Agent-based delay determination method, device, and computer program product

Abstract

The present disclosure provides a method, device, electronic device, storage medium and computer program product for delay determination based on an intelligent agent, which relates to the field of computer technology, specifically to the field of artificial intelligence large models, natural language understanding, and delay detection technology, and can be applied to delay detection scenarios. The specific implementation scheme is: determining the set of tested devices corresponding to the delay detection task; for the tested devices in the tested device set, sending a detection data packet to the destination device corresponding to the tested device in the tested device set through the delay detection intelligent agent in the tested device, and determining the time point at which the detection data packet and the response data packet of the detection data packet are at each network node during the transmission process; according to the time point, determining the delay data corresponding to the delay detection task. The present disclosure enables the delay detection method to be applicable to various delay detection scenarios, and ensures the accuracy of the determined delay data.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, specifically to the field of artificial intelligence big models, natural language understanding, and delay analysis technology, and more particularly to an agent-based delay determination method, device, electronic device, storage medium, and computer program product, which can be applied in delay analysis scenarios. Background Art

[0002] Network latency measurement is used to determine the time required for data transmission within a network and is an important means of evaluating network performance. However, current network latency measurement methods cannot achieve both accuracy and versatility. Summary of the Invention

[0003] The present disclosure provides an agent-based delay determination method, device, electronic device, storage medium, and computer program product.

[0004] According to a first aspect, an agent-based delay determination method is provided, comprising: determining a set of tested devices corresponding to a delay detection task; for a tested device in the tested device set, sending a probe data packet to a destination device corresponding to the tested device in the tested device set through a delay detection agent in the tested device, and determining a time point at which the probe data packet and a response data packet to the probe data packet are at each network node during the transmission process; and determining the delay data corresponding to the delay detection task based on the time point.

[0005] According to the second aspect, an agent-based delay determination device is provided, including: a set determination unit, configured to determine a set of tested devices corresponding to a delay detection task; a detection unit, configured to send a detection data packet to a destination device corresponding to the tested device in the tested device set through a delay detection agent in the tested device, and determine the time points at which the detection data packet and a response data packet to the detection data packet are at each network node during the transmission process; the delay determination unit is configured to determine the delay data corresponding to the delay detection task according to the time points.

[0006] According to a third aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so as to enable the at least one processor to execute the method described in any implementation manner of the first aspect.

[0007] According to a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, where the computer instructions are used to cause a computer to execute the method as described in any implementation of the first aspect.

[0008] According to a fifth aspect, a computer program product is provided, comprising: a computer program, which implements the method described in any implementation manner of the first aspect when executed by a processor.

[0009] According to the technology disclosed in the present invention, a delay determination method and device based on an intelligent agent are provided. A delay detection intelligent agent is deployed in the device under test. The delay detection intelligent agent can determine the time points of each network node in the transmission process of the detection data packet and the response data packet between the device under test and the destination device to determine the delay data, so that the delay detection method can be applied to various delay detection scenarios and the accuracy of the determined delay data is guaranteed.

[0010] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0012] Figure 1 is an exemplary system architecture diagram in which an embodiment of the present disclosure may be applied;

[0013] Figure 2 is a flow chart of an embodiment of an agent-based delay determination method according to the present disclosure;

[0014] Figure 3 is a schematic diagram of the transmission path of the probe data packet and the response data packet according to this embodiment;

[0015] Figure 4 is a schematic diagram of a network card pair according to this embodiment;

[0016] Figure 5 is a schematic diagram of the data aggregation process according to this embodiment;

[0017] Figure 6 is a schematic diagram of an application scenario of the agent-based delay determination method according to this embodiment;

[0018] Figure 7 is a schematic diagram of a heat map of delay data according to this embodiment;

[0019] Figure 8 is a flowchart of another embodiment of the agent-based delay determination method according to the present disclosure;

[0020] Figure 9 is a structural diagram of an embodiment of an agent-based delay determination device according to the present disclosure;

[0021] Figure 10 It is a schematic diagram of the structure of a computer system suitable for implementing the embodiments of the present disclosure. DETAILED DESCRIPTION

[0022] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0023] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0024] Figure 1 An exemplary architecture 100 is shown to which the agent-based delay determination method and apparatus of the present disclosure can be applied.

[0025] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. The communication connections between terminal devices 101, 102, and 103 constitute a topological network, and network 104 is used to provide a medium for communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0026] Terminal devices 101, 102, and 103 can be hardware devices or software that support network connection for data interaction and data processing. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices that support network connection, information acquisition, interaction, display, processing, and other functions, including but not limited to smartphones, tablet computers, e-book readers, laptop computers, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the electronic devices listed above. They can be implemented as multiple software or software modules, for example, to provide distributed services, or they can be implemented as a single software or software module. No specific limitations are given here.

[0027] Server 105 can be a server that provides various services, such as a backend processing server that receives latency detection tasks sent by terminal devices 101, 102, and 103, determines the set of devices under test corresponding to the latency detection tasks, and then performs latency detection operations through latency detection agents deployed in the devices under test. Optionally, the server can feed back the finalized latency data to the terminal device. As an example, server 105 can be a cloud server.

[0028] It should be noted that a server can be either hardware or software. When a server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When a server is software, it can be implemented as multiple software programs or software modules (for example, software or software modules used to provide distributed services), or as a single software program or software module. This is not specifically limited here.

[0029] It should also be noted that the agent-based delay determination method provided in the embodiments of the present disclosure is generally executed by a server, but this does not preclude execution by a terminal device, or the possibility of a server and terminal device cooperating to perform the method. Accordingly, the various components (e.g., various units) of the agent-based delay determination apparatus can be entirely located in the server, entirely located in the terminal device, or even separately located in the server and the terminal device.

[0030] It should be understood that Figure 1 The number of terminal devices, networks, and servers in the system is merely illustrative. Any number of terminal devices, networks, and servers may be provided depending on implementation requirements. When the electronic device on which the agent-based delay determination method is running does not need to transmit data with other electronic devices, the system architecture may include only the electronic device (e.g., a terminal device or server) on which the agent-based delay determination method is running.

[0031] Please refer to Figure 2 , Figure 2 This is a flow chart of an agent-based delay determination method provided in an embodiment of the present disclosure. Process 200 includes the following steps:

[0032] Step 201: Determine a set of devices under test corresponding to a delay detection task.

[0033] In this embodiment, the execution subject of the agent-based delay determination method (for example, Figure 1 The server in the embodiment of the present invention can obtain the delay detection task remotely or locally through a wired network connection or a wireless network connection, and determine the set of devices under test corresponding to the delay detection task.

[0034] Latency detection tasks are used to detect network latency. They can be used in various scenarios and systems with high network latency requirements, including but not limited to:

[0035] 1. Cloud computing and data centers: Network monitoring of virtual machines / containers, especially scenarios involving high-performance transport and storage networks. Examples of high-performance transport networks include SR-IOV (Single Root I / O Virtualization and Sharing) networks and DPDK (Data Plane Development Kit) acceleration networks; and high-performance storage networks include RoCE (RDMA over Converged Ethernet) and iWARP (Internet Wide Area RDMA Protocol).

[0036] 2. HPC (High Performance Computing): MPI (Multi Point Interface) communication optimization, supplementing or replacing native IB (InfiniBand) tools, providing a unified view across nodes and Ethernet-based components or easier-to-integrate monitoring solutions.

[0037] 3. Financial trading systems: Monitoring and rapid fault location of low-latency networks (including Ethernet and RDMA).

[0038] 3. Online games and real-time audio and video: server cluster network quality monitoring.

[0039] 4. Large-scale distributed systems / microservice architecture: monitoring network latency between service calls.

[0040] 5. Artificial Intelligence / Machine Learning Clusters: Performance analysis of communication networks between nodes in distributed training.

[0041] 6. CDN (Content Delivery Network): Detect network quality between nodes.

[0042] 7. Intranet: Performance monitoring between key business system servers.

[0043] A latency detection task can be targeted at all devices in the scenario or system to be detected, or it can be targeted at a subset of devices. For example, a latency detection task can be targeted at only a subset of key devices.

[0044] As an example, a latency detection task is associated with a user's business and is used to characterize business-related information. The aforementioned execution entity can determine the target business associated with the latency detection task, and then determine the devices involved in the target business based on the business-related information to determine the set of devices under test. For example, a user applies for 80 GPU (Graphics Processing Unit) servers to deploy a large model cluster to run a large model business. The latency detection task can be targeted at the large model business. In this case, the set of devices under test corresponding to the latency detection task is the set including the aforementioned 80 GPU servers.

[0045] As another example, the execution entity may determine the set of devices under test based on the device ID (identity document) included in the latency detection task. The set of devices under test includes the devices corresponding to the device ID in the latency detection task. For example, the execution entity may visualize the devices under test using a display device and determine the set of devices under test based on a selection operation by the user.

[0046] The tested devices in the tested device set include but are not limited to network terminal devices such as servers and workstations, network connection devices such as switches and routers, and network access devices such as modems and wireless access points.

[0047] In step 202, for the device under test in the device under test set, a delay detection agent in the device under test sends a detection data packet to the destination device corresponding to the device under test in the device under test set, and determines the time point at which the detection data packet and the response data packet of the detection data packet are at each network node during the transmission process.

[0048] In this embodiment, the execution entity may transmit a probe data packet to a destination device corresponding to the device under test in the device set through a latency detection agent in the device under test, and determine the time point at which the probe data packet reaches each network node during the transmission process. The network nodes during the transmission process may be, for example, various intermediate devices along the transmission path.

[0049] A latency detection agent is deployed on each device under test. The latency detection agent has the following functions:

[0050] 1. Possess autonomous detection capabilities

[0051] Automatic latency detection tasks: Automatically send probe packets to the destination device based on pre-set rules or policies without manual intervention. For example, probe packets can be continuously sent at a specified interval (e.g., every 10 seconds) to continuously monitor network latency.

[0052] Automatically generate probe packets: Probe packets with specific formats and content are automatically generated based on detection requirements. For example, ICMP (Internet Control Message Protocol) or custom protocol packets containing unique identifiers, transmission timestamps, and possibly sequence numbers are generated to accurately identify and calculate latency.

[0053] 2. Accurately record timestamps

[0054] When sending a probe packet, the timestamp of the probe packet leaving the sender is accurately recorded. When receiving a response packet from the destination device, the timestamp of the response packet arriving at the receiver is also accurately recorded. During packet transmission, the timestamps of the probe and response packets at each network node are also accurately recorded. The higher the timestamp accuracy, the more accurate the calculated latency result. Generally, timestamp accuracy is required to be in the microsecond or even nanosecond range.

[0055] As an example, for each device under test in the set of devices under test, the destination device corresponding to the device under test is determined through the task requirements of the delay detection task; then, for each device under test in the set of devices under test, the delay detection agent in the device under test sends a detection data packet to the destination device corresponding to the device under test in the set of devices under test, and receives a response data packet returned by the target device based on the detection data packet, and determines the time point at which the detection data packet and the response data packet are at each network node during the transmission process.

[0056] As another example, for each device under test in the set of devices under test, the destination device corresponding to the device under test is determined based on the routes involved in the device under test in the network, and a delay detection agent in the device under test is used to send a probe data packet to the destination device corresponding to the device under test in the set of devices under test, and a response data packet returned by the target device based on the probe data packet is received to determine the time points at which the probe data packet and the response data packet are at each network node during the transmission process.

[0057] Continue to refer Figure 3, shows a schematic diagram of the transmission path of probe and response packets. Application 3011 in device under test 301 sends a probe packet to destination device 302 via network card 3012 in device under test 301. The probe packet arrives at time points T1 and T2 on network card 3012 of device under test 301 and network card 3022 of destination device 302, respectively. Application 3021 in destination device 302 generates a response packet based on the received probe packet and returns the response packet to device under test 301 via network card 3022 in destination device 302. The response packet arrives at time points T3 and T4 on network card 3022 of destination device 302 and network card 3012 of device under test 301, respectively.

[0058] In some optional implementations of this embodiment, the execution entity may perform the process of sending the detection data packet in step 202 as follows:

[0059] In the first step, for a device under test in the device under test set, a device pair is determined, which has the device under test as a starting device under test and another device under test outside the device under test in the device under test set as a destination device under test.

[0060] As an example, for a device under test in a set of devices under test, multiple device pairs corresponding to the device under test are determined, with the device under test serving as the starting device under test and each other device under test in the set of devices under test other than the device under test serving as the destination device under test. Each device pair includes the device under test and one other device under test in the set of devices under test other than the device under test.

[0061] By generating a device pair between the device under test and each other device under test in the device under test set except the device under test, full interconnected detection of the devices under test in the device under test set can be achieved through subsequent steps.

[0062] In the second step, a delay detection agent in the starting device under test in the device pair sends a detection data packet to the destination device under test in the device pair.

[0063] Each device under test in the device under test set corresponds to multiple device pairs. For all device pairs corresponding to all devices under test in the device under test set, the delay detection agent in the starting device under test in the device pair sends a detection data packet to the destination device under test in the device pair.

[0064] As an example, the delay detection agent is implemented based on software and is mainly composed of a sending module, a receiving module, a clock synchronization module, and a data processing module. Among them, the sending module generates a detection data packet according to preset rules and sends it to the destination device under test via the network interface of the starting device under test. The receiving module receives the response data packet sent by the destination device under test based on the detection data packet. The clock synchronization module uses a clock synchronization algorithm to ensure that the clocks of the starting device under test and the destination device under test are synchronized, providing an accurate time reference for time point measurement. The data processing module records the time points of the detection data packet and the response data packet at each network node, and analyzes and calculates indicators such as transmission delay.

[0065] As another example, the latency detection agent is implemented based on hardware and is primarily composed of a sending control unit, a receiving control unit, a hardware timestamp unit, and a data processing unit. The sending control unit generates a probe data packet and sends it to the target device under test via a hardware interface, triggering the hardware timestamp unit to record the sending time. The receiving control unit receives a response data packet from the target device under test and triggers the hardware timestamp unit to record the receiving time. The hardware timestamp unit is highly accurate and can accurately record the time points at which the probe and response data packets are at each network node during transmission. The data processing unit calculates the transmission delay based on the recorded time points and analyzes network performance indicators.

[0066] In this implementation, full interconnection delay detection of the devices under test in the device set is provided, which expands the delay detection range and improves the comprehensiveness of the delay detection.

[0067] In some optional implementations of this embodiment, the multiple network cards provided in the starting device under test correspond to the multiple network cards provided in the destination device under test in a one-to-one manner, forming multiple network card pairs. Figure 4 , shows a schematic diagram of a network card pair. The first server 401 is equipped with network cards 4011-4018, and the second server 402 is equipped with network cards 4021-4028. Among them, the network cards 4011-4018 correspond to the network cards 4021-4028 one by one, forming 8 network card pairs.

[0068] In this implementation method, the above-mentioned execution entity can execute the above-mentioned second step in the following manner: for multiple network card pairs, through the delay detection intelligent agent in the starting device under test in the device pair, the network card corresponding to the starting device under test in the network card pair is used as the starting network card, and the network card corresponding to the destination device under test in the network card pair is used as the destination network card, and a detection data packet is sent.

[0069] For each device pair, in response to the network device in that device pair having multiple network card pairs, a latency detection agent in the starting device under test in that device pair sends a probe packet to each of the multiple network card pairs, using the network card corresponding to the starting device under test in that device pair as the starting network card and the network card corresponding to the destination device under test in the network card pair as the destination network card. For example, the transmission of probe packets and response packets is required between all eight network card pairs between first server 401 and second server 402.

[0070] In this implementation, a fully interconnected detection method with network card granularity is provided, which further expands the delay detection range and improves the comprehensiveness of delay detection.

[0071] In some optional implementations of this embodiment, the above-mentioned execution entity can perform the above-mentioned second step in the following manner: through the delay detection intelligent agent in the starting device under test in the device pair, using the original socket interface, send a detection data packet using the user datagram protocol to the destination device under test in the device pair.

[0072] As an example, in the starting device under test, first, use the socket() function to create a raw socket. Its socket type is SOCK_RAW, the protocol family is typically AF_INET (indicating the use of IPv4 (Internet Protocol version 4)), and the protocol number is specified as IPPROTO_UDP (indicating the use of UDP (User Datagram Protocol)). Then, use the setsockopt() function to set the IP_HDRINCL option to 1, indicating that the user constructs the IP (Internet Protocol) header, rather than having it automatically populated by the kernel. Then, according to the IP protocol specification, various IP header fields are populated, such as the version number, header length, total length, identifier, time to live, protocol type, source IP address, destination IP address, and the IP header checksum is calculated. Finally, according to the UDP protocol specification, the source port, destination port, length, and checksum fields of the UDP header are populated. The checksum calculation takes into account the pseudo header, UDP header, and UDP data. However, in some cases, it can be set to 0 to indicate that the checksum is not used. Then, the data to be transmitted is added after the UDP header. Finally, the sendto() function is called to send the constructed probe packet through the raw socket, specifying information such as the destination IP address and destination port number.

[0073] On the target device under test, first, use the socket() function to create a raw socket, specifying the protocol number as IPPROTO_UDP, to receive UDP packets. Then, use the bind() function to bind the socket to a specific IP address and port number, allowing it to receive UDP packets sent to that address and port. Next, use the recvfrom() function to receive packets from the raw socket. This function returns the packet's data portion, along with information such as the sender's IP address and port number. The received packet is then processed, such as parsing the IP and UDP headers and extracting the UDP data.

[0074] After receiving the detection data packet from the originating device under test, the destination device under test needs to use a similar processing process as described above to feed back a response data packet to the originating device under test.

[0075] In this implementation, Raw sockets bypass the kernel protocol stack, allowing users to directly manipulate underlying data packets and customize their content and format, reducing transmission latency, improving throughput, and enhancing latency detection accuracy. The UDP protocol's connectionless nature eliminates the need for connection establishment and maintenance, resulting in low overhead and fast transmission. Its real-time nature makes it well-suited for latency detection. The combination of these two improves the flexibility and accuracy of latency detection.

[0076] In some optional implementations of this embodiment, the above-mentioned execution entity can perform the above-mentioned second step in the following manner: through the delay detection intelligent agent in the starting device under test in the device pair, using the original socket interface, based on the preset detection frequency, send a detection data packet using the user datagram protocol to the destination device under test in the device pair.

[0077] The above execution entity can flexibly set the preset detection frequency according to actual conditions.

[0078] As an example, the above-mentioned execution entity can dynamically adjust the preset detection frequency based on network traffic and delay stability. Specifically, first, according to the normal load and delay of the network, an initial detection frequency is pre-set, for example, a detection data packet is sent every 10 seconds as the basic detection frequency. Then, use network monitoring tools or intelligent agents to continuously monitor the traffic conditions and delay data in the network to obtain real-time load and delay information of the current network. When the network traffic is low and the delay is stable, the detection frequency is appropriately lowered to reduce the occupation of network resources. For example, the detection frequency is adjusted to send a detection data packet every 20 seconds. When the network traffic increases or the delay fluctuates: increase the detection frequency accordingly to capture the changes in delay more promptly. For example, increase the detection frequency to send a detection data packet every 5 seconds.

[0079] As another example, the above-mentioned execution entity can dynamically adjust the preset detection frequency based on the delay change rate. Specifically, first, a threshold value of the delay change rate is pre-set. For example, a delay change rate within 10% is considered stable, and a delay change rate exceeding 10% is considered to have a large fluctuation. Then, based on the delay data obtained from several consecutive detections, the delay change rate is calculated to reflect the fluctuation of the delay. When the real-time delay change rate is less than the threshold, the detection frequency is reduced, such as from once every 5 seconds to once every 15 seconds, to save resources. When the real-time delay change rate is greater than or equal to the threshold, the detection frequency is increased, such as from once every 15 seconds to once every 3 seconds, so as to more accurately track the delay change trend.

[0080] In this implementation, a method for sending a detection data packet based on a preset detection frequency is provided, which reduces the occupation of system resources under a high detection scale.

[0081] In some optional implementations of this embodiment, the above-mentioned execution subject can send a detection message in combination with the above-mentioned multiple implementations. For example, for multiple network card pairs, through the delay detection intelligent agent in the starting device under test in the device pair, using the original socket interface, based on the preset detection frequency, with the network card corresponding to the starting device under test in the network card pair as the starting network card, and the network card corresponding to the destination device under test in the network card pair as the destination network card, a detection data packet using the User Datagram Protocol is sent.

[0082] In some optional implementations of this embodiment, the execution entity may perform the time point determination operation in step 202 as follows: determine the time point based on the hardware timestamps of the probe data packet and the response data packet at each network node during transmission.

[0083] Compared with software timestamps, hardware timestamps have the following advantages:

[0084] Higher accuracy: Hardware timestamps use high-precision counters in hardware chips to obtain time, with an accuracy of nanoseconds or even picoseconds. Software timestamps usually rely on the operating system's clock, with an accuracy of only milliseconds or microseconds, which cannot meet the needs of scenarios with extremely high time accuracy requirements.

[0085] More accurate time capture: Hardware timestamps accurately record the moment a packet enters or leaves a network interface, directly marking the moment a packet is sent or received. In contrast, software timestamps are written when the protocol stack processes the packet. They only mark the time when the packet enters or exits the protocol stack, but cannot accurately reflect the actual time of transmission or reception, resulting in certain delays and errors.

[0086] Greater reliability: Hardware timestamps are based on the hardware clock and are relatively independent of software operations. They are not affected by factors such as operating system scheduling, system load, and process priority, and can provide stable and accurate timestamps. Software timestamps are susceptible to interference from these factors, resulting in delays or inaccuracies in timestamp acquisition and processing.

[0087] Low System Resource Consumption: Hardware timestamp generation and processing are primarily performed by hardware, reducing software interrupt overhead and processing burden, thereby lowering system resource usage and improving overall system performance. Software timestamps, on the other hand, require programming within the operating system or application, consuming a certain amount of system resources.

[0088] Easier synchronization: When synchronizing time between different devices, hardware timestamps can achieve high-precision synchronization through protocols, effectively reducing time deviations between devices. Software timestamps, due to their relatively low accuracy and reliability, are more difficult to achieve high synchronization accuracy when synchronizing between devices.

[0089] In this implementation, the time point is determined by using a hardware timestamp, thereby improving the accuracy and efficiency of the determined time point.

[0090] Step 203: Determine the delay data corresponding to the delay detection task according to the time point.

[0091] In this embodiment, the execution entity may determine the delay data corresponding to the delay detection task according to a time point.

[0092] In this embodiment, for the time point corresponding to the transmission process between the device under test and the destination device, the delay corresponding to each transmission link in the transmission process and the total delay corresponding to the transmission process are determined to obtain the delay data corresponding to the device under test; combined with the delay data corresponding to each device under test in the set of devices under test, the delay data corresponding to the delay detection task is determined.

[0093] Continue to refer Figure 3 The delay between the device under test 301 and the destination device 302 is T2-T1, the delay between the destination device 302 and the device under test 301 is T4-T3, the data processing time of the destination device 302 is T3-T2, the round-trip delay between the device under test 301 and the destination device 302 is (T4-T1)-(T3-T2), and the one-way delay is ((T2-T1)-(T3-T4)) / 2.

[0094] In some optional implementations of this embodiment, the execution entity may perform step 203 as follows:

[0095] First, based on the time point, the delay value corresponding to each transmission link in the transmission process is determined; then, based on the preset detection frequency corresponding to the delay detection task, the delay values ​​corresponding to the set of tested devices are aggregated to determine the delay data corresponding to the delay detection task.

[0096] Continue to refer Figure 5 , which shows a schematic diagram of the data aggregation process. The aggregation platform 501 aggregates the data collected by each delay detection agent 502 and stores it in the database 503.

[0097] Specifically, the delay detection agent uploads data to the data collector at regular intervals; the collector writes the received data to the Kafka cluster, and the delay data corresponding to the same or similar detection frequency are aggregated together; the aggregation program pulls data from the Kafka cluster and writes the average delay, maximum delay, and timeout information of each device pair to the aggregation database according to a fixed period.

[0098] In this implementation, a specific method for determining delay data is provided, which improves the real-time performance and accuracy of the delay data.

[0099] Continue to see Figure 6 , Figure 6 6 is a schematic diagram of an application scenario 600 of the agent-based delay determination method according to the present embodiment. User 601 sends a delay detection task to server 603 via terminal device 602. The server first determines the set of tested devices corresponding to the delay detection task; then, for each tested device in the tested device set, the server sends a detection data packet to the destination device corresponding to the tested device in the tested device set through the delay detection agent in the tested device, and determines the time points at which the detection data packet and the response data packet of the detection data packet are at each network node during the transmission process. Specifically, the time points at which the detection data packet is at the network card of the tested device and the network card of the destination device are T1 and T2 respectively, and the time points at which the response data packet is at the network card of the destination device and the network card of the tested device are T3 and T4 respectively; finally, according to the time points, the delay data corresponding to the delay detection task is determined.

[0100] In this embodiment, an agent-based delay determination method is provided. A delay detection agent is deployed in the device under test. The delay detection agent can determine the time points of each network node during the transmission of the detection data packet and the response data packet between the device under test and the destination device to determine the delay data, so that the delay detection method can be applied to various delay detection scenarios and the accuracy of the determined delay data is guaranteed.

[0101] In some optional implementations of this embodiment, the execution entity may further perform the following operations:

[0102] First, according to the received query request, a target detection task and a query time range corresponding to the target detection task are determined from the task set.

[0103] For each latency detection task in the task set, including its corresponding historical latency data, users can send query requests to the above execution entities based on prompt words, selection operations, etc. The query request includes data representing the target detection task and the query time range.

[0104] Then, a heat map is generated and displayed based on the latency data of the target detection task within the query time range.

[0105] As an example, the rows and columns in a heatmap represent the starting and destination devices in a device pair, respectively, and the values ​​in the matrix represent the corresponding latency values. Use the heatmap function to plot the data, selecting an appropriate colormap to represent the magnitude of different latency values.

[0106] In this implementation, users can flexibly query the latency data within a specified time range of the target detection task and quickly understand the latency characteristics based on the heat map, which improves the user's operational flexibility and information acquisition efficiency.

[0107] In some optional implementations of this embodiment, the execution entity may generate a heat map in the following manner:

[0108] First, according to the descending order of multiple nested topological structure levels, the target devices under test in the same topological unit in the topological structure level are arranged adjacently layer by layer to obtain a target sorting method. Among them, the target device under test is the device under test corresponding to the target detection task.

[0109] The above-mentioned execution entities can flexibly set up multiple nested topology levels based on actual conditions. For example, the nested topology levels, from largest to smallest, include cluster, pod (container), and ToR (Top of Rack). The topology units at different topology levels vary in size. For example, the topology unit at the cluster level includes multiple topology units at the pod level.

[0110] Then, a heat map is generated and displayed based on the target ranking method and the latency data of the target detection task within the query time range.

[0111] Continue to refer Figure 7 , which shows a schematic diagram of a heat map of latency data. The heat map shows the latency data corresponding to the device under ToR numbered 04 under Pod numbered 09 in the C3 cluster.

[0112] Through the heat map, you can easily understand the following information:

[0113] 1. Intra-ToR and inter-ToR latency differences: The latency differences within the same switch and between switches can be clearly identified.

[0114] 2. The latency detection data for the same ToR is distributed on both sides of the diagonal line: This reflects the interoperability between devices within the same ToR.

[0115] 3. Symmetry of network transmission: It can be observed whether the delay of the bidirectional link is consistent.

[0116] 4. Server crash or network disconnection: Abnormally high latency or no data can indicate device failure or network connection problems.

[0117] In this implementation, the heat map sorting method (cluster, Pod, ToR) helps users understand the impact of network topology on latency from different dimensions, further improving the efficiency of users' information acquisition based on the heat map.

[0118] In some optional implementations of this embodiment, the above-mentioned execution entity can also perform the following operations: through a large language model, generate network characteristic analysis results based on the delay data of the target detection task within the query time range and the network topology data corresponding to the target detection task.

[0119] Network topology data includes the connection relationships of network devices, link properties (such as bandwidth), etc. This data can be obtained through network configuration files, network management systems, or topology discovery tools.

[0120] Network characteristics include network bottlenecks, symmetry, and stability.

[0121] Network bottleneck analysis mainly includes the following three aspects:

[0122] Latency Distribution Analysis: This approach uses a large language model to analyze the distribution of latency data and calculate statistical metrics such as average, maximum, and minimum latency between different network node pairs. Combined with network topology data, it identifies node pairs or links with high latency, which may be potential locations of network bottlenecks.

[0123] Traffic load analysis: Analyzes the traffic load of each network node and link using network topology data. If a node or link experiences excessive traffic load and high latency, it is likely a network bottleneck. Large language models can predict traffic load trends based on historical and real-time data, identifying potential bottlenecks in advance.

[0124] Path analysis: Analyzes data flow paths based on network topology. For paths with high latency, examines the nodes and links along them to identify potential bottlenecks. Large language models can learn the relationship between paths and latency, identifying critical nodes or links that significantly impact latency.

[0125] Symmetry analysis mainly includes the following three aspects:

[0126] Network structure symmetry analysis: This approach uses large language models to analyze the symmetry of network topology. By analyzing the network structure, symmetric nodes, links, and subnetworks can be identified. For example, some networks may have mirror-symmetric structures, and this symmetry can affect network performance and behavior.

[0127] Delay symmetry analysis: Analyzes delay data in different directions to determine whether network delay is symmetrical. In a symmetrical network, delays in opposite directions should be roughly equal. Asymmetric delays may indicate network issues, such as varying link quality or configuration errors.

[0128] Traffic distribution symmetry analysis: This study examines the distribution of network traffic in different directions and along different paths. Symmetric networks typically have relatively even traffic distribution. Severely asymmetric traffic distribution can impact network performance and stability. Large language models can help identify such asymmetric traffic patterns.

[0129] Network stability analysis mainly includes the following three aspects:

[0130] Latency Fluctuation Analysis: This analyzes latency data fluctuations and calculates statistical metrics such as the standard deviation and variance of latency to assess network stability. Networks with smaller latency fluctuations are relatively more stable. Large language models can learn latency fluctuation patterns and predict trends in network stability.

[0131] Historical data comparison analysis: Compare current latency and network topology data with historical data. If the current latency distribution and network behavior deviate significantly from historical norms, this may indicate that network stability is being compromised. Large language models can build models of normal network behavior based on extensive historical data, allowing for rapid detection of anomalies.

[0132] Fault prediction and diagnosis: Leveraging the fault diagnosis capabilities of large language models, we analyze latency data and network topology data to predict potential network failures. For example, if a node experiences a sudden increase in latency, and that node is in a critical position in the network topology, the network may be at risk of failure. Large language models can identify the root cause of the failure through correlation analysis, helping network administrators take timely action.

[0133] In this implementation, the accuracy and comprehensiveness of network characteristic analysis results are improved by combining a large language model with delay data and network topology data.

[0134] In some optional implementations of this embodiment, the above-mentioned execution entity can also perform the following operations: through a large language model, based on the delay data of the target detection task within the query time range and the network topology data corresponding to the target detection task, determine the transmission path of the data to be transmitted in the network corresponding to the target detection task.

[0135] As an example, first, the network topology structure represented by the network topology data is represented as a graph model, in which the nodes represent network devices and the edges represent links. The weight of the edges can be determined comprehensively based on factors such as delay data and bandwidth. The delay data can be used as an important reference indicator, and the weight of links with smaller bandwidth or larger delay is relatively low. Then, a large language model is used to analyze the distribution and change trend of the delay data to find out the links or nodes with higher delays. At the same time, combined with the network topology data, the key nodes and links that may affect the transmission path are identified, and the delay characteristics of different paths are analyzed, including average delay, delay fluctuation, etc. Then, based on the constructed graph model and the delay characteristic analysis results, a path calculation algorithm is used to calculate multiple candidate paths from the source node to the destination node. Path calculation algorithms include Dijkstra algorithm (an algorithm for solving the shortest path problem in a weighted graph), A The large language model can predict the transmission performance of different paths based on network topology and latency data, thereby helping to select the optimal path. It can also further optimize the path by taking into account factors such as real-time network traffic conditions and link stability. After determining a preliminary transmission path, the performance of the selected path is verified through actual data transmission tests or simulations. Based on the verification results, the large language model is used to analyze potential issues, such as a sudden increase in latency or congestion on a link along the path, and to adjust the transmission path in a timely manner.

[0136] Large language models can mine deep relationships between network nodes, analyze the mutual influence between different pairs of nodes, and identify key nodes that have a significant impact on transmission paths. These key nodes may be core routers, switches, or links prone to congestion in the network.

[0137] Leveraging the large language model's ability to understand network topology, it analyzes the overall structure and layout of the network, identifying bottlenecks or areas prone to transmission delays. Furthermore, combined with latency data, it infers the transmission characteristics of different paths within the network, providing a more comprehensive basis for transmission path selection.

[0138] At the same time, the large language model can learn from and draw upon a wealth of historical network data and transmission path selection cases, summarizing the optimal transmission path patterns under similar network conditions. When faced with new transmission tasks, it can provide reference suggestions based on historical experience, helping to quickly determine the appropriate transmission path.

[0139] In this implementation, a large language model is used in combination with delay data and network topology data to predict the transmission path, providing a feasible transmission path for data transmission, which helps to improve the effectiveness and reliability of data transmission.

[0140] In some optional implementations of this embodiment, the above-mentioned execution subject may further perform the following operations: generating network operation and maintenance results based on the network characteristic analysis results using a large language model.

[0141] For example, the large language model generates an operations and maintenance report based on network characteristic analysis results, including network bottlenecks, symmetry, and stability. Based on these network characteristic analysis results, the large language model provides optimization recommendations. For bottleneck issues, it recommends upgrading link bandwidth or optimizing device configuration; for asymmetry issues, it checks and corrects configuration errors; and to improve stability, it recommends implementing flow control measures.

[0142] In this implementation, the large language model provides network operation and maintenance results based on network feature analysis, which improves the comprehensiveness of the output data, helps operation and maintenance personnel perform operation and maintenance operations, and improves the data processing efficiency of operation and maintenance personnel.

[0143] Continue to refer Figure 8 , shows a schematic process 800 of another embodiment of the agent-based delay determination method according to the present disclosure. In the process 800, the following steps are included:

[0144] Step 801: Determine a set of devices under test corresponding to a delay detection task.

[0145] Step 802 : For a device under test in the device under test set, determine a device pair with the device under test as a starting device under test and another device under test other than the device under test in the device under test set as a destination device under test.

[0146] Step 803: For multiple network card pairs corresponding to the device pair, the delay detection agent in the starting device under test in the device pair sends a detection data packet with the network card corresponding to the starting device under test in the network card pair as the starting network card and the network card corresponding to the destination device under test in the network card pair as the destination network card.

[0147] The multiple network cards provided in the source device under test and the multiple network cards provided in the destination device under test correspond one to one, forming multiple network card pairs.

[0148] Step 804 : determining a time point according to the hardware timestamps of the probe data packet and the response data packet at each network node during the transmission process.

[0149] Step 805: Determine the delay value corresponding to each transmission link in the transmission process according to the time point.

[0150] Step 806 : According to the preset detection frequency corresponding to the delay detection task, the delay values ​​corresponding to the set of devices under test are aggregated to determine the delay data corresponding to the delay detection task.

[0151] Step 807 : According to the received query request, determine the target detection task and the query time range corresponding to the target detection task from the task set.

[0152] Step 808 , according to the order of the multiple nested topological structure levels from large to small, the target tested devices in the same topological unit in the topological structure level are arranged adjacently layer by layer to obtain a target sorting method.

[0153] The target device under test is the device under test corresponding to the target detection task.

[0154] Step 809 : Generate and display a heat map based on the target sorting method and the latency data of the target detection task within the query time range.

[0155] The process 800 of the agent-based delay determination method in this embodiment specifically illustrates the delay detection process and the heat map generation process, which improves the user's information acquisition efficiency and experience while taking into account the versatility and accuracy of delay detection.

[0156] Continue to refer Figure 9 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a time delay determination device based on an intelligent agent. Figure 2 Corresponding to the method embodiment shown, the system can be specifically applied to various electronic devices.

[0157] like Figure 9 As shown, the agent-based delay determination device 900 includes: a set determination unit 901, configured to determine the set of tested devices corresponding to the delay detection task; a detection unit 902, configured to send a detection data packet to the destination device corresponding to the tested device in the tested device set through the delay detection agent in the tested device, and determine the time point at which the detection data packet and the response data packet of the detection data packet are at each network node during the transmission process; a delay determination unit 903, configured to determine the delay data corresponding to the delay detection task according to the time point.

[0158] In some optional implementations of this embodiment, the detection unit 902 is further configured to: for a device under test in the set of devices under test, determine a device pair with the device under test as the starting device under test and other devices under test outside the device under test in the set of devices under test as the destination device under test; and send a detection data packet to the destination device under test in the device pair through the delay detection agent in the starting device under test in the device pair.

[0159] In some optional implementations of this embodiment, the multiple network cards set in the starting device under test and the multiple network cards set in the destination device under test correspond one-to-one to constitute multiple network card pairs; and the detection unit 902 is further configured to: for multiple network card pairs, through the delay detection intelligent agent in the starting device under test in the device pair, with the network card corresponding to the starting device under test in the network card pair as the starting network card, and the network card corresponding to the destination device under test in the network card pair as the destination network card, send a detection data packet.

[0160] In some optional implementations of this embodiment, the detection unit 902 is further configured to: send a detection data packet using the user datagram protocol to the destination device under test in the device pair through the delay detection agent in the starting device under test in the device pair using the raw socket interface.

[0161] In some optional implementations of this embodiment, the detection unit 902 is further configured to: through the delay detection agent in the starting device under test in the device pair, use the original socket interface, based on a preset detection frequency, to send a detection data packet using the user datagram protocol to the destination device under test in the device pair.

[0162] In some optional implementations of this embodiment, the detection unit 902 is further configured to determine the time point according to the hardware timestamps of the probe data packet and the response data packet at each network node during the transmission process.

[0163] In some optional implementations of this embodiment, the delay determination unit 903 is further configured to: determine the delay value corresponding to each transmission link in the transmission process according to the time point; aggregate the delay values ​​corresponding to the set of tested devices according to the preset detection frequency corresponding to the delay detection task, and determine the delay data corresponding to the delay detection task.

[0164] In some optional implementations of this embodiment, the above-mentioned device also includes: a query unit (not shown in the figure), configured to determine the target detection task and the query time range corresponding to the target detection task from the task set based on the received query request; a heat map unit (not shown in the figure), configured to generate and display a heat map based on the delay data of the target detection task within the query time range.

[0165] In some optional implementations of this embodiment, the heat map unit is further configured to: arrange the target tested devices in the same topological unit in the topological structure hierarchy adjacent to each other layer by layer in descending order of multiple nested topological structure hierarchies to obtain a target sorting method, wherein the target tested device is the tested device corresponding to the target detection task; generate and display a heat map based on the target sorting method and the delay data of the target detection task within the query time range.

[0166] In some optional implementations of this embodiment, the above-mentioned device also includes: an analysis unit (not shown in the figure), which is configured to generate a network characteristic analysis result based on the delay data of the target detection task within the query time range and the network topology data corresponding to the target detection task through a large language model.

[0167] In some optional implementations of this embodiment, the above-mentioned device also includes: a path determination unit (not shown in the figure), which is configured to determine the transmission path of the data to be transmitted in the network corresponding to the target detection task through a large language model based on the delay data of the target detection task within the query time range and the network topology data corresponding to the target detection task.

[0168] In some optional implementations of this embodiment, the above-mentioned device also includes: an operation and maintenance unit (not shown in the figure), which is configured to generate network operation and maintenance results based on network characteristic analysis results through a large language model.

[0169] In this embodiment, an agent-based delay determination device is provided. A delay detection agent is deployed in the device under test. The delay detection agent can determine the time points of each network node during the transmission of the detection data packet and the response data packet between the device under test and the destination device to determine the delay data, so that the delay detection method can be applied to various delay detection scenarios and the accuracy of the determined delay data is guaranteed.

[0170] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can implement the agent-based delay determination method described in any of the above embodiments when executing.

[0171] According to an embodiment of the present disclosure, the present disclosure further provides a readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to implement the agent-based delay determination method described in any of the above embodiments when executed.

[0172] An embodiment of the present disclosure provides a computer program product, which, when executed by a processor, can implement the agent-based delay determination method described in any of the above embodiments.

[0173] Figure 10 A schematic block diagram of an example electronic device 1000 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0174] like Figure 10 As shown, electronic device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a random access memory (RAM) 1003. Various programs and data required for the operation of electronic device 1000 may also be stored in RAM 1003. Computing unit 1001, ROM 1002, and RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to bus 1004.

[0175] Multiple components in the electronic device 1000 are connected to the I / O interface 1005, including an input unit 1006, such as a keyboard, a mouse, etc.; an output unit 1007, such as various types of displays, speakers, etc.; a storage unit 1008, such as a magnetic disk, an optical disk, etc.; and a communication unit 1009, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1009 allows the electronic device 1000 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0176] Computing unit 1001 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Computing unit 1001 performs the various methods and processes described above, such as the agent-based latency determination method. For example, in some embodiments, the agent-based latency determination method can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 1008. In some embodiments, part or all of the computer program can be loaded and / or installed onto electronic device 1000 via ROM 1002 and / or communication unit 1009. When the computer program is loaded into RAM 1003 and executed by computing unit 1001, one or more steps of the agent-based latency determination method described above can be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to execute the agent-based delay determination method in any other appropriate manner (eg, by means of firmware).

[0177] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0178] The program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. The program code may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable agent-based delay determination device, such that when the program code is executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0179] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0180] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0181] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0182] A computer system may include a client and a server. The client and server are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, establishing a client-server relationship. The server can be a cloud server, also known as a cloud computing server or cloud host. This server is a host product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and virtual private server (VPS) services. It can also be a server in a distributed system or a server integrated with blockchain.

[0183] According to the technical solution of the embodiments of the present disclosure, a delay determination method and device based on an intelligent agent are provided. A delay detection intelligent agent is deployed in the device under test. The delay detection intelligent agent can determine the time points of each network node in the transmission process of the detection data packet and the response data packet between the device under test and the destination device to determine the delay data, so as to make the delay detection method applicable to various delay detection scenarios and ensure the accuracy of the determined delay data.

[0184] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions provided by this disclosure can be achieved. This is not a limitation herein.

[0185] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for determining delay based on an agent, comprising: Determine the set of devices under test corresponding to the delay detection task; For a device under test in the set of devices under test, a delay detection agent in the device under test generates a probe data packet of a specific format and content according to the detection requirements of the delay detection task, sends the probe data packet to a destination device corresponding to the device under test in the set of devices under test, and determines the time points at which the probe data packet and a response data packet to the probe data packet are at each network node during the transmission process; Determining, according to the time point, delay data corresponding to the delay detection task; Generate a graph model based on the network topology data corresponding to the target detection task in the delay detection task set; Determine the distribution and change trend of the delay data of the target detection task within the query time range through a large language model, and obtain the delay characteristic analysis results; Determining, by means of the large language model, a transmission path for the data to be transmitted in the network according to the graph model, the delay characteristic analysis results, and the real-time traffic status and link stability of the network represented by the network topology data; Through the large language model, network characteristic analysis is performed according to the delay data of the target detection task within the query time range and the network topology data corresponding to the target detection task, and a network characteristic analysis result is generated, wherein the network characteristic analysis includes network bottleneck analysis, symmetry analysis and stability analysis, the network bottleneck analysis includes delay distribution analysis for analyzing the distribution of the delay data, traffic load analysis for analyzing the traffic load of each network node and link in the network, and path analysis for analyzing the data flow path in the network, the symmetry analysis includes network structure symmetry analysis for analyzing the symmetry of the network topology, delay symmetry analysis for analyzing delay data in different directions to determine whether the delay of the network is symmetrical, and traffic distribution symmetry analysis for studying the distribution of network traffic in different directions and paths, the stability analysis includes delay fluctuation analysis for analyzing the fluctuation of delay data, historical data comparison analysis for comparing the delay data and the network topology data with historical data, and fault prediction and diagnosis for predicting network faults; The network operation and maintenance results are generated based on the network characteristic analysis results through the large language model.

2. The method according to claim 1, wherein The step of sending, for a device under test in the set of devices under test, a delay detection agent in the device under test, a detection data packet to a destination device corresponding to the device under test in the set of devices under test, includes: For a device under test in the set of devices under test, determining a device pair with the device under test as a starting device under test and other devices under test other than the device under test in the set of devices under test as destination devices under test; The detection data packet is sent to the destination device under test in the device pair through the delay detection agent in the starting device under test in the device pair.

3. The method according to claim 2, wherein: The multiple network cards provided in the starting device under test correspond to the multiple network cards provided in the destination device under test in a one-to-one manner, forming multiple network card pairs; as well as The sending of the detection data packet to the destination device under test in the device pair through the delay detection agent in the starting device under test in the device pair includes: For multiple network card pairs, the delay detection agent in the starting device under test in the device pair is used to send the detection data packet with the network card corresponding to the starting device under test in the network card pair as the starting network card and the network card corresponding to the destination device under test in the network card pair as the destination network card.

4. The method according to claim 2, wherein: The sending of the detection data packet to the destination device under test in the device pair through the delay detection agent in the starting device under test in the device pair includes: A delay detection agent in the starting device under test in the device pair is used to send a detection data packet using the User Datagram Protocol to the destination device under test in the device pair using a raw socket interface.

5. The method according to claim 4, wherein The method of sending a detection data packet using the User Datagram Protocol to a destination device under test in the device pair by using a raw socket interface through a delay detection agent in the starting device under test in the device pair comprises: The delay detection agent in the starting device under test in the device pair uses a raw socket interface to send a detection data packet using the user datagram protocol to the destination device under test in the device pair based on a preset detection frequency.

6. The method according to claim 1, wherein Determining the time points at which the probe data packet and the response data packet to the probe data packet are located at each network node during the transmission process includes: The time point is determined according to the hardware timestamps of the probe data packet and the response data packet at each network node during the transmission process.

7. The method according to any one of claims 1 to 6, wherein The determining, according to the time point, the delay data corresponding to the delay detection task includes: Determining, based on the time point, a delay value corresponding to each transmission link in the transmission process; According to the preset detection frequency corresponding to the delay detection task, the delay values ​​corresponding to the set of devices under test are aggregated to determine the delay data corresponding to the delay detection task.

8. The method according to claim 7, wherein: Also includes: Determining a target detection task and a query time range corresponding to the target detection task from a task set according to the received query request; Generate and display a heat map based on the latency data of the target detection task within the query time range.

9. The method according to claim 8, wherein Generating and displaying a heat map based on the latency data of the target detection task within the query time range includes: According to the order of multiple nested topological structure levels from large to small, target devices under test in the same topological unit in the topological structure levels are arranged adjacently layer by layer to obtain a target sorting method, wherein the target devices under test are devices under test corresponding to the target detection task; The heat map is generated and displayed according to the target sorting method and the delay data of the target detection task within the query time range.

10. An agent-based delay determination device, comprising: a set determining unit, configured to determine a set of devices under test corresponding to the delay detection task; a detection unit configured to, for a device under test in the set of devices under test, generate, through a delay detection agent in the device under test, a detection data packet of a specific format and content according to a detection requirement of the delay detection task, send the detection data packet to a destination device corresponding to the device under test in the set of devices under test, and determine a time point at which the detection data packet and a response data packet to the detection data packet are at each network node during a transmission process; a delay determination unit, configured to determine delay data corresponding to the delay detection task according to the time point; The path determination unit is configured to: Generate a graph model based on the network topology data corresponding to the target detection task in the delay detection task set; Determine the distribution and change trend of the delay data of the target detection task within the query time range through the large language model to obtain a delay characteristic analysis result; determine the transmission path of the data to be transmitted in the network through the large language model based on the real-time traffic status and link stability of the network represented by the graph model, the delay characteristic analysis result, and the network topology data; an analysis unit configured to perform network characteristic analysis based on the delay data of the target detection task within the query time range and the network topology data corresponding to the target detection task through the large language model, and generate a network characteristic analysis result, wherein the network characteristic analysis includes network bottleneck analysis, symmetry analysis and stability analysis, the network bottleneck analysis includes delay distribution analysis for analyzing the distribution of the delay data, traffic load analysis for analyzing the traffic load of each network node and link in the network, and path analysis for analyzing the data flow path in the network, the symmetry analysis includes network structure symmetry analysis for analyzing the symmetry of the network topology, delay symmetry analysis for analyzing delay data in different directions to determine whether the delay of the network is symmetrical, and traffic distribution symmetry analysis for studying the distribution of network traffic in different directions and paths, the stability analysis includes delay fluctuation analysis for analyzing the fluctuation of delay data, historical data comparison analysis for comparing the delay data and the network topology data with historical data, and fault prediction and diagnosis for predicting network faults; The operation and maintenance unit is configured to generate a network operation and maintenance result according to the network characteristic analysis result through the large language model.

11. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 9.

12. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 9.

13. A computer program product comprising: A computer program which, when executed by a processor, implements the method according to any one of claims 1 to 9.

Images