to a network device, the device comprising:
[0039] The detection module is used to determine the data stream experiencing congestion and congestion information if congestion is detected. The congestion information includes one or more of the following: sender, receiver, time, data link, and latency.
[0040] The transmission module is used to transmit the congestion information hop-by-hop in reverse order so that the first hop device corresponding to the sending device receives the congestion information and adjusts the transmission rate based on the congestion information.
[0041] A communication device includes: a transmitter, a processor, and a receiver, wherein:
[0042] The processor is configured to determine, if congestion is detected, the congested data stream and congestion information, the congestion information including one or more of the following: sender, receiver, time, data link, and delay;
[0043] The transmitter is used to transmit the congestion information hop-by-hop in reverse order so that the first-hop device corresponding to the transmitting device receives the congestion information and adjusts the transmission rate based on the congestion information.
[0044] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the embodiments of this application.
[0045] A computer program product includes a computer program that, when executed by a processor, implements the steps described in the embodiments of this application.
[0046] The aforementioned flow control method, apparatus, communication equipment, storage medium, and computer program product, wherein the method includes: network equipment performing real-time congestion detection; upon detecting congestion, determining the congested data stream and corresponding congestion information; and transmitting the congestion information hop-by-hop back to the first-hop device of the data stream's sending end device. The first-hop device can, based on the received congestion information and a backtracking flow control algorithm for remote congestion control, adjust the data stream's transmission rate in a timely and precise manner. This enables cooperation between routers and switches, achieving independent monitoring and dynamic adjustment of each data stream in the communication network. By adjusting the data stream's transmission rate in real time, congestion can be further avoided, ensuring timely detection of congestion and prompt handling after detection to prevent congestion spread and guarantee the stability and reliability of network operation. Attached Figure Description
[0047] Figure 1 This is a diagram illustrating the application environment of the flow control method in one embodiment;
[0048] Figure 2 This is a flowchart illustrating a flow control method in one embodiment;
[0049] Figure 3 This is a flowchart illustrating the flow control method in another embodiment;
[0050] Figure 4 This is a flowchart illustrating the step of determining the speed limit ratio in one embodiment;
[0051] Figure 5 This is a structural block diagram of a flow control device in one embodiment;
[0052] Figure 6 This is a structural block diagram of the flow control device in another embodiment;
[0053] Figure 7 This is an internal structural diagram of a communication device in one embodiment. Detailed Implementation
[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0055] Figure 1 This is a schematic diagram illustrating an application scenario of a flow control method provided in an embodiment of this application. For example... Figure 1 As shown, this scenario includes a transmitting device 100, a receiving device 200, and multiple relay devices 300 (relay nodes). The transmitting device 100 and each relay device 300 transmit data via a network, and the relay devices 300 also transmit data with each other via the network. Each relay device 300 can also transmit data with the receiving device 200 via the network. This scenario can be a distributed architecture and may also include a network device 400, which can communicate with the transmitting device 100, relay devices 300, and receiving device 200 via the network.
[0056] Among them, network equipment 400 can be access network equipment, such as a base station (BTS) in Global System for Mobile communication (GSM) or Code Division Multiple Access (CDMA), a base station (NodeB, NB) in Wideband Code Division Multiple Access (WCDMA), an evolved Node B (eNB or eNodeB) in LTE, a relay station or access point, or a base station in a 5G network, etc., and is not limited here.
[0057] The transmitting device 100 and the receiving device 200 can be wireless terminals. A wireless terminal can be a device that provides voice and / or other service data connectivity to a user, or a handheld device with wireless connectivity, or other processing device connected to a wireless modem. The wireless terminal can communicate with one or more core networks via a Radio Access Network (RAN). The wireless terminal can be a mobile terminal, such as a mobile phone (or "cellular" phone) or a computer with a mobile terminal, for example, a portable, pocket-sized, handheld, computer-embedded, or vehicle-mounted mobile device. These exchange voice and / or data with the radio access network. The wireless terminal can also be referred to as a system, subscriber unit, subscriber station, mobile station, mobile, remote station, remote terminal, access terminal, user terminal, user agent, or user device or user equipment, without limitation herein.
[0058] In traditional technologies, with the rapid increase in parameters of various models, multiple intelligent computing centers are interconnected through networks to form large-scale virtual intelligent computing clusters to meet the needs of ultra-large-scale model training. In such a distributed architecture, network congestion and packet loss are frequent, becoming major challenges hindering the improvement of intelligent computing efficiency. Even when network equipment is operating normally, abnormal scenarios such as intermittent optical module outages, long-distance link bit error packet loss, and server-side congestion leading to a decrease in data reception capacity may occur during actual business deployment. These abnormal scenarios can all cause congestion problems, resulting in abnormal packet loss. Moreover, intelligent computing business flows are characterized by high synchronization, large traffic, and periodic occurrence. The hash-based load balancing technology in related technologies cannot achieve balance across all paths in the distributed network, resulting in congested links and idle links in the distributed network. This makes it difficult to achieve perfect balance across all paths, leading to some links being congested and others being idle, thus affecting the reliability of communication between devices.
[0059] Based on the aforementioned traditional technologies, this application provides a precise flow control technology for switches to transfer congestion on long links to the first-hop device in the network. Based on congestion information, it controls the transmission rate of data streams, avoids congestion, and can also handle congestion in a timely manner to ensure lossless data transmission and stable network operation.
[0060] The intelligent computing center is an infrastructure integrating high-performance computing, large-scale data processing, and artificial intelligence algorithm support. It aims to provide efficient and intelligent computing services and drive the intelligent transformation of various industries. Flow control technology, within intelligent computing scenarios, addresses potential network congestion and packet loss issues. It utilizes precise flow control via switches and coordinated flow control between routers and switches (achieving monitoring, adjustment, buffer scheduling, and fine-grained rate limiting at the data flow level) to ensure service performance, lossless data transmission, and stable network operation.
[0061] It should be noted that the beneficial effects or technical problems solved by the embodiments of this application are not limited to this one, but may also be other implicit or related problems. For details, please refer to the description of the embodiments below.
[0062] Before introducing the specific embodiments of this example, the technical terms involved in this example will be explained:
[0063] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.
[0064] In one embodiment, such as Figure 2 As shown, a flow control method is provided, which is applied to... Figure 1 Taking network devices as an example, the explanation includes the following steps:
[0065] Step 202: If congestion is detected, determine the data stream that is congested and the congestion information.
[0066] The congestion information includes one or more of the following: sender, receiver, time, data link, and delay. The congestion information is related to the data stream experiencing congestion. The sender in the congestion information represents the sending device of the congested data stream, for example, it could be the identifier of the sending device. The receiver in the congestion information could be the identifier of the receiving device. The time could be the current time or the time when congestion was detected, etc. The data link could be the identifier of the congested data link, for example, the data link between relay device 1 and relay device 2 might be congested, and the corresponding data link in the congestion information represents the identifier of the data link between relay device 1 and relay device 2. The delay could be the transmission delay of the congested data stream. The congested data stream can be a long-distance data stream, i.e., a data stream whose length exceeds a preset length threshold. This disclosure does not limit the specific value of the preset length threshold; those skilled in the art can determine it based on the actual application scenario.
[0067] Specifically, the application scenario of the flow control method provided in this embodiment can be a communication system. In this communication system, the transmitting device is connected to a relay network, and the relay network is connected to the receiving device. The relay network may include one or more interconnected relay devices / relay nodes / communication devices. The communication network also includes network devices. The network devices monitor whether congestion occurs in the communication system in real time, for example, by monitoring whether congestion occurs in each device or each data link in the communication system. Optionally, the network devices can determine whether the current communication system is congested by monitoring the network status of the communication system. If the network devices detect congestion, they can identify the congested / congested data stream and determine congestion information based on the congested / congested data stream.
[0068] Step 204: The congestion information is transmitted hop-by-hop in reverse so that the first-hop device corresponding to the sending device receives the congestion information and adjusts the transmission rate based on the congestion information.
[0069] In this context, "hop-by-hop reverse transmission" means that the congestion information is transmitted hop-by-hop in reverse from the location where the congestion occurs to the first hop device of the sending device. Along the transmission path of the data flow, the direction of data transmission from the sending device to the receiving device is forward, and correspondingly, the direction of data transmission from the receiving device to the sending device is reverse.
[0070] Specifically, after identifying a congested data stream, a network device can transmit the congestion information hop-by-hop in the reverse direction of the data stream's transmission, starting from the location of the congestion, until it reaches the first-hop device, which is the closest device to the sending device in the data stream. Upon receiving the congestion information, the first-hop device can adjust the transmission rate of the congested data stream based on this information. Once the transmission rate of the data stream is adjusted, the congestion situation will be alleviated.
[0071] In the aforementioned flow control method, network devices perform real-time congestion detection. Upon detection, they identify the congested data stream and its corresponding congestion information, and transmit this information hop-by-hop back to the first-hop device that sent the data stream. The first-hop device can adjust the data stream's transmission rate promptly and accurately based on the received congestion information and the remote congestion control backtracking flow control algorithm. This enables cooperation between routers and switches, allowing for independent monitoring and dynamic adjustment of each data stream in the communication network. By adjusting the data stream's transmission rate in real-time, congestion can be further avoided, ensuring timely detection and prompt handling to prevent congestion spread and guaranteeing network stability and reliability.
[0072] In one embodiment, the method further includes:
[0073] Detect network status parameters to determine if congestion has occurred.
[0074] Among them, network status parameters include one or more of link utilization and buffer occupancy.
[0075] Specifically, network devices can monitor in real time whether congestion occurs in various devices or data links within a communication system. For example, network devices can monitor the network status parameters of various communication devices in the communication system, such as link utilization and buffer occupancy. If a network status parameter is determined to meet the conditions for congestion, then the communication device is confirmed to be congested.
[0076] Optionally, the network device monitors the network status parameters of each data link in the communication system, that is, it monitors one or more of the link utilization and buffer occupancy rates of each data link in the communication system. If it is determined that the network status parameter meets the conditions for congestion, then it is determined that the communication device is currently experiencing congestion.
[0077] The conditions for congestion may be that the link utilization rate is greater than or equal to a preset utilization rate threshold, and / or that the buffer occupancy rate is greater than or equal to a preset occupancy rate threshold. The specific values of the preset utilization rate threshold and the preset occupancy rate threshold can be determined by those skilled in the art based on the actual application scenario, and this disclosure does not limit them.
[0078] In this embodiment, indicators that directly reflect the degree of congestion, such as link utilization and buffer occupancy, are used to determine whether congestion has occurred. This allows for accurate and rapid identification of the location of congestion in the communication system. Combining multiple indicators ensures a comprehensive assessment of whether congestion has occurred.
[0079] In one embodiment, the method further includes:
[0080] The network status parameters of the transmission link are monitored to determine if congestion has occurred. Alternatively, the devices in the transmission path are monitored to determine if congestion has occurred.
[0081] In this context, the transmission link can be the connection link between various devices in the communication system, and the devices in the transmission path can be various devices in the communication system.
[0082] Specifically, network devices can monitor in real time whether congestion occurs in individual devices or data links within a communication system. For example, a network device monitors the network status parameters of each communication device in the system; if it determines that a network status parameter meets the conditions for congestion, it determines that the communication device is currently congested. Optionally, a network device monitors the network status parameters of each data link in the communication system; if it determines that a network status parameter meets the conditions for congestion, it determines that the communication device is currently congested.
[0083] Optionally, the system can detect port queue backlog information and port backpressure of the transmission link, and determine whether congestion has occurred based on the monitored data.
[0084] In this embodiment, it is possible to accurately monitor and promptly detect whether congestion occurs at various locations in the communication system, providing a reliable data foundation for timely handling of congestion and preventing its spread.
[0085] In one embodiment, the specific implementation process of the step "determine the data stream experiencing congestion and the congestion information" may include:
[0086] The congested data stream and congestion information are then determined based on the 5-tuple information.
[0087] The 5-tuple information can be the source IP address, source port, destination IP address, destination port, and protocol type. A congested data stream can be any data stream experiencing congestion.
[0088] Specifically, after detecting congestion, network devices can determine the data path to which the congestion occurs based on the source IP address, source port, destination IP address, destination port, and protocol type of the location where the congestion occurs, i.e., determine the data flow experiencing congestion, and determine the congestion information. This congestion information is related to the data flow experiencing congestion, and the specific content of the congestion information has been described in detail in the above embodiments, and will not be repeated here.
[0089] In this embodiment, the congested data stream can be accurately located based on the 5-tuple information, which can ensure the timeliness of congestion detection and the timely handling of congestion, thus ensuring the stability and reliability of the communication system.
[0090] In one embodiment, the specific implementation process of the step "transmitting the congestion information hop-by-hop in reverse so that the first-hop device corresponding to the sending device receives the congestion information" may include:
[0091] Based on the data stream transmission path, congestion information is transmitted in reverse from the congested device to the first-hop device of the data stream sending device.
[0092] Specifically, the data stream is a congested data stream; the transmission path of this data stream can be a path that transmits data from the sending device to the receiving device. Reverse transmission is the opposite direction of data transmission of this data stream. After determining the congested data stream and the congestion information, the network device can transmit the congestion information hop-by-hop back to the sending device from the location of the congestion, in the opposite direction of the data transmission direction, until it reaches the first-hop device of the sending device. In this way, the first-hop device can receive the congestion information and adjust the transmission rate of the congested data stream based on the congestion information.
[0093] In this embodiment, congestion information is transmitted hop-by-hop back to the first-hop device in a timely manner. When the data stream or data link is a long-distance link, the congestion is transferred to the first-hop device of the long network to achieve lossless network. This can ensure timely feedback of congestion information, avoid relying on the feedback delay of remote congestion points, and further improve the timeliness and effectiveness of congestion handling.
[0094] In one embodiment, the first hop device is the device closest to the sending device.
[0095] Specifically, in a congested data stream, the first-hop device is the device closest to the sending device of the data stream, that is, the device directly connected to the sending device.
[0096] In one embodiment, the first-hop device is a switch or a router.
[0097] Specifically, the first-hop device can be a switch, such as a source Leaf switch, and the router can be a router determined based on the actual application scenario. This disclosure does not limit the type of router.
[0098] In one embodiment, such as Figure 3 As shown, a flow control method is provided. Taking the application of this method in a communication device as an example, the method includes the following steps:
[0099] Step 302: Receive congestion information transmitted by the network device.
[0100] The congestion information includes one or more of the following: sender, receiver, time, data link, and delay. Congestion information is transmitted hop-by-hop in reverse order when network devices detect congestion. The congestion information is related to the data stream experiencing congestion; the sender in the congestion information identifies the sending device of the congested data stream, for example, it could be the identifier of the sending device; the receiver in the congestion information can be the identifier of the receiving device; the time can be the current time or the time when congestion was detected, etc.; the data link can be the identifier of the congested data link, for example, the data link between relay device 1 and relay device 2 is congested, and the corresponding data link in the congestion information identifies the identifier of the data link between relay device 1 and relay device 2; the delay can be the transmission delay of the congested data stream. The congested data stream can be a long-distance data stream, that is, a data stream whose length exceeds a preset length threshold. This disclosure does not limit the specific value of the preset length threshold; those skilled in the art can determine it based on the actual application scenario.
[0101] Specifically, the application scenario of the flow control method provided in this embodiment can be a communication system. In this communication system, the transmitting device is connected to a relay network, and the relay network is connected to the receiving device. The relay network may include one or more interconnected relay devices / relay nodes / communication devices. The communication network also includes network devices. The network devices monitor whether congestion occurs in the communication system in real time, for example, by monitoring whether congestion occurs in each device or each data link in the communication system. Optionally, the network devices can determine whether the current communication system is congested by monitoring the network status of the communication system. If the network devices detect congestion, they can identify the congested / congested data stream and determine congestion information based on the congested / congested data stream.
[0102] After determining that a data stream is congested, the network device can transmit the congestion information hop-by-hop in the opposite direction of the data stream's transmission direction until it reaches the first-hop device of the sending device (i.e., the communication device used in this embodiment).
[0103] Step 304: Adjust the sending rate based on congestion information.
[0104] Specifically, the communication device is the first-hop device, which is the device closest to the sending device in the data stream. After receiving the congestion information, the first-hop device can adjust the transmission rate of the congested data stream based on the congestion information. When the transmission rate of the data stream is adjusted, the congestion of the data stream will be alleviated.
[0105] In this embodiment, the network device performs real-time congestion detection. When congestion is detected, it identifies the congested data stream and its corresponding congestion information, and transmits this congestion information hop-by-hop back to the first-hop device of the data stream's sender. The first-hop device can adjust the data stream's transmission rate in a timely and precise manner based on the received congestion information and the remote congestion control backtracking flow control algorithm. This enables cooperation between the router and the switch, allowing for independent monitoring and dynamic adjustment of each data stream in the communication network. By adjusting the data stream's transmission rate in real time, congestion can be further avoided, ensuring timely detection and processing after detection to prevent congestion from spreading and guaranteeing the stability and reliability of the network operation.
[0106] In one embodiment, the congestion information includes identification information of the data stream experiencing congestion. Specifically, the identification information of the data stream may be the name of the data stream, etc.; correspondingly, the specific implementation process of the step "adjusting the transmission rate based on the congestion information" may include:
[0107] Based on the data table and the identification information in the congestion information, the rate limiting ratio of the data stream is determined, and the sending rate of the data stream is adjusted based on the rate limiting ratio.
[0108] The data table is generated based on data from the communication system over a historical period. This data table can be a historical congestion information table for data streams. The identification information in the congestion information represents the identification information of the data path where congestion occurred; for example, it can be the name information of the data stream where congestion occurred.
[0109] Specifically, the first-hop device can query a data table based on the identification information in the congestion information, and obtain the rate-limiting ratio for the congested data stream based on the query results. Thus, the first-hop device can adjust the transmission rate of the congested data stream based on the obtained rate-limiting ratio. Optionally, after obtaining the rate-limiting ratio, the first-hop device can adjust the transmission rate of the congested data stream through a flow control mechanism. When the transmission rate of the data stream is adjusted to the transmission rate corresponding to the rate-limiting ratio, the congestion of the data stream can be alleviated.
[0110] In one example, the flow control mechanism could be Explicit Congestion Notification (ECN) or Quantified Congestion Notification (QCN), etc.
[0111] In this embodiment, intelligent decision-making can be performed through data tables, which is suitable for complex network environments and can adapt to changes in traffic patterns. This can reduce network congestion and packet loss, and improve the robustness and stability of the network.
[0112] In one embodiment, the data table includes at least the identification information, rate-limiting ratio, and congestion count for each data stream. The identification information, rate-limiting ratio, and congestion count for each data stream included in the data table are statistically derived from data collected over a historical time period.
[0113] The specific implementation process of the step "determine the rate limiting ratio of the data stream based on the identification information in the data table and congestion information" may include:
[0114] If identification information is found in the data table, and the number of congestion events corresponding to the identification information is greater than the first target value, then the rate limiting ratio corresponding to the identification information in the data table will be used as the rate limiting ratio of the data stream.
[0115] Specifically, the identification information of the congested data stream is retrieved from the data table, indicating that the data table stores information related to the adjustment of the transmission rate for that data stream. The first-hop device queries the data table for the identification information of the congested data stream. If the identification information is found in the data table, the congestion count corresponding to that identification information is obtained. If the congestion count is greater than a first target value, the first-hop device can determine the rate limiting ratio corresponding to that identification information in the data table as the current rate limiting ratio for that data stream. The first target value can be 0.
[0116] In this embodiment, the data table can record congestion-related information of the data stream in multiple dimensions, providing a reliable data foundation for subsequent intelligent decision-making on the transmission rate; when the number of congestion events exceeds the first target value, the previous rate limiting ratio is used, which can achieve efficient and accurate adjustment of the transmission rate.
[0117] In one embodiment, the data table includes at least the identification information, rate-limiting ratio, and congestion count for each data stream. The identification information, rate-limiting ratio, and congestion count for each data stream included in the data table are statistically derived from data collected over a historical time period.
[0118] The specific implementation process of the step "determine the rate limiting ratio of the data stream based on the identification information in the data table and congestion information" may include:
[0119] If the identification information is not found in the data table, or if the number of congestion events corresponding to the identification information is equal to the first target value, the rate limiting ratio of the data flow is determined based on the real-time network status.
[0120] Specifically, the identification information of the congested data stream is retrieved from the data table, indicating that the data table stores information related to the adjustment of the transmission rate of that data stream. The first-hop device retrieves the identification information of the congested data stream from the data table. If the identification information is not found in the data table, the rate limiting ratio of the congested data stream is determined based on the real-time network status. Optionally, if the identification information is found in the data table, and the number of congestions corresponding to the identification information is equal to a first target value, the rate limiting ratio of the congested data stream is determined based on the real-time network status. The first target value can be 0. Real-time network status s t This can include the port queue depth q t Link utilization u t Back pressure signal b t One or more of these, where t can be the current time.
[0121] In this embodiment, the data table can record congestion-related information of the data stream in multiple dimensions, providing a reliable data foundation for subsequent intelligent decision-making on the transmission rate, and adjusting the transmission rate based on the real-time network status to ensure the adaptability of the transmission rate to the actual application scenario.
[0122] In one embodiment, the real-time network status includes at least port queue depth and link utilization. The specific implementation process of the step "determining the rate limiting ratio of data flows based on the real-time network status" includes:
[0123] The rate limiting ratio of the data flow is determined based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization.
[0124] Specifically, the second target value can be a pre-configured target value, such as 1. The port queue depth represents the queue length of the data stream, and the link utilization represents the link utilization of the data stream. The corresponding first-hop device can process the data based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization to obtain the rate limiting ratio corresponding to the congested data stream. The first-hop device can adjust the data transmission rate of the data stream based on the rate limiting ratio, and the data stream after adjusting the transmission rate can alleviate congestion.
[0125] In one embodiment, such as Figure 4 As shown, the specific implementation process of the step "determine the rate limiting ratio of the data flow based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization" may include:
[0126] Step 402: Calculate the remaining space ratio of the port queue depth based on the port queue depth and the maximum value of the port queue depth.
[0127] Step 404: Calculate the remaining space ratio of link utilization based on the link utilization rate and the maximum value of the link utilization rate.
[0128] Step 406: Calculate the sum of the remaining space ratio of port queue depth and the remaining space ratio of link utilization.
[0129] Step 408: Among the second target value and the sum value, the minimum value is taken as the rate limiting ratio of the data stream.
[0130] Specifically, the port queue depth can be the port queue depth q of the data stream at the current time t. t The maximum value q of the port queue depth maxThis could be the maximum port queue depth of the data stream within a historical time period. In this way, the first-hop device can calculate the difference between the maximum port queue depth and the current port queue depth, and then calculate the ratio (q) of this difference to the port queue depth. max -q t ) / q max The obtained ratio is determined as the proportion of remaining space for the port queue depth.
[0131] In one example, link utilization could be the link utilization u of the data stream at the current time t. t The maximum value of link utilization u max This could be the maximum link utilization of the data stream within a historical time period. In this way, the first-hop device can calculate the difference between the maximum link utilization and the current link utilization, and then calculate the ratio of this difference to the current link utilization (u). max -u t ) / u max The ratio obtained is determined as the remaining space ratio of the link utilization.
[0132] In this way, the first-hop device can calculate the sum of the remaining space ratio of the port queue depth and the remaining space ratio of the link utilization, and compare this sum with a second target value. The smaller value is determined as the rate limiting ratio for the data flow. That is, if the sum is greater than the second target value, the second target value is determined as the rate limiting ratio for the congested data flow; if the sum is less than the second target value, the sum is determined as the rate limiting ratio for the congested data flow.
[0133] In this embodiment, considering both queue depth and link utilization, the rate limiting ratio is dynamically adjusted to maximize transmission efficiency while ensuring network uncongestedness.
[0134] In one embodiment, the data table also includes the time of the most recent congestion corresponding to the identification information.
[0135] Specifically, the time of the most recent congestion corresponding to the identification information is the time of the most recent congestion of the data stream that experienced congestion.
[0136] In one embodiment, the communication device is the first-hop device of the data stream sending device.
[0137] In one embodiment, the first-hop device is the device closest to the sending device, and / or the first-hop device is a switch or a router.
[0138] Specifically, in a congested data stream, the first-hop device is the device closest to the sending device of the data stream, i.e., the device directly connected to the sending device. Optionally, the first-hop device can be a switch, such as a source Leaf switch, or a router determined based on the actual application scenario. This disclosure does not limit the type of router.
[0139] In one embodiment, the specific execution process of the above-described flow control method is described in conjunction with a detailed embodiment:
[0140] The flow control method provided in this embodiment can solve the problem of packet loss due to faults in communication systems. It primarily transfers congestion on long-distance links to the first-hop device in the network. Network devices determine congestion by checking port queue backpressure information and port backpressure. Non-first-hop devices notify the source Leaf switch (first-hop device) of the congestion information. The source Leaf switch runs an algorithm based on the degree of congestion to determine the rate-limiting ratio, and then implements rate control by sending flow control protocol messages. Secondly, this embodiment proposes a retrospective flow control algorithm for remote congestion control. This algorithm combines historical congestion information and real-time network status, dynamically adjusting the rate-limiting strategy to "transfer" congestion on long-distance links to the first-hop device in the network, thereby achieving lossless network operation. The core idea of the algorithm is to use a historical information table to record the congestion patterns of periodic traffic and combine this with real-time network status to make rate-limiting decisions, avoiding reliance on feedback delays from remote congestion points.
[0141] The precise flow control technology of the switch in this embodiment solves the problems of congestion, packet loss and service performance degradation caused by anomalies in intelligent computing service scenarios. By transferring long-distance link congestion to the source Leaf switch, the switch checks the network status, limits the rate according to the algorithm, and trains a traffic periodically maintenance information table based on a large model, so as to achieve lossless traffic in subsequent periods.
[0142] Compared to congestion within data centers, congestion across long-distance links has a significantly different mechanism: when congestion occurs in the local network, existing flow control mechanisms can quickly respond and suppress congestion spread due to the shorter feedback path, while remote congestion takes longer to process. By analyzing the differences in the impact of congestion location on network performance, the flow control method in this embodiment can "move forward" the control concept of remote congestion to the source device. This technical solution monitors the port queue status in real time, feeds back congestion information to the source Leaf switch, and combines the periodic characteristics of traffic trained by a large model to establish a historical congestion information table to achieve preventative rate control.
[0143] When congestion inevitably occurs on long-distance links, the concept of precise flow control by switches is to "migrate" this congestion to the network's originating device, i.e., the first-hop device. The specific operation process is as follows: network devices monitor the network status in real time, including the backlog of port queues and the relevant backpressure conditions on ports, to determine whether network congestion has occurred.
[0144] Once congestion is detected, and the device is not the originating device causing the congested traffic, it will pass congestion-related information to the first-hop device generating the congested traffic—the source Leaf switch. Next, the source Leaf switch will use a backtracking flow control algorithm to rate-limit the congested traffic based on the severity of the congestion. Finally, the source Leaf switch will effectively control the traffic rate by sending flow control protocol messages.
[0145] Specifically, the flow control method in this embodiment can achieve intelligent decision-making through a backtracking flow control algorithm. It can dynamically calculate the rate limiting ratio by comprehensively considering two key network state factors: port queue depth and link utilization. Compared with flow control algorithms that only consider a single factor, this multi-factor comprehensive decision-making approach can more comprehensively reflect the actual network load, make more reasonable rate limiting decisions, effectively balance network transmission efficiency and congestion control, and improve the overall network performance.
[0146] 1. Multi-dimensional congestion status recording, recording the congestion status of each link:
[0147]
[0148] Where f represents the flow identifier, cf represents the number of congestion events, rf represents the rate limiting ratio, and tf represents the time to the nearest congestion event. ct represents the congestion state at time t, and its Z can be a multi-dimensional vector containing information such as the number of congestion events nt, the congestion duration dt, the congestion severity st (e.g., packet loss rate, latency increase), and the congestion location lt.
[0149]
[0150] 2. Record real-time network status. t :
[0151]
[0152] Where, q t It is the port queue depth, u t It is the link utilization rate, b t It is a backpressure signal. When the buffer of a network device (such as a port queue) is close to full, the backpressure signal will notify the upstream device to reduce the sending rate to prevent buffer overflow and packet loss.
[0153] 3. Speed determination: Speed limit ratio a t According to historical information table and real-time network status t Calculated dynamically:
[0154]
[0155] When the flow identifier f exists in the historical information table H(f), and the corresponding congestion count c f Greater than 0, speed limit ratio a t Take the speed limit ratio r directly from the historical records. f This function indicates that if the link has a history of congestion, the rate limiting ratio set in the previous congestion response will be used to control the current traffic in order to avoid congestion again.
[0156] When the flow identifier f does not exist in the historical information table H(f), or although it is in the table, the congestion count c f When =0, the speed limit ratio is taken as 1 and (q max -q t ) / q max +(u max -u t ) / u max The minimum value of . Where, (q max -q t ) / q max This represents the proportion of remaining space at the port queue depth, (u max -u t ) / u max This reflects the remaining space ratio that indicates link utilization. The sum of these two values and the minimum value obtained by adding them together with 1 is a comprehensive consideration of queue depth and link utilization, dynamically adjusting the rate limiting ratio to maximize transmission efficiency while ensuring network congestion.
[0157] The specific implementation process can be as follows: First, determine congestion based on network status: Network devices monitor network status to determine if congestion has occurred. For example, this can be done by detecting metrics such as link utilization and buffer occupancy. Second, identify the specific traffic: When congestion is detected, the network device uses the five-tuple (source IP address, source port, destination IP address, destination port, and protocol type) to determine which specific traffic caused the congestion, i.e., to identify the data flow causing the congestion. Third, conduct hop-by-hop reverse notification: The network device notifies the first-hop device of the congested traffic information hop-by-hop. This means that congestion information is transmitted backward from the congested device along the data flow path to the sending device. Fourth, the first-hop device adjusts the traffic: The first-hop device (i.e., the switch or router closest to the sending end) adjusts the transmission rate of the corresponding traffic based on the received congestion level and traffic information. In this way, the first-hop device can dynamically adjust the transmission rate, thereby effectively alleviating network congestion.
[0158] The flow control method in this embodiment can achieve congestion advance and enable rapid congestion response:
[0159] (1) Fast response: After the congestion point is moved forward, the response time is shortened from more than 1 millisecond to 10 microseconds, which enables the network to react to congestion more quickly, reduces the duration of congestion, and avoids a large amount of data queuing in the link, thus ensuring the real-time performance of data transmission.
[0160] (2) Reduce Buffer Pressure: Remote congestion can cause a large amount of buffer accumulation, which can have a significant impact on cache resources and may lead to buffer overflow and packet loss. However, after the congestion point is moved forward, the amount of buffer accumulation is greatly reduced, which alleviates the burden on the cache, reduces the risk of packet loss, and ensures lossless data transmission.
[0161] (3) Active rate control: When the congestion point is near the switch, the switch can actively control the rate and adjust the transmission rate in a timely manner according to the degree of congestion and traffic information to alleviate the congestion. However, when the congestion is far away, the long response time makes it difficult to control the traffic quickly and effectively, which can easily lead to the aggravation of congestion.
[0162] This embodiment proposes two key technical points: First, precise flow control technology for switches. By monitoring network status in real time, it transfers congestion at the far end of long-distance links to the source Leaf switch. It combines a large-scale model to periodically maintain a historical congestion information table based on traffic periodicity and applies an algorithm for rate limiting, thus solving the problems of congestion, packet loss, and performance degradation caused by anomalies in intelligent computing service scenarios. Second, a retrospective flow control algorithm. This algorithm calculates the rate limiting ratio through multi-dimensional congestion status records and real-time dynamic monitoring of network status, offering advantages such as high precision, high flexibility, strong robustness, high stability, and better resource utilization. In other words, this embodiment proposes precise flow control technology for switches, which transfers congestion at the far end of long-distance links to the source Leaf switch by monitoring network status in real time. It combines a large-scale model to periodically maintain a historical congestion information table based on traffic periodicity and applies an algorithm for rate limiting, thereby solving the problems of congestion, packet loss, and performance degradation caused by anomalies in intelligent computing service scenarios. The backtracking flow control algorithm proposed in this embodiment can achieve higher accuracy and flexibility: through multi-dimensional congestion state recording and fine-grained network state monitoring, the algorithm can more accurately perceive and respond to changes in network state, thereby realizing a more refined flow control strategy; it can achieve stronger robustness and stability: the intelligent rate limiting decision mechanism and adaptive backpressure mechanism can cope with complex network environments and traffic pattern changes, reduce network congestion and packet loss, and improve network robustness and stability; it can achieve better resource utilization: cross-layer optimization and collaboration can make full use of network resources, improve bandwidth utilization and transmission efficiency, while reducing energy consumption and operating costs.
[0163] It should be understood that, although Figure 1-4 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1-4 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.
[0164] In one embodiment, such as Figure 5 As shown, a flow control device 500 is provided, applied to network equipment, comprising:
[0165] The detection module 502 is used to determine the data stream that is congested and the congestion information if congestion is detected. The congestion information includes one or more of the following: sender, receiver, time, data link, and delay.
[0166] The transmission module 504 is used to transmit congestion information hop-by-hop in reverse order so that the first-hop device corresponding to the sending device receives the congestion information and adjusts the transmission rate based on the congestion information.
[0167] In one embodiment, such as Figure 6 As shown, a flow control device 600 is provided, applied to a communication device, and includes:
[0168] The first receiving module 602 is used to receive congestion information transmitted by a network device. The congestion information includes one or more of the following: sending end, receiving end, time, data link, and delay. The congestion information is transmitted hop-by-hop in reverse by the network device when it detects congestion.
[0169] The first adjustment module 604 is used to adjust the transmission rate based on the congestion information.
[0170] For specific limitations regarding the flow control device, please refer to the limitations on the flow control method above, which will not be repeated here. Each module in the aforementioned flow control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in the computer device, or stored in software in the memory of the computer device, so that the processor can call and execute the operations corresponding to each module.
[0171] Figure 7 This is a schematic diagram of the structure of the access network device provided in an embodiment of the present invention. Figure 7 The access network device 700 shown includes at least one processor 701, a memory 702, and at least one network interface 704. The various components in the access network device 700 are coupled together via a bus system 705. It is understood that the bus system 705 is used to implement communication between these components. In addition to a data bus, the bus system 705 also includes a power bus, a control bus, and a status signal bus. However, for clarity, ... Figure 7 Various buses are designated as bus system 705. Additionally, this embodiment of the invention includes a transceiver 706, which may consist of multiple components, including a transmitter and a receiver, providing a unit for communicating with various other devices over a transmission medium.
[0172] It is understood that the memory 702 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 702 of the systems and methods described in this embodiment is intended to include, but is not limited to, these and any other suitable types of memory.
[0173] In some implementations, memory 702 stores executable modules or data structures, or subsets thereof, or extended sets thereof, such as operating system 7021. Operating system 7021 includes various system programs, such as a framework layer, core library layer, and driver layer, used to implement various basic business functions and handle hardware-based tasks.
[0174] In this embodiment of the invention, by calling the program or instructions stored in the memory 702, the processor is configured to determine the congested data stream and congestion information if congestion is detected. The congestion information includes one or more of the following: the sending end, the receiving end, time, data link, and delay. The transmitter is configured to transmit the congestion information hop-by-hop in reverse order so that the first hop device corresponding to the sending end device receives the congestion information and adjusts the transmission rate based on the congestion information.
[0175] The methods disclosed in the above embodiments of the present invention, in part or in all of them, can also be applied to processor 701, implemented by processor 701, or implemented by processor 701 in conjunction with other components (e.g., transceivers). Processor 701 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above methods can be completed by the integrated logic circuit of the hardware in processor 701 or by instructions in the form of software. The processor 701 described above may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present invention can be directly manifested as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 702, and processor 701 reads the information from memory 702 and, in conjunction with its hardware, completes the steps of the above method.
[0176] It is understood that the embodiments described in this invention can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described in this application, or combinations thereof.
[0177] For software implementation, the technology described in the embodiments of the present invention can be implemented by modules (e.g., procedures, functions, etc.) that perform the functions described in the embodiments of the present invention. The software code can be stored in memory and executed by processor 701. The memory can be implemented in processor 701 or external to processor 701.
[0178] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of this embodiment.
[0179] This application also provides a computer program product containing instructions that, when run on a computer, cause the computer to execute the steps in this embodiment.
[0180] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.
[0181] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0182] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A flow control method, characterized in that, Applied to network devices, the method includes: If congestion is detected, the data stream that is congested and the congestion information are determined, including one or more of the following: sender, receiver, time, data link, and delay. The congestion information is transmitted hop-by-hop in reverse so that the first-hop device corresponding to the sending device receives the congestion information and adjusts the transmission rate based on the congestion information.
2. The method according to claim 1, characterized in that, The method further includes: The network status parameters are detected to determine whether congestion has occurred. These network status parameters include one or more of link utilization and buffer occupancy.
3. The method according to claim 1, characterized in that, The method further includes: The network status parameters of the transmission link are checked to determine if congestion has occurred; or, The devices in the transmission path are inspected to determine if congestion has occurred.
4. The method according to claim 1, characterized in that, The process of determining the congested data stream and congestion information includes: The congested data stream and congestion information are then determined based on the 5-tuple information.
5. The method according to claim 1, characterized in that, The step of transmitting the congestion information hop-by-hop in reverse so that the first-hop device corresponding to the sending device receives the congestion information includes: Based on the transmission path of the data stream, the congestion information is transmitted in reverse from the congested device until it reaches the first hop device of the data stream's sending device.
6. The method according to claim 3, characterized in that, The first hop device is the device closest to the transmitting device.
7. The method according to claim 3, characterized in that, The first hop device is a switch or a router.
8. A flow control method, characterized in that, Applied to communication equipment, the method includes: The network device receives congestion information transmitted from a network device. The congestion information includes one or more of the following: sender, receiver, time, data link, and delay. The congestion information is transmitted hop-by-hop in reverse order when the network device detects congestion. Based on the congestion information, the transmission rate is adjusted.
9. The method according to claim 8, characterized in that, The congestion information includes identification information of the data stream experiencing congestion; adjusting the transmission rate based on the congestion information includes: Based on the data table and the identification information in the congestion information, the rate limiting ratio of the data stream is determined, and the transmission rate of the data stream is adjusted based on the rate limiting ratio.
10. The method according to claim 9, characterized in that, The data table includes at least the identification information, rate limiting ratio, and congestion count for each data stream; determining the rate limiting ratio of the data stream based on the data table and the identification information in the congestion information includes: If the identification information is found in the data table, and the number of congestion events corresponding to the identification information is greater than the first target value, then the rate limiting ratio corresponding to the identification information in the data table is used as the rate limiting ratio of the data stream.
11. The method according to claim 9, characterized in that, The data table includes at least the identification information, rate limiting ratio, and congestion count for each data stream; determining the rate limiting ratio of the data stream based on the data table and the identification information in the congestion information includes: If the identification information is not found in the data table, or if the number of congestion events corresponding to the identification information is equal to the first target value, then the rate limiting ratio of the data stream is determined based on the real-time network status.
12. The method according to claim 11, characterized in that, The real-time network status includes at least port queue depth and link utilization; determining the rate limiting ratio of the data flow based on the real-time network status includes: The rate limiting ratio of the data stream is determined based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization.
13. The method according to claim 12, characterized in that, The determination of the data flow rate limiting ratio based on the second target value, the remaining space ratio of the port queue depth, and the remaining space ratio of the link utilization includes: Calculate the remaining space ratio of the port queue depth based on the port queue depth and the maximum value of the port queue depth; Based on the link utilization rate and the maximum link utilization rate, calculate the remaining space ratio of the link utilization rate. Calculate the sum of the remaining space ratio of the port queue depth and the remaining space ratio of the link utilization; The minimum value between the second target value and the sum is determined as the rate limiting ratio of the data stream.
14. The method according to claim 10, characterized in that, The data table also includes the time of the most recent congestion corresponding to the identification information.
15. The method according to claim 8, characterized in that, The communication device is the first-hop device of the data stream sending device.
16. The method according to claim 15, characterized in that, The first hop device is the device closest to the sending device, and / or the first hop device is a switch or a router.
17. A flow control device, characterized in that, Applied to network devices, the device includes: The detection module is used to determine the data stream experiencing congestion and congestion information if congestion is detected. The congestion information includes one or more of the following: sender, receiver, time, data link, and latency. The transmission module is used to transmit the congestion information hop-by-hop in reverse order so that the first hop device corresponding to the sending device receives the congestion information and adjusts the transmission rate based on the congestion information.
18. A communication device, characterized in that, include: Transmitter, processor, and receiver, wherein: The processor is configured to determine, if congestion is detected, the congested data stream and congestion information, the congestion information including one or more of the following: sender, receiver, time, data link, and delay; The transmitter is used to transmit the congestion information hop-by-hop in reverse order so that the first-hop device corresponding to the transmitting device receives the congestion information and adjusts the transmission rate based on the congestion information.
19. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 16.
20. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 16.
