Task processing method and device of computing power scheduling mechanism based on container arrangement platform, computer equipment and storage medium

Abstract

The invention relates to a task processing method and device of a computing power scheduling mechanism based on a container arrangement platform, computer equipment, a storage medium and a computer program product. The method comprises the following steps: extracting key attribute information corresponding to heterogeneous computing power resources from a configuration file corresponding to the heterogeneous computing power resources; adjusting component information in the container arrangement platform according to the key attribute information to obtain an adjusted container arrangement platform; deploying the heterogeneous computing power resources in the adjusted container arrangement platform, and taking the heterogeneous computing power resources and the isomorphic computing power resources in the adjusted container arrangement platform as candidate computing power resources in the adjusted container arrangement platform; determining a target computing power resource corresponding to the to-be-processed task according to the processing demand information of the to-be-processed task and the state information of the candidate computing power resource; and allocating the to-be-processed task to the target computing power resource, so that the target computing power resource processes the to-be-processed task. By adopting the method, the task processing efficiency can be improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a task processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product based on a computing power scheduling mechanism of a container orchestration platform. Background Art

[0002] At present, facing the increasing amount of tasks, how to process the tasks quickly is crucial.

[0003] In traditional technology, in the process of task processing, homogeneous computing resources are generally used for processing; however, the type of computing resources utilized by this method is relatively single, which easily leads to a low match between computing resources and tasks, thereby resulting in low task processing efficiency. Summary of the invention

[0004] Based on this, it is necessary to provide a task processing method, device, computer equipment, computer-readable storage medium and computer program product based on a computing power scheduling mechanism of a container orchestration platform that can improve the processing efficiency of tasks in response to the above technical problems.

[0005] In a first aspect, the present application provides a task processing method based on a computing power scheduling mechanism of a container orchestration platform, comprising:

[0006] Obtain the configuration files corresponding to heterogeneous computing resources;

[0007] Extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file;

[0008] According to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform;

[0009] Deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform;

[0010] Obtaining a task to be processed, and screening out a target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources;

[0011] Allocate the tasks to be processed to the target computing resources so that the target computing resources process the tasks to be processed.

[0012] In one embodiment, adjusting component information in the container orchestration platform according to the key attribute information to obtain an adjusted container orchestration platform includes:

[0013] Querying the correspondence between the attribute information and the adjustment method, obtaining the adjustment method corresponding to the key attribute information as the target adjustment method corresponding to the component information in the container orchestration platform;

[0014] According to the target adjustment method, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform.

[0015] In one embodiment, the step of selecting a target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources includes:

[0016] Extracting the computing resource type corresponding to the task to be processed from the processing requirement information;

[0017] Selecting a candidate computing resource corresponding to the computing resource type from among the candidate computing resources as the computing resource to be analyzed;

[0018] According to the status information of each computing power resource to be analyzed, the target computing power resource corresponding to the task to be processed is screened out from each computing power resource to be analyzed.

[0019] In one embodiment, the step of selecting a target computing resource corresponding to the task to be processed from each computing resource to be analyzed according to the status information of each computing resource to be analyzed includes:

[0020] Extracting the current availability information and current load information of each computing resource to be analyzed from the status information of each computing resource to be analyzed;

[0021] From the computing power resources to be analyzed, filter out the computing power resources to be analyzed whose current availability information is in an idle state as standby computing power resources;

[0022] According to the current load information of each of the standby computing resources, the target computing resources corresponding to the task to be processed are screened out from each of the standby computing resources.

[0023] In one embodiment, the step of selecting a target computing resource corresponding to the task to be processed from each of the standby computing resources according to the current load information of each of the standby computing resources includes:

[0024] Inputting the current load information of each of the standby computing power resources into the trained load prediction model to obtain the predicted load information of each of the standby computing power resources;

[0025] The current load information and the predicted load information of each of the backup computing resources are respectively integrated to obtain the target load information of each of the backup computing resources;

[0026] From each of the standby computing power resources, select the standby computing power resource with the smallest target load information as the target computing power resource corresponding to the task to be processed.

[0027] In one embodiment, extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file includes:

[0028] Extracting attribute information corresponding to the heterogeneous computing resources from the configuration file;

[0029] Inputting the attribute information into a plurality of trained importance prediction models to obtain a plurality of predicted importances corresponding to each of the attribute information;

[0030] fusing the multiple predicted importances corresponding to the attribute information respectively to obtain the target importances corresponding to the attribute information;

[0031] From each of the attribute information, the attribute information whose target importance is greater than a preset importance is screened out as the key attribute information.

[0032] In a second aspect, the present application also provides a task processing device based on a computing power scheduling mechanism of a container orchestration platform, including:

[0033] The file acquisition module is used to obtain the configuration files corresponding to the heterogeneous computing resources;

[0034] An information extraction module, used to extract key attribute information corresponding to the heterogeneous computing resources from the configuration file;

[0035] An information adjustment module, used to adjust the component information in the container orchestration platform according to the key attribute information to obtain an adjusted container orchestration platform; the heterogeneous computing power resources are matched with the adjusted container orchestration platform;

[0036] A resource processing module, configured to deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform;

[0037] A resource screening module, used to obtain the task to be processed, and screen out the target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources;

[0038] The task processing module is used to allocate the tasks to be processed to the target computing resources so that the target computing resources can process the tasks to be processed.

[0039] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0040] Obtain the configuration files corresponding to heterogeneous computing resources;

[0041] Extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file;

[0042] According to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform;

[0043] Deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform;

[0044] Obtaining a task to be processed, and screening out a target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources;

[0045] Allocate the tasks to be processed to the target computing resources so that the target computing resources process the tasks to be processed.

[0046] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0047] Obtain the configuration files corresponding to heterogeneous computing resources;

[0048] Extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file;

[0049] According to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform;

[0050] Deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform;

[0051] Obtaining a task to be processed, and screening out a target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources;

[0052] Allocate the tasks to be processed to the target computing resources so that the target computing resources process the tasks to be processed.

[0053] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0054] Obtain the configuration files corresponding to heterogeneous computing resources;

[0055] Extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file;

[0056] According to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform;

[0057] Deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform;

[0058] Obtaining a task to be processed, and screening out a target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources;

[0059] Allocate the tasks to be processed to the target computing resources so that the target computing resources process the tasks to be processed.

[0060] The task processing method, apparatus, computer equipment, storage medium and computer program product based on the computing power scheduling mechanism of the container orchestration platform first obtain the configuration file corresponding to the heterogeneous computing power resources, and extract the key attribute information corresponding to the heterogeneous computing power resources from the configuration file, and then adjust the component information in the container orchestration platform according to the key attribute information to obtain the adjusted container orchestration platform matching the heterogeneous computing power resources. Then, the heterogeneous computing power resources are deployed in the adjusted container orchestration platform, and the heterogeneous computing power resources and homogeneous computing power resources in the adjusted container orchestration platform are used as candidate computing power resources in the adjusted container orchestration platform. Then, the task to be processed is obtained, and according to the processing requirement information of the task to be processed and the status information of the candidate computing power resources, the target computing power resources corresponding to the task to be processed are screened out from each candidate computing power resource. Finally, the task to be processed is allocated to the target computing power resource so that the target computing power resource processes the task to be processed. In this way, during the task processing process, the container orchestration platform is adjusted according to the key attribute information of the heterogeneous computing resources, so that the container orchestration platform can be matched with the heterogeneous computing resources, and the heterogeneous computing resources are deployed in the container orchestration platform, so that when processing tasks, the heterogeneous computing resources in the container orchestration platform can be considered at the same time, avoiding the defect of using homogeneous computing resources for processing in traditional technologies, which easily leads to a low matching degree between computing resources and tasks, and is conducive to improving the processing efficiency of tasks; moreover, during the task processing process, the processing requirement information of the task to be processed and the status information of the candidate computing resources are considered at the same time, so that the most suitable computing resources can be matched for the task more accurately, which is conducive to improving the matching degree between computing resources and tasks, and further improving the processing efficiency of tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 It is a flowchart of a task processing method based on a computing power scheduling mechanism of a container orchestration platform in one embodiment;

[0063] Figure 2 A flowchart of the steps of selecting target computing resources corresponding to the task to be processed in one embodiment;

[0064] Figure 3 It is a flowchart of a task processing method based on a computing power scheduling mechanism of a container orchestration platform in another embodiment;

[0065] Figure 4 It is a structural block diagram of a task processing device based on a computing power scheduling mechanism of a container orchestration platform in one embodiment;

[0066] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0068] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0069] In an exemplary embodiment, Figure 1 As shown, a task processing method based on the computing power scheduling mechanism of the container orchestration platform is provided. This embodiment uses the method applied to the server as an example for illustration; it can be understood that the method can also be applied to the terminal, and can also be applied to the system including the terminal and the server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be but not limited to various personal computers, laptops, smart phones and tablets; the server can be implemented with an independent server or a server cluster composed of multiple servers. In this embodiment, the method includes the following steps:

[0070] Step S101, obtain the configuration file corresponding to the heterogeneous computing resources.

[0071] Among them, heterogeneous computing resources are used to represent a computing resource set that includes computing units of multiple different architectures, such as a computing resource set that includes a CPU (Central Processing Unit), a GPU (Graphics Processing Unit) and an FPGA (Field Programmable Gate Array).

[0072] The configuration file is a file used to define parameters related to heterogeneous computing resources. It should be noted that the format of the configuration file is YAML (a data serialization language).

[0073] Exemplarily, the server obtains identification information corresponding to the heterogeneous computing resources (such as model information corresponding to the heterogeneous computing resources); then, the server selects a configuration file corresponding to the identification information from multiple candidate configuration files as the configuration file corresponding to the heterogeneous computing resources.

[0074] Step S102: extract key attribute information corresponding to the heterogeneous computing resources from the configuration file.

[0075] Among them, key attribute information refers to attribute information whose corresponding importance is greater than the preset importance among the attribute information corresponding to heterogeneous computing resources, such as architecture information, number of CPU cores, CPU main frequency, cache information, etc. corresponding to heterogeneous computing resources.

[0076] Among them, attribute information refers to all information related to the attributes of heterogeneous computing resources.

[0077] Exemplarily, the server extracts attribute information corresponding to heterogeneous computing resources from a configuration file; then, the server determines the importance of the attribute information corresponding to the heterogeneous computing resources, and filters out attribute information whose corresponding importance is greater than a preset importance from the attribute information corresponding to the heterogeneous computing resources as key attribute information corresponding to the heterogeneous computing resources.

[0078] Step S103: According to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform.

[0079] Among them, the container orchestration platform refers to the K8s (Kubernetes, a container orchestration platform) platform.

[0080] The component information is used to represent the scheduler in the container orchestration platform and related components of the container runtime environment.

[0081] The adjusted container orchestration platform is used to represent the container orchestration platform after the component information is adjusted.

[0082] Exemplarily, the server determines the component information that needs to be adjusted from the component information in the container orchestration platform based on the key attribute information, as the component information to be adjusted; then, the server adjusts the component information to be adjusted based on the key attribute information to obtain an adjusted container orchestration platform that matches the heterogeneous computing resources.

[0083] Step S104: deploy heterogeneous computing resources in the adjusted container orchestration platform, and use both heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform.

[0084] Among them, homogeneous computing resources refer to a set of computing resources that include computing units of the same architecture, such as a group of computing resources composed of CPU servers that are all based on the same architecture.

[0085] Among them, candidate computing resources refer to heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform.

[0086] Exemplarily, the server determines the computing resource type of the heterogeneous computing resources; then, the server queries the correspondence between the computing resource type and the deployment method, obtains the deployment method corresponding to the computing resource type of the heterogeneous computing resources, and uses it as the target deployment method corresponding to the heterogeneous computing resources; then, the server deploys the heterogeneous computing resources in the adjusted container orchestration platform according to the target deployment method; then, the server selects the corresponding computing resources with normal health conditions from the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform, and uses these computing resources as candidate computing resources in the adjusted container orchestration platform.

[0087] Step S105, obtaining the task to be processed, and screening out the target computing resources corresponding to the task to be processed from the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources.

[0088] The pending tasks refer to tasks that need to be processed.

[0089] The processing requirement information is used to indicate the specific processing requirements corresponding to the task to be processed, such as whether the task explicitly requires a CPU of a specific architecture or a specific type of accelerator resources.

[0090] The status information includes current availability information and current load information corresponding to the task to be processed.

[0091] The target computing resources are used to represent the computing resources that match the tasks to be processed.

[0092] Exemplarily, the server obtains multiple candidate tasks and determines the task priority of each candidate task; then, the server filters out the candidate tasks with high task priority from each candidate task as tasks to be processed; then, the server filters out the target computing power resources corresponding to the tasks to be processed from each candidate computing power resource based on the processing requirement information of the tasks to be processed and the status information of the candidate computing power resources.

[0093] Step S106, allocating the task to be processed to the target computing resource, so that the target computing resource processes the task to be processed.

[0094] Exemplarily, the server generates a task processing instruction for the task to be processed, and allocates the task to be processed and the task processing instruction to the target computing power resource, so that the target computing power resource processes the task to be processed according to the task processing instruction.

[0095] In the task processing method of the computing power scheduling mechanism based on the container orchestration platform, the configuration file corresponding to the heterogeneous computing power resources is first obtained, and the key attribute information corresponding to the heterogeneous computing power resources is extracted from the configuration file. Then, according to the key attribute information, the component information in the container orchestration platform is adjusted to obtain the adjusted container orchestration platform matching the heterogeneous computing power resources. Then, the heterogeneous computing power resources are deployed in the adjusted container orchestration platform, and the heterogeneous computing power resources and homogeneous computing power resources in the adjusted container orchestration platform are used as candidate computing power resources in the adjusted container orchestration platform. Then, the task to be processed is obtained, and according to the processing requirement information of the task to be processed and the status information of the candidate computing power resources, the target computing power resources corresponding to the task to be processed are screened from each candidate computing power resource. Finally, the task to be processed is allocated to the target computing power resource so that the target computing power resource processes the task to be processed. In this way, during the task processing process, the container orchestration platform is adjusted according to the key attribute information of the heterogeneous computing resources, so that the container orchestration platform can be matched with the heterogeneous computing resources, and the heterogeneous computing resources are deployed in the container orchestration platform, so that when processing tasks, the heterogeneous computing resources in the container orchestration platform can be considered at the same time, avoiding the defect of using homogeneous computing resources for processing in traditional technologies, which easily leads to a low matching degree between computing resources and tasks, and is conducive to improving the processing efficiency of tasks; moreover, during the task processing process, the processing requirement information of the task to be processed and the status information of the candidate computing resources are considered at the same time, so that the most suitable computing resources can be matched for the task more accurately, which is conducive to improving the matching degree between computing resources and tasks, and further improving the processing efficiency of tasks.

[0096] In an exemplary embodiment, the above step S103, adjusting the component information in the container orchestration platform according to the key attribute information to obtain the adjusted container orchestration platform, specifically includes the following contents: querying the correspondence between the attribute information and the adjustment method, obtaining the adjustment method corresponding to the key attribute information as the target adjustment method corresponding to the component information in the container orchestration platform; adjusting the component information in the container orchestration platform according to the target adjustment method to obtain the adjusted container orchestration platform.

[0097] The corresponding relationship between the attribute information and the adjustment method is used to represent the association information between the attribute information and the adjustment method. For example, for the computing power architecture attribute, the corresponding adjustment method is to extend the Kubernetes scheduler algorithm; for the resource performance attribute, the corresponding adjustment method is to develop resource monitoring and management components; for the software dependency attribute, the corresponding adjustment method is to establish a software dependency management mechanism.

[0098] The adjustment method is used to indicate the method of adjusting the component.

[0099] Among them, the target adjustment method is used to represent the adjustment method corresponding to the key attribute information of heterogeneous computing resources.

[0100] Exemplarily, the server queries the correspondence between the attribute information and the adjustment method, obtains the adjustment method corresponding to the key attribute information, and uses the adjustment method as the target adjustment method corresponding to the component information in the container orchestration platform; then, the server determines the component information that needs to be adjusted from the component information in the container orchestration platform based on the key attribute information, as the component information to be adjusted; then, the server adjusts the component information to be adjusted based on the key attribute information, and obtains the adjusted container orchestration platform that matches the heterogeneous computing resources.

[0101] In this embodiment, by querying the correspondence between attribute information and adjustment methods, the corresponding adjustment methods can be accurately determined according to the unique attributes of different types of heterogeneous computing resources, so that heterogeneous computing resources can be seamlessly connected and efficiently run in the container orchestration platform, thereby improving the platform's adaptability to heterogeneous resources.

[0102] In an exemplary embodiment, Figure 2 As shown, the above step S105, based on the processing requirement information of the task to be processed and the status information of the candidate computing resources, selects the target computing resources corresponding to the task to be processed from the candidate computing resources, and specifically includes the following steps:

[0103] Step S201, extracting the computing resource type corresponding to the task to be processed from the processing requirement information.

[0104] Step S202: Filter out candidate computing resources corresponding to the computing resource type from the candidate computing resources as computing resources to be analyzed.

[0105] Step S203, based on the status information of each computing resource to be analyzed, select the target computing resource corresponding to the task to be processed from each computing resource to be analyzed.

[0106] The computing resource type is used to indicate the type of computing resources required to process the task to be processed, such as CPU, GPU, etc.

[0107] The computing resources to be analyzed refer to candidate computing resources that match the computing resource type corresponding to the task to be processed.

[0108] Exemplarily, the server parses the processing demand information to obtain parsed information; then, the server extracts the computing power resource type corresponding to the task to be processed from the parsed information; then, the server determines the computing power resource type of the candidate computing power resources, and based on the computing power resource type of the candidate computing power resources, screens out candidate computing power resources corresponding to the computing power resource type from each candidate computing power resource, and uses the candidate computing power resources as the computing power resources to be analyzed; then, based on the status information of each computing power resource to be analyzed, the server screens out computing power resources to be analyzed whose status information meets the preset status information from each computing power resource to be analyzed, and uses the computing power resources to be analyzed as the target computing power resources corresponding to the task to be processed.

[0109] In this embodiment, by extracting the computing power resource type corresponding to the task to be processed from the processing requirement information, the essential demand for computing power of the task to be processed can be clarified, the selection range of computing power resources is narrowed, and the most suitable computing power resources can be matched for the task to be processed, which is conducive to improving the efficiency of resource scheduling and reducing unnecessary resource evaluations.

[0110] In an exemplary embodiment, the above step S203, based on the status information of each computing power resource to be analyzed, selects the target computing power resources corresponding to the task to be processed from each computing power resource to be analyzed, and specifically includes the following contents: extracting the current availability information and current load information of each computing power resource to be analyzed from the status information of each computing power resource to be analyzed; selecting the computing power resources to be analyzed whose current availability information is idle from each computing power resource to be analyzed as backup computing power resources; selecting the target computing power resources corresponding to the task to be processed from each backup computing power resource based on the current load information of each backup computing power resource.

[0111] Among them, the current availability information is used to indicate the current idle status of the computing resources to be analyzed, including idle state and busy state.

[0112] The idle state is used to indicate that the computing resources to be analyzed are available.

[0113] The busy state is used to indicate that the computing resources to be analyzed are unavailable.

[0114] Among them, the current load information is used to indicate the load value of the computing resource to be analyzed at the current time.

[0115] Among them, the standby computing resources refer to the computing resources to be analyzed whose current availability information is idle.

[0116] Exemplarily, the server performs an integrity check on the status information of each computing power resource to be analyzed, and obtains an inspection result corresponding to the status information of each computing power resource to be analyzed; when the inspection result corresponding to the status information of each computing power resource to be analyzed indicates that the status information of each computing power resource to be analyzed is complete, the server extracts the current availability information and current load information of each computing power resource to be analyzed from the status information of each computing power resource to be analyzed; then, the server judges the current availability information of each computing power resource to be analyzed; then, the server filters out the computing power resources to be analyzed whose current availability information is idle from each computing power resource to be analyzed, and uses the computing power resources to be analyzed as backup computing power resources; then, the server sorts each backup computing power resource according to its current load information, and obtains sorted backup computing power resources; then, the server screens out the backup computing power resource with the smallest current load information from the sorted backup computing power resources, and uses the backup computing power resource as the target computing power resource corresponding to the task to be processed.

[0117] In this embodiment, by performing secondary screening of computing resources based on the current availability information and current load information of the computing resources to be analyzed, it is possible to avoid the situation where some resources are overly busy while other resources are idle, thereby balancing the load between different computing resources, making the load of the entire computing resource system more balanced, and thereby improving the overall utilization efficiency of resources.

[0118] In an exemplary embodiment, based on the current load information of each standby computing power resource, the target computing power resource corresponding to the task to be processed is screened out from each standby computing power resource, which specifically includes the following contents: the current load information of each standby computing power resource is input into a trained load prediction model to obtain the predicted load information of each standby computing power resource; the current load information and the predicted load information of each standby computing power resource are respectively fused to obtain the target load information of each standby computing power resource; and the standby computing power resource with the smallest target load information is screened out from each standby computing power resource as the target computing power resource corresponding to the task to be processed.

[0119] Among them, the load prediction model refers to a network model that can use the current load information of the backup computing power resources to obtain the predicted load information of the backup computing power resources, such as a convolutional neural network model.

[0120] The predicted load information is used to indicate the load value of the computing resource to be analyzed in the future (such as the next hour, the next day, etc.).

[0121] The target load information refers to the load information obtained by fusing the current load information and predicted load information of each backup computing resource.

[0122] Exemplarily, the server performs feature extraction processing on the current load information of each standby computing power resource, respectively, to obtain feature vectors corresponding to the current load information of each standby computing power resource; then, the server inputs the feature vectors corresponding to the current load information of each standby computing power resource into the trained load prediction model, and obtains the predicted load information of each standby computing power resource through the trained load prediction model; then, the server determines the first weight of the current load information of each standby computing power resource, and the second weight of the predicted load information of each standby computing power resource, and performs weighted sum processing on the current load information and the predicted load information of each standby computing power resource according to the first weight and the second weight, respectively, to obtain the target load information of each standby computing power resource; then, the server sorts the standby computing power resources according to the target load information of each standby computing power resource, to obtain the sorted standby computing power resources; then, the server selects the standby computing power resource with the smallest target load information from the sorted standby computing power resources, and uses the standby computing power resource as the target computing power resource corresponding to the task to be processed.

[0123] In this embodiment, the predicted load information of the backup computing power resources is obtained by utilizing the trained load prediction model, thereby taking into account the future change trend of the resource load; moreover, the current load information and the predicted load information are fused, which is equivalent to integrating information of two dimensions, namely, the immediate status and the future trend, to provide a more reliable basis for screening the target computing power resources, which is beneficial to improving the accuracy of determining the target computing power resources.

[0124] In an exemplary embodiment, the above step S102 extracts key attribute information corresponding to heterogeneous computing resources from the configuration file, which specifically includes the following contents: extracting attribute information corresponding to heterogeneous computing resources from the configuration file; inputting the attribute information into multiple trained importance prediction models to obtain multiple predicted importances corresponding to each attribute information; fusing the multiple predicted importances corresponding to each attribute information respectively to obtain the target importance corresponding to each attribute information; and screening out attribute information whose target importance is greater than the preset importance from each attribute information as key attribute information.

[0125] Among them, the importance prediction model refers to a network model that can predict the predicted importance corresponding to attribute information, such as a random forest model.

[0126] The predicted importance is used to represent the predicted value corresponding to the importance of the attribute information.

[0127] The target importance is used to represent the predicted importance obtained by fusing multiple predicted importances corresponding to each attribute information.

[0128] The preset importance refers to a preset importance threshold. It should be noted that the preset importance depends on the situation.

[0129] Exemplarily, the server obtains identification information of heterogeneous computing resources, and extracts attribute information corresponding to the heterogeneous computing resources from a configuration file based on the identification information of the heterogeneous computing resources; then, the server inputs the attribute information into multiple trained importance prediction models, and obtains multiple predicted importances corresponding to each attribute information through each trained importance prediction model; then, the server fuses the multiple predicted importances corresponding to each attribute information according to the model weight of each trained importance prediction model, and obtains the target importance corresponding to each attribute information; then, the server filters out attribute information whose target importance is greater than the preset importance from each attribute information, and uses the attribute information as the key attribute information.

[0130] In this embodiment, by screening attribute information whose target importance is greater than the preset importance as key attribute information, it is possible to effectively highlight those factors that have a significant impact on heterogeneous computing resources from among numerous attributes, thereby enabling more reasonable decisions to be made in the subsequent use and management of resources, which is conducive to improving the accuracy of resource processing.

[0131] In an exemplary embodiment, Figure 3 As shown, another task processing method based on the computing power scheduling mechanism of the container orchestration platform is provided. The method is applied to a server as an example for explanation, and includes the following steps:

[0132] Step S301, obtain the configuration file corresponding to the heterogeneous computing resources.

[0133] Step S302: extract attribute information corresponding to heterogeneous computing resources from the configuration file; input the attribute information into multiple trained importance prediction models to obtain multiple predicted importances corresponding to each attribute information.

[0134] Step S303, respectively fuse the multiple predicted importances corresponding to each attribute information to obtain the target importance corresponding to each attribute information; from each attribute information, select the attribute information whose target importance is greater than the preset importance as the key attribute information.

[0135] Step S304, querying the correspondence between the attribute information and the adjustment method, obtaining the adjustment method corresponding to the key attribute information, and using it as the target adjustment method corresponding to the component information in the container orchestration platform.

[0136] Step S305: adjust the component information in the container orchestration platform according to the target adjustment method to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform.

[0137] Step S306: deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform.

[0138] Step S307, obtain the task to be processed; extract the computing power resource type corresponding to the task to be processed from the processing requirement information; and select the candidate computing power resources corresponding to the computing power resource type from each candidate computing power resource as the computing power resource to be analyzed.

[0139] Step S308, extract the current availability information and current load information of each computing resource to be analyzed from the status information of each computing resource to be analyzed; from each computing resource to be analyzed, select the computing resources to be analyzed whose current availability information is idle as backup computing resources.

[0140] Step S309, respectively input the current load information of each backup computing resource into the trained load prediction model to obtain the predicted load information of each backup computing resource.

[0141] Step S310, respectively fuse the current load information and predicted load information of each backup computing resource to obtain the target load information of each backup computing resource.

[0142] Step S311, from each standby computing power resource, select the standby computing power resource with the smallest target load information as the target computing power resource corresponding to the task to be processed.

[0143] Step S312: Allocate the task to be processed to the target computing resource so that the target computing resource processes the task to be processed.

[0144] In the above-mentioned task processing method based on the computing power scheduling mechanism of the container orchestration platform, during the process of task processing, the container orchestration platform is adjusted according to the key attribute information of the heterogeneous computing power resources, so that the container orchestration platform can be matched with the heterogeneous computing power resources, and the heterogeneous computing power resources are deployed in the container orchestration platform, so that when processing tasks, the heterogeneous computing power resources in the container orchestration platform can be considered at the same time, avoiding the defect of using homogeneous computing power resources for processing in traditional technologies, which easily leads to a low matching degree between computing power resources and tasks, and is conducive to improving the processing efficiency of tasks; moreover, during the process of task processing, the processing requirement information of the task to be processed and the status information of the candidate computing power resources are considered at the same time, so that the most suitable computing power resources can be matched for the task more accurately, which is conducive to improving the matching degree between computing power resources and tasks, and further improving the processing efficiency of tasks.

[0145] In an exemplary embodiment, in order to more clearly illustrate the task processing method of the computing power scheduling mechanism based on the container orchestration platform provided in the embodiment of the present application, the task processing method of the computing power scheduling mechanism based on the container orchestration platform is specifically described below with a specific embodiment. In one embodiment, the present application also provides a computing power scheduling mechanism based on K8s. In the process of task processing, the configuration file corresponding to the heterogeneous computing power resources is first obtained, and the key attribute information corresponding to the heterogeneous computing power resources is extracted from the configuration file. Then, according to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform that matches the heterogeneous computing power resources. Then, the heterogeneous computing power resources are deployed in the adjusted container orchestration platform, and the heterogeneous computing power resources and homogeneous computing power resources in the adjusted container orchestration platform are used as candidate computing power resources in the adjusted container orchestration platform. Then, the task to be processed is obtained, and according to the processing requirement information of the task to be processed and the status information of the candidate computing power resources, the target computing power resources corresponding to the task to be processed are screened from each candidate computing power resource. Finally, the task to be processed is assigned to the target computing power resource so that the target computing power resource processes the task to be processed. Specifically including the following contents:

[0146] The heterogeneous computing resource pooling adaptation framework in this embodiment is built on the K8s platform and mainly includes three core modules: a custom resource definition (CRD, CustomResourceDefinition) module, a custom controller module, and a module for collaboration with K8s native functions. The overall architecture is dynamically adjusted through a monitoring and feedback mechanism to ensure efficient use of resources and stable operation of the system.

[0147] 1. Custom Resource Definition (CRD) module:

[0148] In-depth analysis of the core characteristics of computing resources of different chip architectures (such as GPU, FPGA, etc.), covering multi-dimensional information such as computing performance indicators, instruction set characteristics, memory access modes, and scalability of hardware resources. Based on the analysis results, determine the key attribute parameters that need to be described in detail in custom resources so that K8s can fully and accurately understand and manage heterogeneous computing resources. Among them, computing performance indicators include FLOPS (Floating-point Operations PerSecond), IPC (Instructions Per Cycle), etc.

[0149] By using the CRD mechanism of K8s and writing corresponding YAML-formatted configuration files, multiple custom resource types for describing heterogeneous computing resources can be created. The specific example is as follows: define the name of the resource type as "HeterogeneousCPUResource" and define each key attribute in detail in its specification (spec) field. For example, the "architecture" field is set to clarify the CPU architecture, the "cores" field indicates the number of CPU cores, the "frequency" field indicates the CPU main frequency, and the "cacheSize" field indicates the cache size. At the same time, other related attributes can be added according to actual needs, such as supported instruction set extension features, etc., to more comprehensively describe the CPU resource characteristics.

[0150] Apply the written custom resource configuration file to the K8s cluster and register it through the K8sAPIServer (Kubernetes API server) to make these custom resource types identifiable and manageable objects on the K8s platform, just like native Pod (the smallest management computing unit in Kubernetes), Node (node) and other resources. They can support common operations such as creation, query, update and deletion, laying the foundation for subsequent resource management and adaptation.

[0151] 2. Custom controller module:

[0152] For each custom resource type created, use the appropriate programming language to write the corresponding custom controller code. The core function of the controller is to monitor various event changes of the corresponding custom resources, including but not limited to creation, update, deletion and other operations, and execute the corresponding adaptation logic according to the specific attributes of the resources, so as to achieve seamless connection between heterogeneous computing resources and K8s platform and adaptive operation of applications. Take the "HeterogeneousCPUResource (heterogeneous CPU resource) controller" as an example: when this type of resource is updated (such as GPU memory expansion, FPGA logic unit reconfiguration, etc.), the controller obtains the latest resource attributes and performs adaptation operations.

[0153] Notify the K8s scheduler to update the scheduling policy related to the accelerator resources. For example, according to the newly added computing power of the accelerator, adjust the task priority, allocation rules, and resource quota of the accelerator resources to ensure that the enhanced computing power can be fully utilized in the subsequent task scheduling process and improve the utilization rate of accelerator resources.

[0154] Coordinate with the container orchestration system to correctly configure the connection and calling method between the application and the accelerator when starting the containerized application that needs to use the accelerator. For example, for a GPU accelerator, ensure that the application in the container can correctly load the corresponding GPU driver, set appropriate video memory allocation and CUDA (Compute Unified Device Architecture) environment variables, etc., to ensure that the application can smoothly use the accelerator resources for efficient computing.

[0155] The written custom controller code is compiled and packaged into an executable file, and then deployed to the appropriate node in the K8s cluster through the K8s resource object, so that it can run continuously and stably in the cluster environment, monitor the corresponding custom resource events in real time, and automatically execute the corresponding adaptation logic to achieve dynamic management and adaptation of heterogeneous computing resources.

[0156] 3. Collaborative module with K8s native functions:

[0157] When K8s schedules tasks, the scheduler will comprehensively consider native resources (such as conventional CPUs, memory, etc.) and heterogeneous computing resources defined through custom resource extensions. Based on the specific requirements of the task (for example, whether the task explicitly requires a CPU of a specific architecture or a specific type of accelerator resources) and the current availability and load status of each resource, the task is accurately assigned to the appropriate node, achieving unified scheduling of heterogeneous computing resources and conventional resources, optimizing overall resource utilization efficiency, and ensuring that computing resources of different architectures can play a role in appropriate tasks.

[0158] In the above-mentioned embodiment, during the process of task processing, the container orchestration platform is adjusted according to the key attribute information of the heterogeneous computing power resources, so that the container orchestration platform can be matched with the heterogeneous computing power resources, and the heterogeneous computing power resources are deployed in the container orchestration platform, so that when processing tasks, the heterogeneous computing power resources in the container orchestration platform can be considered at the same time, avoiding the defect of using homogeneous computing power resources for processing in traditional technologies, which easily leads to a low matching degree between computing power resources and tasks, which is beneficial to improving the processing efficiency of tasks; moreover, during the process of task processing, the processing requirement information of the task to be processed and the status information of the candidate computing power resources are considered at the same time, so that the most suitable computing power resources can be matched for the task more accurately, which is beneficial to improving the matching degree between computing power resources and tasks, and further improving the processing efficiency of tasks; at the same time, through custom resource definition, different chips can be accurately characterized. Based on the characteristics of the architecture computing resources, combined with the custom controller to realize the automated adaptation logic, heterogeneous computing resources can be seamlessly connected to the K8s resource pool, and ensure that containerized applications of different architectures can run smoothly, giving full play to the computing advantages of each architecture; based on the K8s native mechanism extension, it naturally cooperates with K8s's existing scheduling, network, storage and other functional modules, without the need for large-scale changes to the K8s core architecture, which is convenient for operation and management by operation and maintenance personnel; it can easily respond to emerging chip architectures or computing resource forms by adding corresponding CRDs and controllers, maintain good adaptability to new technologies, and reduce development and operation and maintenance costs; it realizes the unified scheduling of heterogeneous computing resources and conventional resources, and combined with resource monitoring and dynamic adjustment mechanisms, it can flexibly allocate resources according to resource status and business needs, improve overall resource utilization, effectively meet the diverse and complex business computing needs, and ensure efficient and stable operation of the business.

[0159] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0160] Based on the same inventive concept, the embodiment of the present application also provides a task processing device based on the computing power scheduling mechanism of the container orchestration platform for implementing the task processing method of the computing power scheduling mechanism based on the container orchestration platform involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of one or more task processing devices based on the computing power scheduling mechanism of the container orchestration platform provided below can be referred to the limitations of the task processing method based on the computing power scheduling mechanism of the container orchestration platform above, and will not be repeated here.

[0161] In an exemplary embodiment, Figure 4 As shown, a task processing device based on a computing power scheduling mechanism of a container orchestration platform is provided, including: a file acquisition module 401, an information extraction module 402, an information adjustment module 403, a resource processing module 404, a resource screening module 405 and a task processing module 406, wherein:

[0162] The file acquisition module 401 is used to obtain the configuration files corresponding to the heterogeneous computing resources.

[0163] The information extraction module 402 is used to extract key attribute information corresponding to the heterogeneous computing resources from the configuration file.

[0164] The information adjustment module 403 is used to adjust the component information in the container orchestration platform according to the key attribute information to obtain the adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform.

[0165] The resource processing module 404 is used to deploy heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform.

[0166] The resource screening module 405 is used to obtain the task to be processed, and screen out the target computing resources corresponding to the task to be processed from each candidate computing resource according to the processing requirement information of the task to be processed and the status information of the candidate computing resources.

[0167] The task processing module 406 is used to allocate the tasks to be processed to the target computing resources so that the target computing resources can process the tasks to be processed.

[0168] In an exemplary embodiment, the information adjustment module 403 is also used to query the correspondence between the attribute information and the adjustment method, obtain the adjustment method corresponding to the key attribute information, and use it as the target adjustment method corresponding to the component information in the container orchestration platform; according to the target adjustment method, adjust the component information in the container orchestration platform to obtain the adjusted container orchestration platform.

[0169] In an exemplary embodiment, the resource screening module 405 is also used to extract the computing power resource type corresponding to the task to be processed from the processing requirement information; screen out the candidate computing power resources corresponding to the computing power resource type from each candidate computing power resource as the computing power resources to be analyzed; and screen out the target computing power resources corresponding to the task to be processed from each computing power resource to be analyzed based on the status information of each computing power resource to be analyzed.

[0170] In an exemplary embodiment, the resource screening module 405 is also used to extract the current availability information and current load information of each computing power resource to be analyzed from the status information of each computing power resource to be analyzed; to screen out the computing power resources to be analyzed whose current availability information is idle as backup computing power resources from each computing power resource to be analyzed; and to screen out the target computing power resources corresponding to the task to be processed from each backup computing power resource based on the current load information of each backup computing power resource.

[0171] In an exemplary embodiment, the resource screening module 405 is also used to input the current load information of each standby computing power resource into the trained load prediction model to obtain the predicted load information of each standby computing power resource; fuse the current load information and the predicted load information of each standby computing power resource to obtain the target load information of each standby computing power resource; and screen out the standby computing power resource with the smallest target load information from each standby computing power resource as the target computing power resource corresponding to the task to be processed.

[0172] In an exemplary embodiment, the information extraction module 402 is also used to extract attribute information corresponding to heterogeneous computing resources from a configuration file; input the attribute information into multiple trained importance prediction models to obtain multiple predicted importances corresponding to each attribute information; fuse the multiple predicted importances corresponding to each attribute information to obtain a target importance corresponding to each attribute information; and filter out attribute information whose target importance is greater than a preset importance from each attribute information as key attribute information.

[0173] Each module in the task processing device of the computing power scheduling mechanism based on the container orchestration platform can be implemented in whole or in part by software, hardware and their combination. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0174] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 5As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as key attribute information and component information. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a task processing method based on a computing power scheduling mechanism of a container orchestration platform is implemented.

[0175] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0176] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0177] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0178] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0179] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0180] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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.

[0181] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A task processing method based on a computing power scheduling mechanism of a container orchestration platform, characterized in that: The method comprises: Obtain the configuration files corresponding to heterogeneous computing resources; Extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file; According to the key attribute information, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform; the heterogeneous computing resources are matched with the adjusted container orchestration platform; Deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform; Acquire the task to be processed, and select the target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources; Allocate the tasks to be processed to the target computing resources so that the target computing resources process the tasks to be processed.

2. The method according to claim 1, characterized in that The step of adjusting component information in the container orchestration platform according to the key attribute information to obtain an adjusted container orchestration platform includes: Querying the correspondence between the attribute information and the adjustment method, obtaining the adjustment method corresponding to the key attribute information as the target adjustment method corresponding to the component information in the container orchestration platform; According to the target adjustment method, the component information in the container orchestration platform is adjusted to obtain an adjusted container orchestration platform.

3. The method according to claim 1, characterized in that The step of selecting a target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources includes: Extracting the computing resource type corresponding to the task to be processed from the processing requirement information; Selecting a candidate computing resource corresponding to the computing resource type from among the candidate computing resources as the computing resource to be analyzed; According to the status information of each computing power resource to be analyzed, the target computing power resource corresponding to the task to be processed is screened out from each computing power resource to be analyzed.

4. The method according to claim 3, characterized in that: The step of selecting a target computing resource corresponding to the task to be processed from the computing resources to be analyzed according to the status information of the computing resources to be analyzed includes: Extracting the current availability information and current load information of each computing resource to be analyzed from the status information of each computing resource to be analyzed; From the computing power resources to be analyzed, filter out the computing power resources to be analyzed whose current availability information is in an idle state as standby computing power resources; According to the current load information of each of the standby computing resources, the target computing resources corresponding to the task to be processed are screened out from each of the standby computing resources.

5. The method according to claim 4, characterized in that The step of selecting a target computing resource corresponding to the task to be processed from each of the standby computing resources according to the current load information of each of the standby computing resources comprises: Inputting the current load information of each of the standby computing power resources into the trained load prediction model respectively to obtain the predicted load information of each of the standby computing power resources; The current load information and the predicted load information of each of the backup computing resources are respectively integrated to obtain the target load information of each of the backup computing resources; From each of the standby computing power resources, select the standby computing power resource with the smallest target load information as the target computing power resource corresponding to the task to be processed.

6. The method according to any one of claims 1 to 5, characterized in that: Extracting key attribute information corresponding to the heterogeneous computing resources from the configuration file includes: Extracting attribute information corresponding to the heterogeneous computing resources from the configuration file; Inputting the attribute information into a plurality of trained importance prediction models to obtain a plurality of predicted importances corresponding to each of the attribute information; fusing the multiple predicted importances corresponding to the attribute information respectively to obtain the target importances corresponding to the attribute information; From each of the attribute information, the attribute information whose target importance is greater than a preset importance is screened out as the key attribute information.

7. A task processing device based on a computing power scheduling mechanism of a container orchestration platform, characterized in that: The device comprises: The file acquisition module is used to obtain the configuration files corresponding to the heterogeneous computing resources; An information extraction module, used to extract key attribute information corresponding to the heterogeneous computing resources from the configuration file; An information adjustment module, used to adjust the component information in the container orchestration platform according to the key attribute information to obtain an adjusted container orchestration platform; the heterogeneous computing power resources are matched with the adjusted container orchestration platform; A resource processing module, configured to deploy the heterogeneous computing resources in the adjusted container orchestration platform, and use the heterogeneous computing resources and homogeneous computing resources in the adjusted container orchestration platform as candidate computing resources in the adjusted container orchestration platform; A resource screening module, used to obtain the task to be processed, and screen out the target computing resource corresponding to the task to be processed from each of the candidate computing resources according to the processing requirement information of the task to be processed and the status information of the candidate computing resources; The task processing module is used to allocate the tasks to be processed to the target computing resources so that the target computing resources can process the tasks to be processed.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images