A task cooperation scheduling method, an edge server and a storage medium

Abstract

The application relates to a task cooperation scheduling method, an edge server and a storage medium. The task cooperation scheduling method comprises the following steps: after receiving a task request, determining whether the predicted completion time length of the edge server is less than a preset time length; if the predicted completion time length of the edge server is less than the preset time length, shortening the predicted completion time length of the edge server; determining whether the predicted completion time length of the edge server after being shortened is greater than the predicted completion time length of a cloud server; if the predicted completion time length of the edge server after being shortened is greater than the predicted completion time length of the cloud server, sending the task request to the cloud server; and if the predicted completion time length of the edge server after being shortened is not greater than the predicted completion time length of the cloud server, executing the task request. Through the task cooperation scheduling method, the calculation efficiency and the resource balance degree can be considered.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a task collaborative scheduling method, an edge server, and a storage medium. Background Technology

[0002] Currently, edge devices (EDs) are responsible for generating and collecting task requests. However, due to limitations in hardware, middleware, heat dissipation, and batteries, EDs need to forward task requests to edge servers (ESs) or cloud servers (CSs) for execution.

[0003] Compared to ES, CS has more powerful computing capabilities and can handle complex computing tasks. As a result, a large number of computing tasks have flocked to CS, leading to a resource imbalance between CS and ES.

[0004] To address this, a strategy of fully offloading resources can be adopted, which prioritizes sending task requests to Elasticsearch (ES) for execution, thereby alleviating the computational pressure on CS and reducing resource waste in ES. However, this method cannot fully utilize the powerful computing capabilities of CS, resulting in low computational efficiency. Summary of the Invention

[0005] Based on this, this application provides a task collaborative scheduling method, an edge server, and a storage medium, which balances computational efficiency and resource balance by partially sinking resources.

[0006] In a first aspect, this application provides a task collaboration scheduling method applied to an edge server. The task collaboration scheduling method includes: after receiving a task request, determining whether the estimated completion time of the edge server is less than a preset time; if it is less than the preset time, shortening the estimated completion time of the edge server; determining whether the shortened estimated completion time of the edge server is greater than the estimated completion time of the cloud server; if so, sending a task request to the cloud server; otherwise, executing the task request.

[0007] In conjunction with the first aspect, in a first possible implementation of the first aspect, the step of shortening the estimated completion time of the edge server includes: obtaining at least one of a proportional constant and an expansion coefficient as a resource sinking parameter, wherein the proportional constant is directly proportional to the resource sinking intensity and the expansion coefficient is inversely proportional to the resource sinking intensity; and reducing the estimated completion time of the edge server according to the at least one resource sinking parameter.

[0008] In conjunction with the first possible implementation of the first aspect, in the second possible implementation of the first aspect, the mathematical expression for the step of reducing the estimated completion time of the edge server based on at least one resource sinking parameter includes: ;in, The estimated completion time for the shortened edge server, The estimated completion time for the edge server. It is a proportional constant. It is a positive integer greater than or equal to 2. Based on execution time, The coefficient of thermal expansion is 1 / 3. Greater than 0 and less than 1.

[0009] In conjunction with the first aspect, in the third possible implementation of the first aspect, before determining whether the expected completion time of the edge server is less than the preset time, the method further includes: after receiving the parameter adjustment instruction, updating at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient, according to the parameter adjustment instruction to adjust the resource sinking intensity; and calculating the preset time based on the updated at least one resource sinking parameter.

[0010] In conjunction with the third possible implementation of the first aspect, in the fourth possible implementation of the first aspect, the mathematical expression for the step of calculating the preset duration based on at least one updated resource sinking parameter includes: ;in, For preset duration, For the updated proportional constant, It is a positive integer greater than or equal to 2. Based on execution time, The updated inflation coefficient. Greater than 0 and less than 1.

[0011] In conjunction with the first aspect, in the fifth possible implementation of the first aspect, after receiving the task request, before determining whether the estimated completion time of the edge server is less than the preset time, the method further includes: obtaining the latest historical samples of the edge server and the cloud server; and predicting the estimated completion time of the edge server and the cloud server based on the historical samples.

[0012] In conjunction with the fifth possible implementation of the first aspect, in the sixth possible implementation of the first aspect, the step of obtaining the latest historical samples of the edge server and the cloud server includes: determining the number of independent variables, wherein the independent variables include at least one of the following: total CPU capacity, CPU capacity, total memory capacity, memory capacity, total bandwidth capacity, and bandwidth capacity; determining the sample size to be collected based on the number of independent variables, and collecting historical samples of the edge server and the cloud server according to the sample size, wherein the sample size is greater than or equal to ten times the number of independent variables.

[0013] Secondly, this application also provides an edge server, which includes: a task scheduling management module, a cloud resource monitoring module, and an edge resource monitoring module; wherein, the task scheduling management module is used to execute the task collaborative scheduling method of the first aspect or any embodiment of the first aspect; the cloud resource monitoring module is used to monitor the resources of the cloud server in real time; and the edge resource monitoring module is used to monitor the resources of the edge server in real time.

[0014] Thirdly, this application also provides an edge server, which includes a processor, a transceiver, and a memory, the processor, transceiver, and memory being connected via a bus; the processor is used to execute multiple instructions; the transceiver is used to interact with other devices; and the memory is used to store multiple instructions, the instructions being adapted to be loaded by the processor and executed as a task cooperative scheduling method as described in the first aspect or any embodiment of the first aspect.

[0015] Fourthly, this application also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor and executing a task cooperative scheduling method as described in the first aspect or any embodiment of the first aspect.

[0016] In summary, this application provides a task collaborative scheduling method, an edge server, and a storage medium. The task collaborative scheduling method, when the estimated completion time of the Elasticsearch (ES) is less than a preset time, first shortens the estimated completion time of the ES, and then executes the task request from either the ES or the CS with the shorter estimated completion time. This achieves partial resource decentralization, which not only fully utilizes the powerful computing capabilities of the CS and reduces task bottlenecks in the ES, but also alleviates the computational pressure on the CS and reduces resource waste in the ES. Therefore, the task collaborative scheduling method provided in this application can balance computational efficiency and resource balance. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a task collaboration scheduling method in one embodiment of this application;

[0018] Figure 2 A flowchart illustrating the task collaborative scheduling method in another embodiment provided in this application;

[0019] Figure 3 A flowchart illustrating the task collaborative scheduling method in another embodiment provided in this application;

[0020] Figure 4 A schematic block diagram of an edge server in one embodiment provided in this application;

[0021] Figure 5 This is a structural block diagram of an edge server in one embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0023] Currently, to address the resource imbalance issue in existing task collaborative scheduling methods, existing technologies employ a fully resource-sunk approach. This means that upon receiving a task request, Elasticsearch (ES) computing resources are prioritized, and only when ES resources are insufficient are Task Controllers (CS) considered. While this method alleviates the computational pressure on CS and reduces resource waste in ES, it fails to fully utilize the powerful computing capabilities of CS. This is because it may cause complex tasks to become stuck in ES, unable to complete in a timely manner, while CS computing resources remain idle.

[0024] To address this issue, this application provides a task collaborative scheduling method that balances computational efficiency and resource balance. This method, by partially sinking resources, not only fully leverages the powerful computing capabilities of the edge server (CS) but also alleviates the computational pressure on the CS and reduces resource waste in the elite server (ES). Specifically, after receiving a task request, the edge server determines whether its estimated completion time is less than a preset time. If it is less, the estimated completion time of the edge server is shortened. The shortened estimated completion time of the edge server is then compared with the estimated completion time of the cloud server. If the shortened time is greater than the estimated completion time of the cloud server, a task request is sent to the cloud server; otherwise, the task request is executed.

[0025] In order to better understand the technical solution of this application, such as Figure 1 As shown, this application proposes an embodiment of a task collaborative scheduling method. Next, this application will use an edge server ES as the execution entity to perform tasks such as... Figure 1 The task collaboration scheduling method shown will be explained in detail below:

[0026] 101: After receiving a task request, determine whether the estimated completion time of the edge server is less than the preset time.

[0027] Upon receiving a task request from the ED, the ES (Executable Component) first determines whether its estimated completion time is less than a preset time. If the estimated completion time is less than the preset time, step 102 is executed to perform resource partial decentralization. The estimated completion time of the ES is the time it expects to complete the task request, and the estimated completion time of the CS (Computer Component) is the time it expects to complete the task request. Both the estimated completion times of the ES and CS can be calculated based on the latest historical samples. Furthermore, if the estimated completion time of the ES is greater than or equal to the preset time, the estimated completion times of the ES and CS are compared. If the estimated completion time of the ES is greater than the estimated completion time of the CS, the task request is sent to the CS for execution, fully utilizing the powerful computing capabilities of the CS. If the estimated completion time of the ES is less than or equal to the estimated completion time of the CS, the task is executed by the ES.

[0028] It should be noted that after receiving a task request from the ED, before determining whether the ES's estimated completion time is less than the preset time, the ES can first determine whether the task request is a batch task. If it is a batch task, it is broken down into the smallest independent task requests and placed into the task queue in sequence. Then, after retrieving the task requests from the task queue, the ES determines whether its task collaboration mode is offline. If it is offline, the task request is executed. If it is not offline, it determines whether the task can be relocated. If it cannot be relocated, the task request is sent to the CS. If it can be relocated, it determines whether the ES meets the resource requirements. If it does not meet the resource requirements, the task request is sent to the CS. If it meets the resource requirements, the aforementioned step of determining whether the ES's estimated completion time is less than the preset time is executed. Specifically, when determining whether resources can be relocated, the ES can be checked to see if it meets the hardware resources required to execute the task request, such as memory, disk resources, software resources, third-party modules, network resources, authentication permissions, etc. If so, the task is considered relocatable.

[0029] It should also be noted that, in cases where the estimated completion time of the Elasticsearch (ES) is longer than that of the Computer Search (CS), this application does not perform resource allocation downsizing. Instead, it directly sends the task request to the end with the shorter estimated completion time between ES and CS. This is to fully leverage the advantages of CS in handling complex tasks and reduce the occurrence of tasks getting stuck in ES. Furthermore, if ES, upon receiving a task request, does not compare it with the preset completion time and directly sends the task request to the end with the shorter estimated completion time between ES and CS, resource allocation will be impossible or minimally effective. This is because the computing power of CS is far greater than that of ES. If the server executing the task request is determined solely by the length of the estimated completion time, then a massive number of task requests would be executed by CS.

[0030] It should also be noted that the preset duration can be any small value set in advance, or a value calculated based on at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient. The proportional constant is directly proportional to the resource sinking force, and the expansion coefficient is inversely proportional to the resource sinking force. Both the proportional constant and the expansion coefficient are adjustable. Furthermore, to determine the preset duration, ES can also calculate the preset duration based on at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient, before determining whether the estimated completion time of ES is less than the preset duration.

[0031] Next, this application will illustrate the steps of calculating the preset duration based on at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient, with examples:

[0032] In the first case, based on the preset duration calculated from the proportional constant, the mathematical expression for this step can be: , For preset duration, It is a proportional constant. It is a positive integer greater than or equal to 2. The base execution time is set to, for example, 100 milliseconds. It can be seen that the proportional constant is directly proportional to the preset duration; the larger the proportional constant, the larger the preset duration, and the greater the resource allocation.

[0033] In the second case, based on the preset duration calculated using the proportional constant and the expansion coefficient, the mathematical expression for this step can be: ,in, For preset duration, It is a proportional constant. Greater than or equal to 2, Based on the execution time, for example, 100 milliseconds. The coefficient of thermal expansion is 1 / 3. It is greater than 0 and less than 1. It can be seen that the expansion coefficient is inversely proportional to the preset duration. The smaller the expansion coefficient, the larger the preset duration, and the greater the resource sinking force.

[0034] 102: If the time is less than the preset time, shorten the estimated completion time of the edge server.

[0035] In cases where the estimated completion time of the ES is less than the preset time, the estimated completion time of the ES is shortened to achieve partial resource sinking. For example, the estimated completion time of the ES can be shortened by any random value, or the estimated completion time of the ES can be shortened by a fixed value, or the estimated completion time of the ES can be reduced according to at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient.

[0036] Next, this application will illustrate with examples the steps of reducing the estimated completion time of ES based on at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient. Specifically:

[0037] In the first case, the mathematical expression for reducing the estimated completion time of ES based on the scaling constant can be: ,in, The estimated completion time for the shortened edge server, The estimated completion time for the edge server. It is a proportional constant. It is a positive integer greater than or equal to 2. Based on the execution time. It can be seen that the proportional constant is inversely proportional to the expected completion time of the shortened ES. The larger the proportional constant, the smaller the expected completion time of the shortened ES, and the greater the degree of resource decentralization.

[0038] In the second case, based on the proportional constant and the expansion coefficient to reduce the expected completion time of ES, the mathematical expression for this step can be: ,in, The estimated completion time for the shortened edge server, The estimated completion time for the edge server. It is a proportional constant. It is a positive integer greater than or equal to 2. Based on execution time, The coefficient of thermal expansion is 1 / 3. It is greater than 0 and less than 1. It can be seen that the expansion coefficient is directly proportional to the expected completion time of the shortened ES. The smaller the expansion coefficient, the shorter the expected completion time of the shortened ES, and the greater the resource sinking force.

[0039] It should be noted that although both the proportional constant and the expansion coefficient can adjust the resource sinking intensity by adjusting the preset duration and / or shortening the expected completion time of ES, the adjustment intensity differs. The proportional constant has a greater adjustment intensity, while the expansion coefficient has a smaller adjustment intensity. When adjusting the preset duration and shortening the expected completion time of ES are both achieved based on the proportional constant and the expansion coefficient, the mathematical expressions for steps 101 and 102 can be... .

[0040] 103: Determine whether the estimated completion time of the scaled-down edge server is greater than the estimated completion time of the cloud server. If so, send a task request to the cloud server; otherwise, execute the task request.

[0041] Specifically, after shortening the estimated completion time of Elasticsearch (ES), the estimated completion time of the reduced edge server will be... Estimated completion time with cloud servers Compare, if Then ES will execute the task request, if If so, the task request will be sent to the CS, and the CS will execute the task request.

[0042] It should be noted that if the estimated completion time of ES is less than the preset time, the powerful computing capabilities of CS cannot be fully utilized if the task request is executed directly by ES. This is because the method only evaluates the estimated completion time of ES. Although the estimated completion time of ES is less than the preset time, the estimated completion time of CS may be much less than that of ES.

[0043] In summary, this application, by partially offloading resources, not only fully leverages the powerful computing capabilities of the Computer System (CS) but also alleviates its computational burden and reduces resource waste in the Elasticsearch System (ES). Furthermore, compared to complex load balancing algorithms, this application is computationally simple, easy to implement, and requires no additional hardware.

[0044] like Figure 2 As shown, this application also proposes another embodiment of the task collaborative scheduling method. Before determining whether the estimated completion time of the edge server is less than the preset time, the ES can, upon receiving a parameter adjustment instruction, update at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient, to adjust the resource sinking intensity. Then, the preset time is calculated based on the updated at least one resource sinking parameter. Next, this application will use the edge server ES as the execution subject to perform tasks such as... Figure 2 The task collaboration scheduling method shown will be explained in detail below:

[0045] 201: Upon receiving the parameter adjustment instruction, update the proportional constant and expansion coefficient according to the parameter adjustment instruction, and calculate the preset duration based on the updated proportional constant and expansion coefficient.

[0046] The parameter adjustment instructions include user-defined proportional constants and / or expansion coefficients. These instructions update the current proportional constants and expansion coefficients by assigning the user-defined proportional constant value to the current proportional constant, and assigning the user-defined expansion coefficient value to the current expansion coefficient, resulting in an updated expansion coefficient. Furthermore, after updating the proportional constants and expansion coefficients, a preset duration is calculated based on these updated values. For example, the mathematical expression for calculating the preset duration can be expressed as: , The updated preset duration, For the updated proportional constant, It is a positive integer greater than or equal to 2. Based on execution time, The updated inflation coefficient. Greater than 0 and less than 1.

[0047] 202: After receiving the task request, determine whether the estimated completion time of the edge server is less than the preset time.

[0048] 203: If the time is less than the preset time, the estimated completion time of the edge server will be shortened according to the updated proportional constant and the expansion coefficient.

[0049] The mathematical expression for the step of shortening the expected completion time of the edge server based on the updated proportional constant and the expansion coefficient can be expressed as: , The estimated completion time for the shortened edge server, The estimated completion time for the edge server. For the updated proportional constant, It is a positive integer greater than or equal to 2. Based on execution time, The updated inflation coefficient. Greater than 0 and less than 1.

[0050] 204: Determine if the estimated completion time of the scaled-down edge server is greater than the estimated completion time of the cloud server. If so, send a task request to the cloud server; otherwise, execute the task request.

[0051] In summary, this embodiment calculates the preset duration, the estimated completion time of ES, and the estimated completion time of CS based on the resource sinking parameters. Furthermore, the resource sinking parameters are adjustable, which allows for adjustment of the resource sinking intensity, thereby improving resource balance.

[0052] like Figure 3 As shown, this application also proposes another embodiment of the task collaborative scheduling method. Before determining whether the estimated completion time of the edge server is less than a preset time, the ES can obtain the latest historical samples of the edge server and the cloud server, and then predict the estimated completion time of the edge server and the cloud server based on the historical samples to achieve dynamic task collaborative scheduling. Next, this application will use the edge server ES as the execution subject to perform tasks such as... Figure 3 The task collaboration scheduling method shown will be explained in detail below:

[0053] 301: Upon receiving a task request, obtain the latest historical samples of the edge server and cloud server, and predict the estimated completion time of the edge server and cloud server based on the historical samples.

[0054] Upon receiving a task request, the Elasticsearch (ES) calculates its estimated completion time in real time based on its latest historical samples, and the CS calculates its estimated completion time based on its latest historical samples. For example, a regression model is built based on the ES's historical samples, and then this model is used to predict the estimated completion time of the task request. Furthermore, based on the statistical method of Events Per Variable (EPV), the sample size of the collected historical samples can be greater than or equal to ten times the number of independent variables to improve the accuracy of the regression model. Additionally, when the sample size of the historical samples is insufficient, the average completion time of the historical samples can be calculated and used as the estimated completion time of the task request. Finally, if the ES or CS has only executed the task request once in the past, the execution time of that task request is directly used as the estimated completion time of the ES or CS.

[0055] Specifically, the steps for obtaining the latest historical samples of edge servers and cloud servers include: determining the number of independent variables, wherein the independent variables include at least one of the following: total CPU capacity, CPU capacity, total memory, memory capacity, total bandwidth, and bandwidth capacity; determining the sample size to be collected based on the number of independent variables, and collecting historical samples of edge servers and cloud servers according to the sample size, wherein the sample size is greater than or equal to ten times the number of independent variables.

[0056] For example, assuming the independent variables include the total number of CPUs, CPU margin, total memory, memory margin, total bandwidth, and bandwidth margin, which is a total of 6 independent variables, then at least the latest 60 samples from ES and the latest 60 samples from CS need to be collected.

[0057] 302: Determine if the estimated completion time of the edge server is less than the preset time.

[0058] 303: If the time is less than the preset time, the estimated completion time of the edge server will be shortened.

[0059] 304: Determine if the estimated completion time of the scaled-down edge server is greater than the estimated completion time of the cloud server. If so, send a task request to the cloud server; otherwise, execute the task request.

[0060] In summary, since the state of server resources is constantly changing, it will experience peaks and troughs due to the execution and unloading of instantaneous tasks. Therefore, this application estimates the expected completion time of ES and CS based on the latest historical samples, which can improve the distortion of instantaneous monitoring data, improve the accuracy of the predicted expected completion time, thereby further alleviating the computational pressure on CS and reducing the waste of ES resources.

[0061] It should be understood that, although Figure 1 , Figure 2 as well as Figure 3 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 , Figure 2 as well as Figure 3 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0062] This application also provides an edge server, such as Figure 4 As shown, the edge server includes a task scheduling management module 410, a cloud resource monitoring module 420, and an edge resource monitoring module 430. The task scheduling management module 410 is used to execute the aforementioned task collaboration scheduling method. The cloud resource monitoring module 420 is used to monitor the resources of the cloud server in real time. The edge resource monitoring module 430 is used to monitor the resources of the edge server in real time. The resources may include, for example, total bandwidth, available bandwidth, bandwidth utilization, CPU utilization, memory utilization, total storage resources, available capacity, and utilization.

[0063] This application also provides an edge server, see [link to application]. Figure 5 As shown in the figure, the edge server in this embodiment may include a processor 510, a transceiver 520, and a memory 530. The processor 510, transceiver 520, and memory 530 are connected via a bus 540. The processor 510 is used to execute multiple instructions; the transceiver 520 is used to interact with other devices; and the memory 530 is used to store multiple instructions adapted to be loaded by the processor 510 and executed using the task collaboration scheduling method described in the above embodiment.

[0064] The processor 510 can be an Electronic Control Unit (ECU), a Central Processing Unit (CPU), a general-purpose processor, a coprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The processor 510 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a 5SP and a microprocessor, etc. In this embodiment, the processor 510 can be a microcontroller. By programming the microcontroller, various control functions can be implemented. The processor has the advantages of powerful computing capabilities and fast processing speed. Specifically: transceiver 520 is used to receive task requests; processor 510 is used to determine whether the estimated completion time of the edge server is less than a preset time; processor 510 is also used to shorten the estimated completion time of the edge server if the estimated completion time of the edge server is less than the preset time; processor 510 is also used to determine whether the estimated completion time of the scaled-down edge server is greater than the estimated completion time of the cloud server; if the estimated completion time of the scaled-down edge server is less than or equal to the estimated completion time of the cloud server, the task request is executed; transceiver 520 is also used to send a task request to the cloud server if the estimated completion time of the scaled-down edge server is greater than the estimated completion time of the cloud server.

[0065] In another possible implementation, the processor 510 is specifically used to: obtain at least one of a proportional constant and an expansion coefficient as a resource sinking parameter, wherein the proportional constant is directly proportional to the resource sinking intensity and the expansion coefficient is inversely proportional to the resource sinking intensity; and reduce the estimated completion time of the edge server based on at least one resource sinking parameter.

[0066] In another feasible approach, the mathematical expression for the step of reducing the expected completion time of the edge server based on at least one resource sinking parameter is: ;in, The estimated completion time for the shortened edge server, The estimated completion time for the edge server. It is a proportional constant. It is a positive integer greater than or equal to 2. Based on execution time, The coefficient of thermal expansion is 1. Greater than 0 and less than 1.

[0067] In another possible implementation, the processor 510 is also used to: upon receiving a parameter adjustment instruction, update at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient, according to the parameter adjustment instruction to adjust the resource sinking intensity; and calculate a preset duration based on the updated at least one resource sinking parameter.

[0068] In another feasible approach, the mathematical expression for the step of calculating the preset duration based on at least one updated resource sinking parameter is: ;in, For preset duration, For the updated proportional constant, It is a positive integer greater than or equal to 2. Based on execution time, The updated inflation coefficient. Greater than 0 and less than 1.

[0069] In another feasible implementation, the processor 510 is also used to: obtain the latest historical samples of the edge server and the cloud server; and predict the expected completion time of the edge server and the cloud server based on the historical samples.

[0070] In another possible implementation, the processor 510 is also used to: determine the number of independent variables, wherein the independent variables include at least one of the following: total CPU capacity, CPU capacity, total memory, memory capacity, total bandwidth, and bandwidth capacity; determine the sample size to be collected based on the number of independent variables, and collect historical samples from the edge server and the cloud server according to the sample size, wherein the sample size is greater than or equal to ten times the number of independent variables.

[0071] In one embodiment, this application also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor and executing the methods in any of the foregoing embodiments. The processor is used to execute the plurality of instructions; the memory is used to store the plurality of instructions, which are loaded by the processor and executed as in the task cooperative scheduling method described in the foregoing embodiments.

[0072] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0073] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A task collaborative scheduling method applied to an edge server, characterized in that, include: Upon receiving a task request, determine whether the estimated completion time of the edge server is less than the preset time. If the estimated completion time of the edge server is greater than or equal to the preset time, then the estimated completion time of the edge server and the cloud server are compared. When the estimated completion time of the edge server is greater than the estimated completion time of the cloud server, the task request is sent to the cloud server for execution; when the estimated completion time of the edge server is less than or equal to the estimated completion time of the cloud server, the task is executed by the edge server. If the time is less than the preset time, the estimated completion time of the edge server is shortened, including: obtaining at least one resource sinking parameter from a proportional constant and an expansion coefficient, wherein the proportional constant is directly proportional to the resource sinking intensity and the expansion coefficient is inversely proportional to the resource sinking intensity; and reducing the estimated completion time of the edge server according to at least one resource sinking parameter. Determine whether the estimated completion time of the scaled-down edge server is greater than the estimated completion time of the cloud server. If so, send the task request to the cloud server; otherwise, execute the task request.

2. The method according to claim 1, characterized in that, The mathematical expression for the step of reducing the estimated completion time of the edge server based on at least one resource sinking parameter is: ； in, The estimated completion time for the shortened edge server, The estimated completion time for the edge server. It is a proportional constant. It is a positive integer greater than or equal to 2. Based on execution time, The coefficient of thermal expansion is 1 / 3. Greater than 0 and less than 1.

3. The method according to claim 1, characterized in that, Before the step of determining whether the estimated completion time of the edge server is less than a preset time, the method further includes: Upon receiving a parameter adjustment instruction, update at least one of the resource sinking parameters, namely the proportional constant and the expansion coefficient, according to the parameter adjustment instruction to adjust the resource sinking intensity. The preset duration is calculated based on at least one of the updated resource sinking parameters.

4. The method according to claim 3, characterized in that, in, The mathematical expression for the step of calculating the preset duration based on at least one updated resource sinking parameter is: ； in, For preset duration, For the updated proportional constant, It is a positive integer greater than or equal to 2. Based on execution time, The updated inflation coefficient. Greater than 0 and less than 1.

5. The method according to claim 1, characterized in that, After receiving the task request, and before determining whether the estimated completion time of the edge server is less than the preset time, the method further includes: Obtain the latest historical samples from the edge server and cloud server; The estimated completion time for the edge server and cloud server is predicted based on the historical samples.

6. The method according to claim 5, characterized in that, The step of obtaining the latest historical samples of the edge server and the cloud server includes: Determine the number of independent variables, wherein the independent variables include at least one of the following: total CPU capacity, CPU margin, total memory, memory margin, total bandwidth, and bandwidth margin. The sample size to be collected is determined based on the number of independent variables, and historical samples of edge servers and cloud servers are collected according to the sample size, wherein the sample size is greater than or equal to ten times the number of independent variables.

7. An edge server, characterized in that, The edge server includes: Task scheduling and management module, cloud resource monitoring module, and edge resource monitoring module; The task scheduling management module is used to execute the task collaboration scheduling method as described in any one of claims 1 to 6; the cloud resource monitoring module is used to monitor the resources of the cloud server in real time; and the edge resource monitoring module is used to monitor the resources of the edge server in real time.

8. An edge server, characterized in that, The edge server includes a processor, a transceiver, and a memory, which are connected via a bus. The processor is used to execute multiple instructions. The transceiver is used to interact with other devices. The memory is used to store the multiple instructions, which are adapted to be loaded by the processor and executed using the task collaborative scheduling method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor and executing the task cooperative scheduling method of any one of claims 1 to 6.

Images