System for enhancing long-distance network transmission quality, method and computer readable medium thereof

Abstract

The present invention is a system for enhancing long-distance network transmission quality and method thereof. A service status management subsystem collects user traffic information of managed services and cloud platform information of each connected cloud platform, so as to generate a correction request when the user traffic information does not meet a regulatory target. After receiving the correction request and cloud platform information from the service status management subsystem, an acceleration network card management subsystem analyzes traffic packets to execute a performance control under abnormal service performance, and / or execute an abnormal transmission correction when traffic packets are dropped. Accordingly, the present invention enables each service to effectively and stably exert its performance. The present invention also provides a computer-readable medium for executing the method of the present invention.

Description

Technical Field

[0001] This invention relates to network transmission technology, and more particularly to a long-distance network transmission quality optimization system, method, and computer-readable medium thereof. Prior Technology

[0002] With the evolution of high-speed network transmission technology, cloud services can exchange information and execute tasks more efficiently, giving rise to various real-time edge cloud services. However, physical delays caused by distance can easily lead to congestion, packet loss, and retransmissions in high-speed transmission services, affecting other stable transmission services on the same cloud platform. This results in a significant decrease in network transmission performance and data transmission anomalies for all cloud services. Current technologies primarily focus on high-speed transmission equipment, such as Remote Direct Memory Access (RDMA) network cards developed to address server-side data processing latency in network transmission. However, these technologies aim to effectively utilize these high-speed transmission devices rather than focusing on the operational performance and stability of the services they support. This often results in services failing to fully realize their potential performance and value.

[0003] Therefore, finding a technology to improve network transmission quality, especially addressing issues such as congestion, packet loss, or retransmission that can easily occur during long-distance transmission, has become a pressing goal for researchers in this field. Summary of the Invention

[0004] To achieve the aforementioned objectives, this invention proposes a long-distance network transmission quality optimization system, comprising: a service status management subsystem, used to collect user traffic information and information from each connected cloud platform under management, to generate a correction request when the user traffic information is compared with pre-stored monitoring targets and does not meet a preset threshold, and to generate cloud platform messages from the information of each cloud platform for performance correction; and an accelerated network interface card management subsystem, connected to the service status management subsystem, used to receive the correction request and cloud platform messages from the service status management subsystem, to analyze traffic packets, and to perform performance control accordingly when a service performance anomaly event is received, and / or to perform abnormal transmission correction accordingly when a packet loss occurs.

[0005] In one embodiment, the service status management subsystem is located on an edge cloud platform, while the accelerated network interface card (NIC) management subsystem is located on an accelerated NIC device.

[0006] In one embodiment, the service status management subsystem includes: a service registration module, which provides a user interface for receiving service information, monitoring indicators, and correction methods input by the user and performing format checks; a service association database, which records data after format checks by the service registration module; a service supervision module, which periodically collects the user's service information and compares it with the supervision targets recorded in the service association database, so as to provide the user's service information for the accelerated network card management subsystem to correct if it does not meet the threshold; and a cloud information synchronization module, which periodically collects information from the cloud platform as a basis for the accelerated network card management subsystem to perform service performance correction.

[0007] In one embodiment, the accelerated network interface card (NIC) management subsystem includes: a target receiving module for receiving basic service information from the service status management subsystem and performing format checks; a NIC association database for recording information after format checks by the target receiving module; a packet query module for continuously monitoring packet loss events, obtaining packet loss service information from the service status management subsystem when packet loss occurs, synchronizing the NIC association database and generating correction messages, and performing packet capture using user information from the NIC association database to generate analysis messages when a service performance anomaly event is received from the target receiving module; and a traffic management module for receiving the correction messages and analysis messages from the packet query module, performing performance control to find available relay points when the transmission situation is congestion and round-trip time (RTT) caused by transmission anomalies, and performing abnormal transmission correction for the traffic when the transmission situation is traffic packet loss.

[0008] In one embodiment, the performance control system includes the traffic management module checking the relay stations recorded in the network interface card (NIC) associated database to find the relay station with an available resource percentage greater than the target traffic and the shortest round-trip communication time. The system then notifies the cloud platform to perform scheduling, uses software-defined networking (Software-Defined Networking) technology to redirect packets from the source to the target cloud platform through the relay station, sets up a listening configuration within the relay station to automatically reply with acknowledgment messages (ACK) and handle retransmissions, writes the relay station information into the NIC associated database, and removes the relay station when the traffic does not exist.

[0009] In one embodiment, the abnormal transmission correction system includes the traffic management module querying the network card associated database through the user information. If the adjustment count in the user information is greater than zero, the adjustment count is reduced by one, and a hardware-supported congestion method is used to obtain a lower speed reduction value than the current value for the traffic. In the event of packet loss or congestion, the adjustment count reduction and the congestion method are continuously executed until the packet loss or congestion is eliminated.

[0010] This invention further discloses a method for optimizing long-distance network transmission quality, which is executed on a computer or server. The method includes the following steps: A service status management subsystem collects user traffic information and information from each connected cloud platform under its management; when the user traffic information is compared with pre-stored monitoring targets and does not meet a pre-set threshold, a correction request is generated; and the information from each cloud platform is used to generate cloud platform messages for performance correction. An accelerated network interface card (NIC) management subsystem receives the correction request and cloud platform messages from the service status management subsystem to analyze traffic packets, so that when a service performance anomaly event is received, performance control is performed accordingly, and / or when packet loss occurs, abnormal transmission correction is performed accordingly.

[0011] In the above method, the service status management subsystem is set on the edge cloud platform, while the accelerated network card management subsystem is set on the accelerated network card device.

[0012] In the above method, the service status management subsystem performs the following: The service registration module provides a user interface to receive service information, monitoring indicators, and correction methods input by the user and performs format checks; the service-related database records the data after format checks by the service registration module; the service monitoring module periodically collects the user's communication information and compares it with the monitoring target recorded in the service-related database, so that if the threshold is not met, the user's communication information is provided to the accelerated network card management subsystem for correction; and the cloud information synchronization module periodically collects cloud platform information to use as the basis for the accelerated network card management subsystem to perform communication performance correction.

[0013] In the above method, the accelerated network interface card (NIC) management subsystem performs the following: instructs the target receiving module to receive basic service information from the service status management subsystem and perform format checks; instructs the NIC association database to record the information after format checks by the target receiving module; instructs the packet query module to continuously monitor packet loss events, so as to obtain packet loss service information from the service status management subsystem when packet loss occurs, so as to synchronize the NIC association database and generate correction messages; and when receiving a service performance anomaly event transmitted by the target receiving module, performs packet capture using user information in the NIC association database to generate analysis messages; and instructs the traffic management module to receive the correction messages and analysis messages from the packet query module, and by analyzing the transmission status of the traffic packets, performs performance control to find available relay points when the transmission status is congestion and round-trip time (RTT) caused by transmission anomalies, and performs abnormal transmission correction for the traffic when the transmission status is traffic packet loss.

[0014] In the above method, the performance control includes the traffic management module checking the relay stations recorded in the network interface card (NIC) associated database to find the relay station with an available resource percentage greater than the target traffic and the shortest round-trip communication time. The cloud platform is then notified to perform scheduling, and the packets from the source and target cloud platforms are redirected to pass through the relay station using software-defined networking (Software-Defined Networking) technology. The relay station is configured to listen for and automatically reply with acknowledgment messages (ACK) and handle retransmissions. The relay station information is written to the NIC associated database, and the relay station is removed when the traffic does not exist.

[0015] In the above method, the abnormal transmission correction system includes the traffic management module querying the network card associated database through the user information. If the adjustment count in the user information is greater than zero, the adjustment count is reduced by one, and a lower speed reduction value is obtained by using the hardware-supported congestion method to configure the traffic. In the event of packet loss or congestion, the adjustment count reduction and the congestion method are continuously executed until the packet loss or congestion is eliminated.

[0016] The present invention further discloses a computer-readable medium, which is used in a computing device or computer and stores instructions to execute the aforementioned long-distance network transmission quality optimization method.

[0017] In summary, the long-distance network transmission quality optimization system, method, and computer-readable medium of the present invention manage and collect the status and requirements of each service through a cloud platform, enabling the exchange of platform status and latency information between cloud platforms. This allows for the evaluation of suitable transmission schemes for services, configuration of data transmission conditions based on the schemes, and network management configuration of smart network cards or data processing units, thereby enabling each service to effectively and stably perform its performance and value. Simple Explanation of the Diagram

[0018] Figure 1 is a system architecture diagram of the long-distance network transmission quality optimization system of the present invention.

[0019] Figure 2 is a system architecture diagram of a specific embodiment of the long-distance network transmission quality optimization system of the present invention.

[0020] Figure 3 is a flowchart illustrating the steps of the long-distance network transmission quality optimization method of the present invention.

[0021] Figure 4 is a flowchart of the service status management subsystem of this invention, which handles user registration and monitoring services and synchronizes cloud information.

[0022] Figure 5 is a flowchart of the process by which the accelerated network card management subsystem receives data provided by the service status management subsystem in accordance with the present invention.

[0023] Figure 6 is a flowchart of the packet loss handling process of the accelerated network card management subsystem of the present invention.

[0024] Figure 7 is a flowchart of the performance control of the communication management module of the present invention.

[0025] Figure 8 is a flowchart of the abnormal transmission correction performed by the communication management module of the present invention.

[0026] Figure 9 shows the data format used in the service-related database of this invention.

[0027] Figure 10 shows the data format used by the user to correct errors in the network card association database of this invention.

[0028] Figure 11 shows the data format used by the network card association database of the present invention for the accelerated network card. Implementation

[0029] The technical content of this invention is described below through specific embodiments. Those skilled in the art can easily understand the advantages and effects of this invention from the content disclosed in this specification. However, this invention can also be implemented or applied through other different embodiments.

[0030] Figure 1 is a system architecture diagram of the long-distance network transmission quality optimization system of the present invention. The present invention proposes a technology and method for stable high-speed transmission of edge cloud services. Its main purpose is to improve the problem of significant degradation in overall transmission network performance and intermittent service caused by unstable transmission environments in high-speed transmission scenarios for edge cloud services, thereby achieving the goal of effectively and stably providing cloud services to users. As shown in the figure, the long-distance network transmission quality optimization system 1 includes a service status management subsystem 11 and an accelerated network interface card (NIC) management subsystem 12.

[0031] The service status management subsystem 11 collects user traffic information and information from connected cloud platforms for managed services. It generates correction requests when user traffic information is compared with pre-stored monitoring targets and does not meet pre-set thresholds. It also generates cloud platform messages from the information of each cloud platform for performance correction. In one embodiment, the service status management subsystem 11 is implemented on the edge cloud platform 2, primarily to monitor services and synchronize information, performing registration, management, and evaluation of user traffic status. When a monitoring target becomes abnormal, the service status management subsystem 11 notifies the accelerated network interface card management subsystem 12 for evaluation and correction. Furthermore, the service status management subsystem 11 can synchronize information between different cloud platforms, enabling subsequent performance correction or its mechanisms to evaluate feasible adjustment plans.

[0032] The accelerated network interface card (NIC) management subsystem 12 is connected to the service status management subsystem 11. The accelerated NIC management subsystem 12 receives correction requests and cloud platform messages from the service status management subsystem 11 to perform packet analysis. Accordingly, upon receiving a notification of a service performance anomaly, it performs corresponding performance control or its mechanism, and / or performs abnormal transmission correction or its mechanism when packet loss occurs. In one embodiment, the accelerated NIC management subsystem 12 resides within the accelerated NIC device 3. Each edge cloud platform 2 can have multiple accelerated NIC devices 3. Messages from each accelerated NIC management subsystem 12 are independent, while service and user information originate from the service status management subsystem 11 of their respective cloud platforms. Furthermore, the primary objective of the accelerated NIC management subsystem 12 is to handle packet loss and performance tuning, adjusting on a user-by-user basis. Unlike current technologies that operate on a service-by-service basis, this invention enables the cloud platform to more accurately and flexibly manage its traffic (i.e., the data flow of users on the service).

[0033] In addition, the accelerated network card management subsystem 12 is a processor that uses the accelerated network card device, which enables the cloud platform to support more services and users. Therefore, the present invention can achieve better service scalability and improve the performance bottleneck caused by physical distance, thereby effectively and stably providing cloud services to users.

[0034] In summary, the service status management subsystem 11 and the accelerated network interface card (NIC) management subsystem 12 of the long-distance network transmission quality optimization system 1 operate on the edge cloud platform 2 and their respective accelerated NIC devices 3, respectively. The main function of the service status management subsystem 11 is to allow service holders to indicate resource requirements, regulatory requirements, user information, and adjust traffic policy information, and to periodically check the traffic status of each service user. If there is an anomaly, it will execute the corresponding traffic correction action and simultaneously notify the accelerated NIC management subsystem 12. In addition, the accelerated NIC management subsystem 12 mainly receives the correction information from the service status management subsystem 11 to improve traffic performance and detect packet loss notifications, so that abnormal traffic of users does not affect the overall transmission network performance.

[0035] Figure 2 is a system architecture diagram of a specific embodiment of the long-distance network transmission quality optimization system of the present invention. As shown in the figure, the service status management subsystem 11 and the accelerated network card management subsystem 12 of the long-distance network transmission quality optimization system 1 are the same as those shown in Figure 1. In this embodiment, the internal architecture of the service status management subsystem 11 and the accelerated network card management subsystem 12 is further explained.

[0036] As shown in Figure 2, the main function of the service status management subsystem 11 is to monitor the communication of service users and transmit information between users and the cloud platform. The service status management subsystem 11 includes a service registration module 111, a service association database 112, a service monitoring module 113, and a cloud information synchronization module 114. The main function of the accelerated network card management subsystem 12 is to perform corrective actions for performance anomaly notifications and packet loss events. The accelerated network card management subsystem 12 includes a target receiving module 121, a network card association database 122, a packet query module 123, and a communication management module 124.

[0037] The service registration module 111 provides a user interface for receiving service information, monitoring indicators, and correction methods input by users, and performs format checks.

[0038] The service-related database 112 is used to record the data after it has been formatted and checked by the service registration module 111. That is, it records the configuration data received by the service registration module 111. The configuration data may include the service identifier (ID), resource usage, user identifier column, target and threshold, number of adjustments and abnormal limit policy, but is not limited to these.

[0039] Further, in one embodiment, as shown in FIG9, the user data format of the service association database 112, i.e., the basic data format, is illustrated. Here, the service ID 901 is the identifier given to the managed service by the cloud platform; resource usage 902 is the hardware requirement of the service; and the user identifier column 903 is the identifier of the user in this service, which is a key-value array, such as [{"key":"value"},...], where the key is the ID given to the user by the service, and the value is a field value that can be used for packet identification, such as VLAN or Differentiated Services Code (DSC). Points, DSCPs, etc.; Target and Threshold 904 are the service's regulatory metrics and thresholds, which are a key-value array, such as [{"key":"value"},...], where key is the metric and value is the threshold; Adjustment Count 905 is the upper limit number of times the service can be corrected or adjusted by its mechanism; Abnormal Limit Policy 906 indicates that after reaching the adjustment limit number, packets should be dropped or a specified rate should be configured as much as possible (the value can be greater than or equal to 0).

[0040] The service monitoring module 113 periodically collects user traffic information and compares it with the monitoring targets recorded in the service association database 112. If any user traffic information does not meet the threshold, the module provides the accelerated network card management subsystem 12 with the information for correction. In one embodiment, the service monitoring module 113 is responsible for periodically collecting user traffic information of the managed services and comparing it with the monitoring targets recorded in the service association database 112. If any user traffic information does not meet the threshold, the module transmits the relevant information to the accelerated network card management subsystem 12 for correction.

[0041] The cloud information synchronization module 114 is used to periodically collect information from cloud platforms for use in accelerating network interface card management subsystem 12 to perform traffic performance correction. In one embodiment, the cloud information synchronization module 114 is responsible for periodically collecting information from each connected cloud platform, including source identifier (ID), destination identifier (ID), latency time, and percentage of available resources. This information is provided to the accelerated network interface card management subsystem 12 for use as traffic performance correction parameters.

[0042] The target receiving module 121 is used to receive basic information about the service from the service status management subsystem 11 and perform format checks. In one embodiment, the main function of the target receiving module 121 is to receive basic information about the service from the service status management subsystem 11 (including service identification code (ID), resource usage, adjustment times and abnormal limit policy), source IP, destination IP and user tag, and write it into the network card association database 122 after checking the format.

[0043] The network interface card (NIC) association database 122 is used to record data after the target receiving module 121 has checked the format. In one embodiment, the NIC association database 122 records data provided by the target receiving module 121, including service identifier (ID), source IP, destination IP, user identifier, and number of adjustments.

[0044] The packet query module 123 is used to continuously monitor packet loss events. When a packet loss occurs, it obtains packet loss service information from the service status management subsystem 11 to synchronize the network card association database 122 and generate correction messages. When it receives a service performance abnormality event transmitted by the target receiving module 121, it performs packet capture with the user information in the network card association database 122 to generate analysis messages.

[0045] Specifically, the packet query module 123 mainly has the following two technologies.

[0046] The first step is to continuously monitor for packet loss events. If a packet loss event occurs, the system will query the service status management subsystem 11 for packet loss service information, synchronize the network card association database 122, and notify the traffic management module 124 to perform analysis and correction processes. The second step is to handle service performance anomaly events notified by the target receiving module 121. Packets will be extracted using user information from the network card association database 122 and forwarded to the traffic management module 124 for analysis.

[0047] The traffic management module 124 is used to receive correction messages and analysis messages from the packet query module 123. By analyzing the transmission status of traffic packets, it performs performance control or its mechanism to find available relay points when the transmission status is congestion or abnormal transmission caused by round-trip time (RTT). It also performs abnormal transmission correction or its mechanism for traffic when the transmission status is packet loss.

[0048] Specifically, the main technology of the traffic management module 124 is to analyze the traffic packet transmission status and then configure network management settings accordingly. There are two types of transmission status: the first is transmission anomalies caused by congestion and RTT time. In this case, performance control or its mechanism will be executed to find usable relay points to improve performance. The second is traffic packet loss. In this case, abnormal transmission correction or its mechanism will be performed to ensure that the overall transmission network performance is not affected by the anomalies.

[0049] Further, in one embodiment, as shown in FIG10, the data format used by the network interface card association database 122, particularly the user correction information format, is as follows: Service ID 1001 is the service identification code that needs to be corrected provided by the service status management subsystem 11; Source IP 1002 is the service IP that needs to be corrected provided by the service status management subsystem 11; Destination IP... 1003 The service status management subsystem 11 provides the destination IP address of the service that needs to be corrected; 1004 The service status management subsystem 11 provides the user identifier of the service that needs to be corrected, which is a key-value structure, such as {"key":"value"}, where key is the ID given to the user by the service, and value is a field value that can be used as packet identification; 1005 Traffic control information is provided by the traffic management module 124, and the value is the transmission rate after abnormal transmission correction or adjustment by its mechanism (e.g., 50 gigabits per second (GB / s)); 1006 Service relay information is provided by the traffic management module 124, and the value is the cloud platform ID selected by the performance control or its mechanism; 1007 The number of adjustments is used by the traffic management module 124 to calculate the number of corrections or adjustments by its mechanism.

[0050] As shown in Figure 11, the data format used by the network interface card (NIC) association database 122 is explained, especially the information format of accelerated NICs from different cloud platforms. This information is provided by the cloud information synchronization module 114. Among them, the source ID 1101 is the port ID of the source accelerated NIC; the destination ID 1102 is the destination accelerated NIC PORT ID; the latency 1003 is the round-trip communication time (RTT) between the two ports; and the available resource percentage 1104 is the available traffic of the accelerated NIC.

[0051] The aforementioned performance control or mechanism includes the traffic management module 124 checking the relay stations recorded in the network interface card (NIC) association database 122 to find the relay station with a percentage of available resources greater than the target traffic and the shortest round-trip communication time. Then, the cloud platform is notified to perform scheduling, and the packets from the source and target cloud platforms are redirected to pass through the relay station through Software Defined Networking (SDN) technology. The relay station is configured to listen and automatically reply with acknowledgment messages (ACK) and handle retransmissions. The relay station information is written into the NIC association database 122. In addition, the relay station is removed when the traffic does not exist.

[0052] The aforementioned abnormal transmission correction or its mechanism includes the traffic management module 124 querying the network card association database 122 through user information. If the adjustment count in the user information is greater than zero, the adjustment count is reduced by one, and a hardware-supported congestion method (such as a congestion algorithm) is used to obtain a lower speed reduction value than the current one for the traffic. Then, if packet loss or congestion still occurs, the aforementioned reduction of the adjustment count and the use of the congestion method (such as a congestion algorithm) are continued until the packet loss or congestion is eliminated.

[0053] Each module of the present invention can be software, hardware, or firmware; if it is hardware, it can be a processing unit, processor, computer, or server with data processing and computing capabilities; if it is software or firmware, it can include instructions executable by the processing unit, processor, computer, or server, and can be installed on the same hardware device or distributed across different multiple hardware devices.

[0054] Figure 3 is a flowchart illustrating the steps of the long-distance network transmission quality optimization method of the present invention. As shown in the figure, the present invention manages and collects the status and requirements of each service through a cloud platform, allowing the cloud platforms to exchange platform status and latency information. This enables the evaluation of suitable transmission schemes for services, and then configures data transmission conditions according to the schemes. The smart network interface card or data processing unit is then configured for network management, thereby enabling each service to effectively and stably perform its performance and value.

[0055] In step S301, the service status management subsystem collects user communication information and information from each connected cloud platform for the managed services. If the user communication information is compared with pre-stored monitoring targets and does not meet a pre-set threshold, a correction request is generated. The subsystem also generates cloud platform messages from the information of each cloud platform for performance correction. This step explains that the service status management subsystem collects user communication information and information from each connected cloud platform for the managed services, compares it with pre-stored monitoring targets, and notifies the accelerated network interface card management subsystem to correct or adjust if any anomalies are found (e.g., discrepancies with thresholds).

[0056] In one embodiment, the execution of the service status management subsystem further includes: having the service registration module provide a user interface to receive service information, monitoring indicators, and correction methods input by the user and perform format checks; having the service association database record the data after format checks by the service registration module; having the service monitoring module periodically collect the user's communication information and compare it with the monitoring targets recorded in the service association database, so as to provide the user's communication information for the accelerated network card management subsystem to make corrections if the threshold is not met; and having the cloud information synchronization module periodically collect the cloud platform information for the accelerated network card management subsystem to perform communication performance corrections.

[0057] In step S302, the accelerated network interface card (NIC) management subsystem receives the correction request from the service status management subsystem and the cloud platform message to analyze the traffic packets. This allows the subsystem to perform performance control when a service performance anomaly notification is received, and / or perform abnormal transmission correction when packet loss occurs. This step explains that the accelerated NIC management subsystem receives the correction request from the service status management subsystem and the cloud platform message, performs traffic packet parsing, and then performs performance control or its mechanism in the event of service performance anomalies, and / or performs abnormal transmission correction or its mechanism when packet loss occurs.

[0058] In the aforementioned method, the service status management subsystem is located on the edge cloud platform, while the accelerated network card management subsystem is located on the accelerated network card device.

[0059] In one embodiment, the execution of the accelerated network interface card (NIC) management subsystem further includes: instructing the target receiving module to receive basic service information from the service status management subsystem and perform format checks; instructing the NIC association database to record the information after format checks by the target receiving module; instructing the packet query module to continuously monitor packet loss events, so as to obtain packet loss service information from the service status management subsystem when packet loss occurs, synchronize the NIC association database and generate correction messages, and when receiving a service performance anomaly event transmitted by the target receiving module, to perform packet capture using user information in the NIC association database to generate analysis messages; and instructing the traffic management module to receive the correction messages and analysis messages from the packet query module, and to perform performance control to find available relay points when the transmission situation is caused by congestion and round-trip time (RTT) anomalies by analyzing the transmission status of the traffic packets, and to perform abnormal transmission correction for the traffic when the transmission situation is traffic packet loss.

[0060] Furthermore, the aforementioned performance regulation or its mechanism and abnormal transmission correction or its mechanism will be explained later with flowcharts.

[0061] Figure 4 is a flowchart of the service status management subsystem of the present invention for processing user registration and supervision services and synchronizing cloud information, mainly illustrating the process of the service status management subsystem for processing user registration and supervision services and synchronizing cloud information.

[0062] In process 401, the service information that needs to be managed is imported through the service registration module. This process explains that the service holder imports service information and user information into the service status management subsystem through the service registration module.

[0063] In process 402, the service registration module verifies whether the operation is complete. This process describes how the service registration module checks the information provided by the service holder, which must include resource usage, user ID entries, targets and thresholds, adjustment counts, and exception limit policies. If any information is missing, proceed to process 403; otherwise, proceed to process 404.

[0064] The process ended at step 403, indicating a failure to be included in the management system. This process explains that the service registration module reported a failure to be included in the management system due to missing necessary information.

[0065] In process 404, the service registration module assigns a service ID and writes the information to the service association database. This process describes how the service registration module assigns a service ID to the input data and writes it to the network interface card (NIC) association database.

[0066] In process 405, the service monitoring module periodically retrieves data from the service-related database, obtains the data and detection thresholds. If the thresholds are exceeded, the acceleration network interface card (NIC) management subsystem is notified. This process explains that the service monitoring module periodically polls all service data from the NIC-related database, and on a user-by-user basis, progressively obtains the target and threshold (i.e., component 904 in Figure 9) indicated by the target and threshold. It then compares the target and threshold values ​​(which can be user experience parameters such as transmission rate, processing speed, or user feedback values) to the threshold values. If any value exceeds the threshold, the acceleration NIC management subsystem is notified to perform corrective actions; otherwise, the process continues to check the next service or user.

[0067] In process 406, the cloud information synchronization module periodically acquires information and latency between cloud platforms and synchronizes it to the accelerated network interface card management subsystem. This process describes how the cloud information synchronization module periodically acquires information such as source ID, destination ID, latency time, and available resource percentage between cloud platforms and synchronizes it to the accelerated network interface card management subsystem for subsequent relay point selection.

[0068] Figure 5 is a flowchart of the process by which the accelerated network card management subsystem receives data provided by the service status management subsystem, illustrating the processing flow of the accelerated network card management subsystem receiving information such as service ID, information of a single user in the user identifier queue, adjustment count, and abnormal limit policy provided by the service status management subsystem.

[0069] In process 501, the target receiving module receives correction and cloud platform information. This process describes how the target receiving module receives correction information and cloud platform information. The correction information includes the service ID, information about a single user in the user identifier queue, the number of adjustments, and the exception limit policy. The cloud platform information includes the source ID, destination ID, latency time, and percentage of available resources.

[0070] In process 502, the target receiving module verifies the correctness of the information. This process describes how the target receiving module verifies the correctness of the information. If it fails, it proceeds to process 503; if it succeeds, it proceeds to process 504.

[0071] At process 503, the processing flow ends, and the report fails. This process indicates that the target receiving module's report service status management subsystem failed to correct the error or that the cloud platform data is abnormal.

[0072] In process 504, the target receiving module writes the information to the network interface card (NIC) associated database. This process describes how the target receiving module writes the data (including service ID, source IP, destination IP, user identifier, traffic control information, service relay station information, and adjustment count) back to the NIC associated database and notifies the packet query module to retrieve packets.

[0073] In process 505, the packet query module obtains service information through the network interface card (NIC) association database and listens for service packets, thereby synchronizing packet information with the traffic management module. This process explains that the packet query module retrieves the packet listening conditions (i.e., source IP, destination IP, and user identifier) ​​from the NIC association database, configures the NIC monitoring settings, obtains packet samples over a period of time, and then organizes the samples into a transmission pattern (i.e., the packet delivery context, not the packet itself) and synchronizes it to the traffic management module.

[0074] In process 506, the traffic management module analyzes the service packet situation. If congestion, ACK response time exceeding RTT, or RTT intervals cause service delays, performance control is performed. This process explains that the traffic management module analyzes the service packet situation. If user congestion, automatic acknowledgment message (ACK) response time exceeding round-trip time (RTT), or RTT intervals cause service delays, performance control or its mechanisms are executed.

[0075] Figure 6 is a flowchart of the packet loss handling process of the accelerated network card management subsystem of the present invention, illustrating the handling process when packet loss is detected after the accelerated network card management subsystem continuously monitors the accelerated network card packets.

[0076] In process 601, the packet query module detects a packet loss event, and the relevant packet information is not being processed. This process explains that when the packet query module monitors the accelerated network interface card (NIC) device, if it detects a packet loss event that is not being processed (i.e., no program is using the corresponding service ID, source IP, destination IP, user identifier, traffic control information, service relay station information, and adjustment count as conditions to correct performance), it will continue to monitor the corresponding event.

[0077] In process 602, determine whether continuous packet loss is occurring. This process describes how to observe the corresponding packet loss event. If it is occurring continuously, proceed to process 603; otherwise, end the observation.

[0078] In process 603, the traffic management module reports back to the service status management subsystem, awaiting feedback information from the system. This process describes how the traffic management module reports back to the service status management subsystem using the source IP and destination IP, and waits for feedback information from the service status management subsystem.

[0079] In process 604, is there any feedback? This process explains that the service status management subsystem provides feedback. If the cloud platform hosts the service using the source IP and destination IP, then process 606 is executed; otherwise, process 605 is executed.

[0080] In process 605, the report is skipped. This process explains that the report is skipped because the cloud platform does not host the service with the corresponding parameters, indicating that the cloud platform is currently a relay point, not the source or destination, and therefore is not processed.

[0081] In process 606, the traffic management module stores information in the network interface card (NIC) associated database and analyzes the service packet context, performing abnormal transmission correction. This process describes how the traffic management module writes data provided by the service status management subsystem (i.e., service ID, resource usage, user identifier queue, target and threshold, adjustment count, and abnormal limit policy) into the NIC associated database (i.e., service ID, source IP, destination IP, user identifier, traffic control information, service relay station information, and adjustment count). After analysis, if the packet is from a transmitting end, abnormal transmission correction or its mechanism is performed; if it is from a receiving end, processing is skipped.

[0082] Figure 7 is a flowchart of the performance control process of the service management module of the present invention, illustrating the control process of performance control for a specified service.

[0083] In process 701, check the relay stations in the network interface card (NIC) association database to see if there are any relay stations with an available resource percentage greater than the target traffic and the minimum RTT. This process explains that performance control or its mechanism queries the transmitting port ID (i.e., the value corresponding to the source ID) and the destination port ID (i.e., the value corresponding to the destination ID) based on the source IP and destination IP. Then, it queries the corresponding latency (i.e., the current RTT) in the NIC association database. Next, it queries the destination ID column (i.e., the relay point PORT ID column, for example, ["uuid1", "uuid2", ...]) whose source ID is the source port ID, latency is less than the current RTT, and available resource percentage is greater than or equal to the current service traffic. Then, using the destination port ID as the destination ID and the relay point PORT ID column as the source ID, it finds the relay point PORT IDs where the latency from the relay point to both endpoints is less than the RTT, the total RTT is less than the minimum, and the available resource percentage is greater than or equal to the current service traffic.

[0084] In process 702, determine if there is an available relay point. This process confirms whether a relay point PORT ID has been found. If so, proceed to process 704; otherwise, proceed to process 703.

[0085] In process 703, the processing flow ends, and the report indicates that no nearby cloud platform is available. This process description aims to improve performance by ending the processing flow and reporting that no relay point is available.

[0086] In process 704, the cloud platform is notified to perform scheduling, and the source and destination cloud platform packets are redirected to this relay station cloud platform via SND technology. This process explains that the traffic management module notifies the cloud information synchronization module to synchronize the corresponding bypass rules configured on the relay end and the destination end to the accelerated network card device, so that service packets will pass through the relay end before reaching the destination end.

[0087] In process 705, the relay station's traffic management module is configured to listen for packets received from the source and forward them to the destination. This process describes how the relay station's traffic management module enables the buffered packet forwarding function, receiving and forwarding specified service packets to the destination.

[0088] In process 706, the relay station's traffic management module configures the user packet listening service to automatically reply to user ACKs and handle retransmissions. The ACK reply requires service support confirmation. This process explains that the relay station's traffic management module enables the automatic reply of ACK packets with specified identifiers (i.e., the source IP, destination IP, and user identifier for the service correction) and the retransmission packet function, but it requires confirmation that the service supports the automatic ACK reply mode and that it does not cause service anomalies.

[0089] In process 707, the traffic adjustment process. This process describes how the source-side traffic management module initiates a traffic adjustment process to adjust the control rate to suit the current situation (further explanation will be provided in Figure 8).

[0090] In process 708, the source-side traffic management module backfills relay station information to the network interface card (NIC) association database. If the user no longer exists, the relay point is removed. This process explains that the source-side traffic management module stores data (i.e., traffic control information, service relay station information, and adjustment count) back to the NIC association database. Furthermore, when the user's service ends, the service status management subsystem notifies the traffic management module to remove the cloud platform relay point setting.

[0091] Figure 8 is a flowchart of the abnormal transmission correction process performed by the traffic management module of the present invention. It illustrates the process by which the traffic management module adjusts the abnormal user traffic by configuring the acceleration network card settings when it detects packet loss, so as to maintain the overall network performance.

[0092] In process 801, the abnormal transmission correction system receives user information and queries the network interface card (NIC) associated database. This process describes how the traffic management module queries the NIC associated database for information (i.e., service ID, source IP, destination IP, user identifier, traffic control information, service relay station information, and adjustment count).

[0093] In process 802, check if the information is correct. This process describes how the communication management module checks if the data exists. If it does, proceed to process 804; otherwise, proceed to process 803.

[0094] At process 803, the correction process ends, and the report fails. This process indicates that the communication management module reported incorrect data, thus ending the correction process.

[0095] In process 804, it is determined that the number of adjustments is greater than 0. This process describes whether the communication management module checks whether the number of adjustments is greater than 0. If it is greater than 0, process 806 is executed; if it is less than or equal to 0, process 805 is executed.

[0096] In process 805, based on the anomaly response strategy, user packets are dropped or the minimum operational traffic is limited, and the adjustment count is set to 0 and stored in the network interface card (NIC) associated database. This process explains that the traffic management module decides its course of action based on the anomaly limit policy value provided by the service status management subsystem. If packet loss still occurs even after configuring the minimum operational rate set by the anomaly limit policy, the NIC is configured to drop user packets to maintain network stability. The service status management subsystem is then notified of the service anomaly, and the results are stored in the NIC associated database (i.e., traffic control information, service relay station information, and adjustment count).

[0097] In process 806, the adjustment count is reduced by one, and a lower speed reduction value is obtained by utilizing hardware-supported congestion methods (such as congestion algorithms). This process explains that the traffic management module reduces the obtained adjustment count by one and uses methods supported by the network card (such as speed reduction algorithms like DCQCN - Data Center Quantized Congestion Notification, HPCC - High Performance Congestion Control, or DCTCP - Data Center Transmission Control Protocol) to obtain a speed reduction value that is lower than the current speed.

[0098] In process 807, configure the speed reduction settings for users of the limited service and observe whether packet loss or congestion continues to occur. This process explains that the traffic management module configures the speed reduction settings to the accelerated network interface card (NIC) device through software-defined networking (SDN) technology and observes the packet transmission situation.

[0099] In process 808, it is determined whether packet loss or congestion has occurred. This process describes how the communication management module checks the packet transmission status. If there is an abnormality, process 804 is executed; otherwise, process 809 is executed.

[0100] In process 809, traffic control information and other information are stored back in the NIC-associated database. This process explains that the traffic management module confirms that the current user traffic will not affect the overall network and stores the data back in the NIC-associated database (i.e., traffic control information, service relay information, and adjustment count).

[0101] Furthermore, this invention discloses a computer-readable medium applied in a computing device or computer having a processor (e.g., CPU, GPU, etc.) and / or memory, storing instructions, and capable of being executed by the computing device or computer through the processor and / or memory to perform the aforementioned methods and steps when executing the computer-readable medium. In one embodiment, the computer-readable medium is a non-transitory computer-readable storage medium.

[0102] As can be seen from the above, the long-distance network transmission quality optimization system, method, and computer-readable medium of the present invention utilize a cloud platform to manage and collect the status and requirements of various services, enabling the exchange of platform status and latency information between cloud platforms. Based on this, a suitable transmission scheme is evaluated for the service. Subsequently, data transmission conditions are configured according to the scheme, and network management configuration is performed on the smart network interface card or data processing unit, allowing each service to effectively and stably perform its performance and value. Therefore, the present invention possesses the following performance and advantages.

[0103] First, this invention provides an edge cloud service quality monitoring technology to address the high-speed and stable transmission characteristics of cloud-based multi-service support. Through this edge cloud service quality monitoring technology, a service status management subsystem is introduced into the cloud platform. This subsystem can obtain the performance and operational status of managed services in real time, and the transmission status of services on the accelerated network interface card (NIC) can be obtained in real time through the underlying accelerated NIC management subsystem. These two pieces of real-time information are then evaluated to monitor the health and quality of the services. Therefore, this invention can instantly detect any service performance degradation, large amounts of packet loss, and continuous retransmission anomalies. These anomalies not only affect the specific service but also all other cloud services sharing the accelerated NIC, as the network path becomes filled with a large number of retransmitted packets. These packets may increase over time, consuming more network resources and causing a severe decline in overall network performance. Compared to existing technologies, current methods often only assess the packet transmission status of currently abnormal services to limit traffic or increase speed. This leads to repeated switching between service availability and inoperability, causing the overall network to be congested and unreliable at times, potentially resulting in network service malfunction or even service corruption. In contrast, this invention manages abnormal transmission behaviors involving significant packet loss and retransmission through a cloud platform, intervenes in abnormal cloud service activities, and then adjusts the transmission status of cloud services to prevent other cloud services from being affected by abnormal transmissions, maintaining the stability and effective operation of the overall cloud service. Furthermore, it assesses the possible causes of abnormal transmissions. If performance degradation, packet loss, or congestion is due to physical distance factors, it identifies relay stations between connected edge cloud platforms to mitigate physical distance limitations, achieving effective and stable provision of cloud services to users.

[0104] Secondly, the service status management subsystem of this invention includes a service registration module, a service supervision module, and a cloud information synchronization module. The service registration module is used to manage existing services or establish new services on the cloud platform, and requires users to provide the user identifier, resource usage, target, and threshold of the service to complete the registration. After successful service registration, the service supervision module will obtain the operational status of the service based on the supervision target. The supervision information can be actively collected by the service supervision module or provided by the service provider. After being digitized, this information will be synchronized to the smart NIC through the management channel, and then analyzed and used by the NIC's traffic management module to manage traffic, thereby achieving the advantages of real-time supervision and task distribution. The cloud information synchronization module is used to synchronize information between edge clouds, enabling the NIC management subsystem to evaluate available solutions.

[0105] Third, the accelerated network interface card (NIC) management subsystem of this invention includes a target receiving module, a packet query module, and a traffic management module. The target receiving module is used to collect monitoring and alarm data provided by the service status management subsystem, while the packet query module is used to collect NIC information and service packet status. This data can be obtained periodically or in real time. The traffic management module will periodically or proactively evaluate the target data and NIC data to determine whether the service goals have been achieved and whether the NIC is operating normally. Finally, based on the evaluation results, the traffic management module classifies the service status into performance control or abnormal transmission. For abnormal transmission, it performs service abnormal transmission correction; for performance control, it performs network performance control to correct the service performance and stability in real time.

[0106] Fourth, this invention features service anomaly transmission correction. It assesses the performance and stability of services on the cloud platform based on network interface card (NIC) operation information, service operation status, and service packet information, and further adjusts the transmission status of services assessed as abnormal. The assessment method relies on service status and NIC operation information. If any anomaly occurs, a transmission correction process is initiated, considering resource usage, transmission rate, congestion, and packet loss to determine if correction is necessary. If correction is required, the transmission correction process is initiated; otherwise, the next piece of information is checked. During the transmission correction process, the service IP, destination IP, and user identifier are used as identifiers to gradually adjust the service's traffic until packet loss and congestion caused by the service packets no longer occur. Ultimately, this ensures that all service users are supported by the most stable transmission method available at present, without affecting overall network service performance.

[0107] Fifth, this invention features network performance regulation, proposing a network enhancement method based on smart NICs or data processing units (DPUs) and software-defined networking (SDN) technology. This method aims to improve network data transmission speed without affecting overall service performance and maintaining stability, thereby effectively and reliably providing cloud services to users. This network performance regulation utilizes the accelerated NIC management subsystem of the smart NIC or DPU, using RTT (Round-Trip Time) as a filtering criterion to identify suitable relay stations (cloud platforms) to address issues such as user traffic congestion, packet loss, or low performance. Furthermore, the relay station's traffic management module listens for and responds to ACK packets and caches packets from the source end to improve transmission performance and reduce latency. When a user is no longer present, the source end's traffic management module notifies the relevant cloud platform to remove the associated relay station configuration.

[0108] The above detailed description is a specific description of one feasible embodiment of the present invention. However, this embodiment is not intended to limit the patent scope of the present invention. All equivalent implementations or modifications that do not depart from the spirit of the present invention should be included in the patent scope of the present invention.

[0109] 1: Long-distance network transmission quality optimization system

[0110] 11: Service Status Management Subsystem

[0111] 111: Service Registration Module

[0112] 112: Service-related database

[0113] 113: Service Supervision Module

[0114] 114: Cloud Information Synchronization Module

[0115] 12: Accelerated Network Interface Card Management Subsystem

[0116] 121: Target receiving module

[0117] 122: Network Interface Card (NIC) Association Database

[0118] 123: Packet Query Module

[0119] 124: Communication Management Module

[0120] 2: Edge Cloud Platform

[0121] 3: Accelerating network card devices

[0122] 901: Service ID

[0123] 902: Resource Usage

[0124] 903: User ID Queue

[0125] 904: Target and Threshold

[0126] 905: Number of adjustments

[0127] 906: Abnormal Limit Strategy

[0128] 1001: Service ID

[0129] 1002: Source IP

[0130] 1003: Destination IP

[0131] 1004: User ID

[0132] 1005: Traffic Control Information

[0133] 1006: Service Relay Station Information

[0134] 1007: Number of adjustments

[0135] 1101: Source ID

[0136] 1102: Destination ID

[0137] 1103: Delay Time

[0138] 1104: Percentage of available resources

[0139] 401-406: Process

[0140] 501-506: Process

[0141] 601-606: Process

[0142] 701-708: Process

[0143] 801-809: Process

[0144] S301-S302: Steps

Claims

1. A long-distance network transmission quality optimization system, comprising: The service status management subsystem is used to collect user traffic information and information from connected cloud platforms for managed services. It generates correction requests when user traffic information is compared with pre-stored monitoring targets and does not meet pre-set thresholds, and generates cloud platform messages from the information of each cloud platform for performance correction. The accelerated network interface card (NIC) management subsystem is connected to the service status management subsystem and receives the correction requests and cloud platform messages from the service status management subsystem to analyze traffic packets. Upon receiving notification of a service performance anomaly, it performs corresponding performance control, and / or performs abnormal transmission correction when packet loss occurs. The accelerated NIC management subsystem includes a target receiving module for receiving basic service data from the service status management subsystem and performing format checks. The network interface card (NIC) association database is used to record data after format checking by the target's receiving module; the packet query module is used to continuously monitor packet loss events, and when packet loss occurs, obtain packet loss service information from the service status management subsystem to synchronize the NIC association database and generate correction messages, and when receiving a service performance anomaly event transmitted by the target's receiving module, perform packet capture using the user information in the NIC association database to generate analysis messages; and the traffic management module is used to receive the correction messages and analysis messages from the packet query module, and by analyzing the transmission status of the traffic packets, perform performance control to find available relay points when the transmission status is congestion and round-trip communication time-related transmission anomalies, and perform abnormal transmission correction for the traffic when the transmission status is traffic packet loss.

2. The long-distance network transmission quality optimization system as described in claim 1, wherein, The service status management subsystem is located on the edge cloud platform, while the accelerated network card management subsystem is located on the accelerated network card device.

3. The long-distance network transmission quality optimization system as described in claim 1, wherein, The service status management subsystem includes: a service registration module, which provides a user interface for receiving service information, monitoring indicators, and correction methods input by the user and performing format checks; a service association database, which records data after format checks by the service registration module; a service monitoring module, which periodically collects the user's communication information and compares it with the monitoring targets recorded in the service association database, so that if the user's communication information does not meet the threshold, the user's communication information is provided to the accelerated network card management subsystem for correction; and a cloud information synchronization module, which periodically collects information from the cloud platform as a basis for the accelerated network card management subsystem to perform communication performance correction.

4. The long-distance network transmission quality optimization system as described in claim 1, wherein, The performance control system includes the traffic management module checking the relay stations recorded in the network interface card (NIC) associated database to find the relay station with an available resource percentage greater than the target traffic and the shortest round-trip communication time. The system then notifies the cloud platform to perform scheduling, using software-defined networking (Software-Defined Networking) technology to redirect packets from the source to the target cloud platform through the relay station. The system also includes setting up a listening configuration within the relay station to automatically reply with acknowledgment messages and handle retransmissions, writing the relay station information into the NIC associated database, and removing the relay station when the traffic is absent.

5. The long-distance network transmission quality optimization system as described in claim 1, wherein, The abnormal transmission correction system includes the traffic management module querying the network card's associated database through the user information. If the adjustment count in the user information is greater than zero, the adjustment count is reduced by one, and a hardware-supported congestion method is used to obtain a lower speed reduction value than the current value for the traffic. This allows the system to continue reducing the adjustment count and using the congestion method even when packet loss or congestion still occurs, until the packet loss or congestion is eliminated.

6. A method for optimizing long-distance network transmission quality, comprising the following steps: a service status management subsystem collects user traffic information and information from each connected cloud platform under its management; generates a correction request when the user traffic information is compared with pre-stored monitoring targets and does not meet a pre-set threshold; and generates cloud platform messages from the information of each cloud platform for performance correction; and an accelerated network interface card (NIC) management subsystem receives the correction request and cloud platform messages from the service status management subsystem to perform traffic packet analysis, so as to perform performance control upon receiving notification of a service performance anomaly event, and / or perform abnormal transmission correction upon the occurrence of traffic packet loss, wherein... The accelerated network interface card (NIC) management subsystem executes the following: instructs the target receiving module to receive basic service information from the service status management subsystem and perform format checks; instructs the NIC association database to record the information after format checks by the target receiving module; instructs the packet query module to continuously monitor packet loss events, so as to obtain packet loss service information from the service status management subsystem when packet loss occurs, so as to synchronize the NIC association database and generate correction messages; and when the target receiving module receives a service performance anomaly event, it performs packet capture using the user information in the NIC association database to generate analysis messages. The communication management module receives the correction message and the analysis message from the packet query module. By analyzing the transmission status of the communication packets, it performs performance control to find available relay points when the transmission status is congestion and round-trip communication time is the cause of the transmission anomaly, and performs abnormal transmission correction for the communication when the transmission status is packet loss.

7. The method for optimizing long-distance network transmission quality as described in claim 6, wherein, The service status management subsystem is located on the edge cloud platform, while the accelerated network card management subsystem is located on the accelerated network card device.

8. The method for optimizing long-distance network transmission quality as described in claim 6, wherein, The service status management subsystem performs the following actions: The service registration module provides a user interface to receive service information, monitoring indicators, and correction methods input by the user and performs format checks; the service-related database records the data after format checks by the service registration module; the service monitoring module periodically collects the user's communication information and compares it with the monitoring targets recorded in the service-related database, so that if the threshold is not met, the user's communication information is provided to the accelerated network card management subsystem for correction; and the cloud information synchronization module periodically collects information from the cloud platform to use as a basis for the accelerated network card management subsystem to perform communication performance corrections.

9. The method for optimizing long-distance network transmission quality as described in claim 6, wherein, The performance control system includes the traffic management module checking the relay stations recorded in the network interface card (NIC) associated database to find the relay station with an available resource percentage greater than the target traffic and the shortest round-trip communication time. The system then notifies the cloud platform to perform scheduling, using software-defined networking (Software-Defined Networking) technology to redirect packets from the source to the target cloud platform through the relay station. The system also includes setting up a listening configuration within the relay station to automatically reply with acknowledgment messages and handle retransmissions, writing the relay station information into the NIC associated database, and removing the relay station when the traffic is absent.

10. The method for optimizing long-distance network transmission quality as described in claim 6, wherein, The abnormal transmission correction system includes the traffic management module querying the network card's associated database through the user information. If the adjustment count in the user information is greater than zero, the adjustment count is reduced by one, and a hardware-supported congestion method is used to obtain a lower speed reduction value than the current value for the traffic. This allows the system to continue reducing the adjustment count and using the congestion method even when packet loss or congestion still occurs, until the packet loss or congestion is eliminated.

11. A computer-readable medium, applied in a computing device or computer, storing instructions for performing the long-distance network transmission quality optimization method as described in any one of claims 6 to 10.

Images