Networking equipment edge computing node transformation method and equipment and computer storage medium

Abstract

The invention discloses a networking device edge computing node transformation method and device and a computer storage medium, and relates to the technical field of edge computing node optimization, and the method comprises the steps that a cloud server obtains all edge computing nodes; dividing all the edge computing nodes into a plurality of edge computing node clusters; two edge computing nodes are dynamically selected from the edge computing node cluster to serve as control ends, the other edge computing nodes serve as execution ends, and the working modes of the execution ends are managed by the control ends; the cloud server distributes a calculation task to a control end in each edge calculation node cluster, and the control end decomposes the distributed calculation task and then dynamically distributes the calculation task to an execution end for execution; according to the method, the robustness can be effectively improved, the utilization rate of edge computing node cluster resources is reduced, idle resources are reduced, and the overall performance is improved.

Description

Technical Field

[0001] The present invention relates to the field of edge computing node optimization technology, and in particular to a method, device, and computer storage medium for modifying edge computing nodes of networked devices. Background Art

[0002] As the scale of IoT devices grows, traditional cloud computing architectures are facing systemic bottlenecks in response latency and energy consumption. In industrial IoT scenarios, 70% of data must be processed within 100 milliseconds, forcing computing to migrate to the edge of the network.

[0003] Edge computing effectively addresses these issues by moving computations to nodes at the edge of the network. However, the distributed nature of edge nodes presents new technical challenges. While this distributed architecture alleviates pressure on the cloud, it also introduces new challenges such as network volatility and high failure rates. Traditional centralized management mechanisms also expose fundamental flaws in dynamic load distribution, node coordination efficiency, and fault tolerance. Summary of the Invention

[0004] The embodiments of the present invention provide a method, device and computer storage medium for modifying edge computing nodes of networked devices, which effectively solve the above-mentioned problems.

[0005] A method for modifying edge computing nodes of networked devices includes the following steps:

[0006] The cloud server obtains all edge computing nodes;

[0007] Divide all edge computing nodes into several edge computing node clusters;

[0008] Two edge computing nodes are dynamically selected from the edge computing node cluster as control terminals, and the remaining edge computing nodes are used as execution terminals. The control terminals manage the working mode of the execution terminals.

[0009] The cloud server assigns computing tasks to the control end in each edge computing node cluster. The control end decomposes the assigned computing tasks and dynamically assigns them to the execution end for execution.

[0010] After execution is completed, each execution end feeds back the results to the control end, and the control end feeds the results back to the cloud server.

[0011] Furthermore, the edge computing node cluster division process includes:

[0012] Obtain the properties of edge computing nodes, including network performance;

[0013] Selecting a number of initial edge computing node cluster cores from the edge computing nodes, wherein the initial edge computing node cluster cores are evenly distributed among the edge computing nodes;

[0014] Each initial edge computing node cluster core uses the first communication delay threshold to screen and absorb edge computing nodes;

[0015] After the initial edge computing node cluster scale has expanded to the set node upper limit, it stops and obtains several edge computing node clusters.

[0016] Furthermore, the remaining edge computing nodes that have not been absorbed use the second communication delay threshold to screen the core of the initial edge computing node cluster. When the screening result is greater than one, they are merged into the edge computing node cluster with low communication delay. When the screening result is one, they are merged into the current edge computing node cluster.

[0017] Furthermore, the control terminal's determination process includes the following steps:

[0018] Use the threshold value to screen each edge computing node in the edge computing node cluster and select the edge computing nodes that meet the conditions as management candidates;

[0019] Use the generated test cases to test the management candidates respectively to obtain the performance information of each management candidate;

[0020] The execution end evaluates the performance information of the management candidates and votes for the management candidates. Based on the number of votes, the management candidates with the first and second highest votes are determined as control ends, among which the control end with the first highest number of votes is the main control end, and the control end with the second highest number of votes is the secondary control end.

[0021] Furthermore, the threshold value is dynamically adjusted according to the computing tasks assigned to the end server, including computing performance threshold and network performance threshold.

[0022] Furthermore, the main control end and the sub-control end manage the working mode of the execution end respectively, wherein the main control end and the sub-control end operate in a collaborative manner, and the main control end and the sub-control end respectively manage part of the execution end. After the cloud server assigns computing tasks to the edge computing node cluster, the main control end and the sub-control end first assign the computing tasks, and then decompose the assigned computing tasks respectively, and then assign them to the execution end for execution. The main control end and the sub-control end respectively obtain the calculation results of the corresponding execution end and transmit them back to the cloud server by the main control end or the sub-control end.

[0023] Furthermore, when an abnormality occurs on one of the main control terminal or the secondary control terminal, the normal other terminal will perform all management tasks and re-determine the control terminal.

[0024] Furthermore, the main control end and the sub-control end are also used to dynamically adjust the computing load of the execution end according to the operating status of the execution end. When it is identified that the computing load of the current execution end exceeds the first load threshold, the execution end with a computing load lower than the second load threshold is identified, and the computing tasks of the execution end with a computing load exceeding the first load threshold are transferred and allocated to the execution end with a computing load lower than the second load threshold. The transfer and allocation is stopped when the computing load of the execution end exceeding the first load threshold is lower than the third load threshold.

[0025] A computer device comprising:

[0026] A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the above method by executing the computer instructions.

[0027] A computer-readable storage medium stores computer instructions, wherein the computer instructions are used to enable a computer to execute the above method.

[0028] The beneficial effects of the above technical solutions provided by the embodiments of the present invention include at least:

[0029] 1. When the edge computing node on the execution side fails, the system ensures the continuous operation of the core functions of the cluster through the real-time task redistribution mechanism on the control side. If the edge computing node on the control side fails, a new edge computing node on the control side is dynamically selected to achieve service availability in the failure scenario, significantly improving the robustness of the system.

[0030] 2. The edge computing nodes on the control side monitor the node load status in real time, dynamically dispatch idle computing resources to high-load nodes, and combine task decomposition to effectively improve the resource utilization of the edge computing node cluster, reduce idle resources, and effectively improve overall performance.

[0031] 3. By building an edge computing node cluster architecture, the control end is elected within the cluster to achieve localized autonomous management, and the decentralized decision-making that originally required direct interaction with the cloud is converted to internal coordination within the cluster, which greatly reduces the frequency of communication with the cloud server. At the same time, the control end node aggregates the calculation results of multiple nodes in the cluster and only transmits refined data to the cloud, significantly reducing the dependence on uplink network bandwidth, thereby effectively alleviating the problem of communication quality degradation caused by network fluctuations.

[0032] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0033] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0035] In the attached figure:

[0036] Figure 1 This is a schematic diagram of the networked device edge computing node system architecture disclosed in an embodiment of the present invention;

[0037] Figure 2 This is a flow chart of the method for modifying edge computing nodes of networked devices disclosed in an embodiment of the present invention;

[0038] Figure 3 A schematic diagram of the process of obtaining an edge computing node cluster disclosed in an embodiment of the present invention;

[0039] Figure 4 A schematic diagram of a process for determining a control terminal according to an embodiment of the present invention;

[0040] Figure 5 This is a schematic diagram of the computing task decomposition efficiency evaluation process disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0042] Figure 1 The edge computing node system architecture of the networked device disclosed in the embodiment of the present invention is shown. In the prior art, the system includes a cloud server, an edge computing node, an edge device and other structures. Among them, the cloud server and the edge computing node are shown in the edge computing node system architecture of the networked device disclosed in the embodiment of the present invention. Since other structures are all prior art, they are not shown in the edge computing node system architecture of the networked device disclosed in the embodiment of the present invention.

[0043] Among them, the technical solution proposed in the embodiment of the present invention is different from the edge computing architecture in the prior art in that this solution divides all edge computing nodes into several edge computing node clusters composed of several edge computing nodes. The number of edge computing nodes in different edge computing node clusters may be the same or different. Depending on the node upper limit and the number of edge computing nodes actually absorbed into the combination, the number of edge computing nodes in different edge computing node clusters may exceed the node upper limit or be lower than the node upper limit or be consistent with the node upper limit. Therefore, the computing performance of different edge computing node clusters may be the same or different.

[0044] Figure 2 The flowchart of the method for modifying an edge computing node of a networked device disclosed in an embodiment of the present invention is shown, and includes the following steps:

[0045] S1, the cloud server obtains all edge computing nodes.

[0046] S2, divide all edge computing nodes into several edge computing node clusters.

[0047] Among them, Figure 3 As shown in the figure, the edge computing node cluster division process includes:

[0048] S21, obtain the attributes of the edge computing node, including network performance.

[0049] S22: Select several initial edge computing node cluster cores from the edge computing nodes, where the initial edge computing node cluster cores are evenly distributed among the edge computing nodes.

[0050] Test the network latency of edge computing nodes. Based on the distribution map of edge computing nodes, select several edge computing nodes whose network latency is within the core latency threshold as the core of the initial edge computing node cluster. The core of the initial edge computing node cluster is evenly distributed among the edge computing nodes to facilitate the construction of the edge computing node cluster.

[0051] S23: Each initial edge computing node cluster core uses the first communication delay threshold to screen and absorb edge computing nodes.

[0052] Obtain communication delay data between the core of the initial edge computing node cluster and adjacent edge computing nodes, use a first communication delay threshold to screen adjacent edge computing nodes, eliminate edge computing nodes that exceed the first communication delay threshold, and retain edge computing nodes that are lower than the first communication delay threshold as edge computing nodes for constructing the initial edge computing node cluster.

[0053] S24, stops after the initial edge computing node cluster scales up to the set node upper limit, and obtains several edge computing node clusters.

[0054] After the number of edge computing nodes in the initial edge computing node cluster reaches the node upper limit, the initial edge computing node cluster core in the initial edge computing node cluster stops using the first communication delay threshold to screen and absorb edge computing nodes. At this time, several edge computing node clusters are formed.

[0055] The stopping condition also includes that the core of the initial edge computing node cluster is less than the edge computing node screened below the first communication delay threshold, at which time the absorption and formation of the edge computing node cluster is stopped.

[0056] The remaining edge computing nodes that have not been absorbed use the second communication delay threshold to screen the core of the initial edge computing node cluster. When the screening result is greater than one, they are merged into the edge computing node cluster with low communication delay. When the screening result is one, they are merged into the current edge computing node cluster.

[0057] When there are still edge computing nodes that have not been absorbed, the edge computing node cluster where the edge computing node with the smallest communication delay is located is selected as the merging object based on the size of the communication delay with any edge computing node in the adjacent edge computing node cluster.

[0058] Through the above operations, the edge computing node cluster is built.

[0059] S3 dynamically selects two edge computing nodes in the edge computing node cluster as control terminals, and the remaining edge computing nodes as execution terminals. The control terminals manage the working mode of the execution terminals.

[0060] like Figure 4 As shown, the control end determination process includes the following steps:

[0061] S31, using the threshold value to screen each edge computing node in the edge computing node cluster, and taking out the edge computing nodes that meet the conditions as management candidates.

[0062] The threshold value is dynamically adjusted according to the computing tasks assigned by the end server, including the computing performance threshold and the network performance threshold. The threshold threshold is used to screen out management candidates from the edge computing node cluster. The performance of the management candidate can meet the work of the control end, so that it can perform additional task decomposition and allocation operations while completing its own tasks, as well as perform management operations on other edge computing nodes.

[0063] S32: Test the management candidates respectively using the generated test cases to obtain performance information of each management candidate.

[0064] The test cases are several computing tasks of random sizes, which are randomly generated by the management candidate with the highest computing resource idle rate. If there are more than two management candidates with the highest and same idle rate, any management candidate is randomly designated to generate the test case. The test cases can also be preset by the cloud server. The management candidates perform the decomposition and feedback operations of the same test case in turn. The cloud server obtains the performance information of the management candidates during the execution of each test case.

[0065] Among them, performance information includes computing task decomposition efficiency evaluation information and communication network delay evaluation information with cloud servers.

[0066] like Figure 5 As shown, the task decomposition efficiency evaluation information is determined based on the preset decomposition quantity and the total time consumed by the decomposition tasks.

[0067] The decomposed quantity is compared with the benchmark quantity. When the decomposed quantity is greater than the benchmark quantity, the evaluation is 4 points; when it is equal to the benchmark quantity, the evaluation is 3 points; when the decomposed quantity is less than the benchmark quantity, the evaluation is insufficient. At this time, if the difference between the benchmark quantity and the decomposed quantity accounts for less than 5% of the benchmark quantity, the evaluation is 2 points; if it is greater than 5%, the evaluation is 1 point.

[0068] The time consumed for decomposition is compared with the benchmark time. When the time consumed for decomposition is less than the benchmark time, it is evaluated as 4 points; when it is equal to the benchmark time, it is evaluated as 3 points; when the time consumed for decomposition is greater than the benchmark time, it is evaluated as time excess. At this time, if the difference between the benchmark time and the time consumed for decomposition accounts for less than 3% of the benchmark time, it is evaluated as 2 points; if it is greater than 3%, it is evaluated as 1 point.

[0069] The task decomposition efficiency is calculated based on the sum of the number of decompositions and the total time spent on the decomposed tasks.

[0070] The network delay time evaluation information for communicating with the cloud server is determined based on the network delay for communicating with the cloud server, wherein when the network delay is lower than the third communication delay threshold, the evaluation is 4 points; when the network delay is lower than the fourth communication delay threshold, the evaluation is 3 points; when the network delay is lower than the fifth communication delay threshold, the evaluation is 2 points; and when the network delay is lower than the sixth communication delay threshold, the evaluation is 1 point.

[0071] S33, the execution end evaluates the performance information of the management candidates and votes for the management candidates. According to the number of votes, the management candidates with the first and second highest votes are determined as control ends, among which the control end with the first highest vote is the main control end, and the control end with the second highest vote is the secondary control end.

[0072] The execution end identifies the performance information of each candidate control end based on the sum of the scores of the decomposition efficiency evaluation information and the evaluation information of the network delay time of communication with the cloud server. In addition, the execution end also calculates the communication delay between itself and each candidate control end and performs a communication delay evaluation. When the communication delay is lower than the seventh communication delay threshold, the evaluation is 1 point, and when the network delay is higher than the seventh communication delay threshold, the evaluation is 0 point. Each execution end votes based on the performance information plus the communication delay evaluation. Among them, the higher the performance information evaluation and the communication delay evaluation score, the stronger the voting ability, and vice versa. Ultimately, two control ends are determined, namely the primary control end and the secondary control end. If the scores of the two control ends are determined to be the same, one of them is randomly designated as the primary control end and the other as the secondary control end.

[0073] In S4, the cloud server assigns computing tasks to the control end in each edge computing node cluster. The control end decomposes the assigned computing tasks and dynamically assigns them to the execution end for execution.

[0074] S5, after the execution is completed, each execution end feeds back the results to the control end, and the control end feeds back the results to the cloud server.

[0075] In the above steps S4 to S5, the main control end and the sub-control end manage the working mode of the execution end respectively, wherein the main control end and the sub-control end operate in a collaborative manner, and the main control end and the sub-control end respectively manage part of the execution end. After the cloud server assigns the computing tasks to the edge computing node cluster, the main control end and the sub-control end first assign the computing tasks, and then decompose the assigned computing tasks respectively, and then assign them to the execution end for execution. The main control end and the sub-control end respectively obtain the calculation results of the corresponding execution end and transmit them back to the cloud server by the main control end or the sub-control end.

[0076] The allocation process includes: the main control end and the sub-control end respectively obtain the operating status of the corresponding execution end, and the operating status of the execution end includes the idle computing resources of the execution end. The main control end or the sub-control end uses the operating status of the corresponding execution ends as the decomposition basis to generate the allocable computing task volume of each edge computing node cluster. The cloud server allocates computing tasks that do not exceed the computing task volume to the edge computing node cluster based on the allocable computing task volume of the edge computing node cluster. The main control end and the sub-control end adaptively decompose the computing tasks according to the operating status of the execution end under their jurisdiction so that they do not exceed the range that the execution end under their jurisdiction can bear. After the decomposition is completed, the decomposed computing tasks are allocated one by one to the corresponding execution end for execution.

[0077] Among them, when an abnormality occurs on one of the main control end or the secondary control end, such as being unable to perform management tasks as the control end, the other normal end will take over and perform all management tasks. At the same time, the control end will be re-determined according to step S3 to achieve the effect of dynamically adjusting and electing the control end. Through the dual-control end management jurisdiction scheme, the failure of the edge computing node cluster caused by the failure of a single management node can be avoided, effectively improving the robustness of the system.

[0078] The main control end and the sub-control end are also used to dynamically adjust the computing load of the execution end according to the operating status of the execution end, wherein the computing load is defined as the computing resources used. For example, if 50% of the computing resources have been used, it is defined as a load of 50%. When it is identified that the computing load of the current execution end exceeds the first load threshold, the execution end with a computing load lower than the second load threshold is identified, and the computing tasks of the execution end with a computing load exceeding the first load threshold are transferred and allocated to the execution end with a computing load lower than the second load threshold. The transfer and allocation is stopped when the computing load of the execution end that exceeds the first load threshold is lower than the third load threshold, thereby avoiding the execution end from being overloaded due to fluctuations in its own tasks, resulting in a decrease in its operating efficiency and the occurrence of computing anomalies. When all execution ends have no idle computing resources to undertake the transfer and allocation, the main control end or the sub-control end will return the computing tasks of the corresponding execution end to the cloud server for reallocation.

[0079] The cloud server is also used to monitor the operating status of the edge computing node cluster, which includes idle computing resources. The computing load of the edge computing node cluster is dynamically adjusted according to the operating status of the edge computing node cluster. When it is identified that the computing load of the current edge computing node cluster exceeds the fourth load threshold, the edge computing node cluster with a computing load lower than the fifth load threshold is identified, and the computing tasks of the edge computing node cluster with a computing load exceeding the fourth load threshold are transferred and allocated to the edge computing node cluster with a computing load lower than the fifth load threshold. The transfer and allocation is stopped when the computing load of the edge computing node cluster that exceeds the fourth load threshold is lower than the sixth load threshold.

[0080] When all edge computing node clusters have no idle computing resources to undertake the transfer allocation, the cloud server temporarily freezes the computing task. When any edge computing node cluster has idle computing resources and can meet the execution of the computing task, the cloud server assigns the computing task to the edge computing node cluster for calculation.

[0081] During the process of cloud server transferring and allocating computing tasks to edge computing node clusters, the edge computing node clusters determine their respective transfer amounts based on the transfer amount of computing tasks, and transfer computing tasks to the execution ends under their respective jurisdictions. The tasks are then transmitted back to the cloud server by the main control end or the sub-control end. The cloud server then allocates the transferred computing tasks to other target edge computing node clusters.

[0082] The main control end and sub-control end in the edge computing node cluster can also determine the computing resources of the edge computing node cluster in real time based on the idle computing resources of the execution end. The cloud server allocates computing tasks according to the real-time computing resources of different edge computing node clusters, effectively improving the resource utilization of the edge computing node cluster, reducing idle resources, and effectively improving overall performance.

[0083] A computer device includes: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the above method by executing the computer instructions.

[0084] A computer-readable storage medium stores computer instructions, wherein the computer instructions are used to enable a computer to execute the above method.

[0085] The above technical solution provided by the embodiment of the present invention ensures the continuous operation of the core functions of the cluster through the real-time task redistribution mechanism of the control end when the edge computing node of the execution end fails; if the edge computing node of the control end fails, a new edge computing node of the control end is dynamically elected to achieve service availability in the failure scenario, which significantly improves the robustness of the system. At the same time, the edge computing node of the control end monitors the node load status in real time and dynamically dispatches idle computing resources to high-load nodes. Combined with task decomposition, it effectively improves the resource utilization rate of the edge computing node cluster, reduces idle resources, and effectively improves the overall performance. By constructing an edge computing node cluster architecture, the control end is elected within the cluster to achieve localized autonomous management, and the decentralized decision-making that originally required direct interaction with the cloud is converted to internal coordination within the cluster, which greatly reduces the frequency of communication with the cloud server. At the same time, the control end node aggregates the calculation results of multiple nodes in the cluster and only transmits refined data to the cloud, which significantly reduces the dependence on the uplink network bandwidth, thereby effectively alleviating the problem of communication quality degradation caused by network fluctuations.

[0086] It should be understood that the specific order or hierarchy of steps in the disclosed processes is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.

[0087] In the foregoing detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that embodiments of the claimed subject matter require more features than are expressly recited in each claim. On the contrary, as reflected in the appended claims, the invention comprises less than all the features of any individual disclosed embodiment. The appended claims are therefore hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

Claims

1. A method for transforming edge computing nodes of networked devices, characterized in that: The steps include: The cloud server obtains all edge computing nodes; Divide all edge computing nodes into several edge computing node clusters; Two edge computing nodes are dynamically selected from the edge computing node cluster as control terminals, and the remaining edge computing nodes are used as execution terminals. The control terminals manage the working mode of the execution terminals. The cloud server assigns computing tasks to the control end in each edge computing node cluster. The control end decomposes the assigned computing tasks and dynamically assigns them to the execution end for execution. After execution is completed, each execution end feeds back the results to the control end, and the control end feeds the results back to the cloud server.

2. The method according to claim 1, wherein The process of dividing edge computing node clusters includes: Obtain the properties of edge computing nodes, including network performance; Selecting a number of initial edge computing node cluster cores from the edge computing nodes, wherein the initial edge computing node cluster cores are evenly distributed among the edge computing nodes; Each initial edge computing node cluster core uses the first communication delay threshold to screen and absorb edge computing nodes; After the initial edge computing node cluster scale has expanded to the set node upper limit, it stops and obtains several edge computing node clusters.

3. The method according to claim 2, wherein The remaining edge computing nodes that have not been absorbed use the second communication delay threshold to screen the core of the initial edge computing node cluster. When the screening result is greater than one, they are merged into the edge computing node cluster with low communication delay. When the screening result is one, they are merged into the current edge computing node cluster.

4. The method according to claim 1, wherein The process of determining the control end includes the following steps: Use the threshold value to screen each edge computing node in the edge computing node cluster and select the edge computing nodes that meet the conditions as management candidates; Use the generated test cases to test the management candidates respectively to obtain the performance information of each management candidate; The execution end evaluates the performance information of the management candidates and votes for the management candidates. Based on the number of votes, the management candidates with the first and second highest votes are determined as control ends, among which the control end with the first highest number of votes is the main control end, and the control end with the second highest number of votes is the secondary control end.

5. The method according to claim 4, wherein The threshold value is dynamically adjusted according to the computing tasks assigned to the end server, including computing performance threshold and network performance threshold.

6. The method according to claim 4, wherein The main control end and the sub-control end manage the working mode of the execution end respectively. The main control end and the sub-control end operate in a collaborative manner. The main control end and the sub-control end manage part of the execution end respectively. After the cloud server assigns computing tasks to the edge computing node cluster, the main control end and the sub-control end first assign the computing tasks, and then decompose the assigned computing tasks respectively, and then assign them to the execution end for execution. The main control end and the sub-control end respectively obtain the calculation results of the corresponding execution end and transmit them back to the cloud server by the main control end or the sub-control end.

7. The method according to claim 6, wherein When an abnormality occurs on either the main control terminal or the secondary control terminal, the normal other terminal will perform all management tasks and re-determine the control terminal.

8. The method according to claim 6, wherein The main control end and the sub-control end are also used to dynamically adjust the computing load of the execution end according to the operating status of the execution end. When it is identified that the computing load of the current execution end exceeds the first load threshold, the execution end with a computing load lower than the second load threshold is identified, and the computing tasks of the execution end with a computing load exceeding the first load threshold are transferred and allocated to the execution end with a computing load lower than the second load threshold. The transfer and allocation is stopped when the computing load of the execution end that exceeds the first load threshold is lower than the third load threshold.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 8 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 8.

Images