Computing power resource scheduling method and related equipment

Abstract

The invention provides a computing power resource scheduling method and related equipment, which can be used in the field of data processing. The method comprises the following steps: receiving a resource scheduling request sent by first computing power demand equipment, and determining an equipment identifier of the first computing power demand equipment according to the resource scheduling request; determining a first similarity between the first computing power demand device and each second computing power demand device in response to the fact that the device identifier indicates that the first computing power demand device is a device which is not allocated with computing power resources; determining a second similarity in the first similarities, and determining allocated first computing power resources of second computing power demand equipment corresponding to the second similarity as target computing power resources of the first computing power demand equipment; and sending a resource allocation instruction to a first computing power supply system according to the target computing power resource. According to the method, the scheduling efficiency of the computing power resources of the computing power demand equipment is improved.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular, to a computing power resource scheduling method and related devices. Background Art

[0002] With the update, replacement, iteration, and upgrade of devices, it is necessary to configure computing power resources for these devices so that the devices can operate normally after the update, replacement, iteration, and upgrade.

[0003] In an exemplary technology, the user calculates the computing power resources required for the device operation based on the relevant attributes in the device, and then applies to the scheduling system for scheduling the computing power resources to supply the device.

[0004] However, the calculation time of the computing power resources required for the device by manual is relatively long, resulting in low scheduling efficiency of the computing power resources of the device. Summary of the Invention

[0005] This application provides a computing power resource scheduling method and related devices, which solves the problem of the scheduling efficiency of the computing power resources of the device.

[0006] In a first aspect, this application provides a computing power resource scheduling method, including:

[0007] Receiving a resource scheduling request sent by a first computing power demand device, and determining the device identifier of the first computing power demand device according to the resource scheduling request;

[0008] In response to the device identifier indicating that the first computing power demand device is a device without allocated computing power resources, determining a first similarity between the first computing power demand device and each second computing power demand device, where the second computing power demand device is used to indicate a device with allocated computing power resources;

[0009] Determining a second similarity among the first similarities, and determining the first allocated computing power resources of the second computing power demand device corresponding to the second similarity as the target computing power resources of the first computing power demand device, where the second similarity is the first similarity greater than a preset similarity;

[0010] According to the target computing power resources, sending a resource allocation instruction to a first computing power supply system for the computing power supply system to provide the target computing power resources to the first computing power demand device.

[0011] In some embodiments, the step of sending a first resource allocation instruction to a first computing power supply system according to the target computing power resources includes:

[0012] Determining the system where the first computing power demand device is deployed as the first computing power supply system;

[0013] Obtain the first computing power parameter of the single-core processing unit in the first computing power supply system, and determine the first ratio between the target computing power resource and the first computing power parameter;

[0014] According to the first ratio, determine the first number of cores of the processing unit required for the first computing power supply system to provide the target computing power resource, and generate a resource allocation instruction according to the first number of cores;

[0015] Send the resource allocation instruction to the first computing power supply system.

[0016] In some embodiments, before obtaining the first computing power parameter of the single-core processing unit in the first computing power supply system, it further includes:

[0017] Obtain the total number of cores of the processing units of each second computing power supply system, and the first computing power supply system is any one of the second computing power supply systems;

[0018] Use a variety of test cases to test the processing units of the second computing power supply system respectively, and obtain the computing power sub-parameters of the processing units of the second computing power supply system in each test case;

[0019] According to the sum of the computing power sub-parameters of the processing units of the second computing power supply system, determine the total computing power parameter of the second computing power supply system;

[0020] According to the second ratio between the total computing power parameter of the second computing power supply system and the total number of cores, determine the second computing power parameter of the single-core processing unit in the second computing power supply system, and store the second computing power parameter in association with the system identifier of the second computing power supply system.

[0021] In some embodiments, the using a variety of test cases to test the processing units of the second computing power supply system respectively includes:

[0022] In response to the processing unit of the second computing power supply system being an image processing unit, determine the label of the image processing unit, and the label is used to indicate the use of the image processing unit in the second computing power supply system;

[0023] Test the image processing unit respectively according to each test case matched by the label.

[0024] In some embodiments, the determining the allocated first computing power resource of the second computing power demand device corresponding to the second similarity includes:

[0025] Obtain the target resource utilization rate of the second computing power demand device, and determine the current computing power resource of the second computing power demand device;

[0026] Modify the current computing power resources according to the target resource utilization rate to obtain the first computing power resources.

[0027] In some embodiments, obtaining the target resource utilization rate of the second computing power demand device includes:

[0028] Determine each third computing power supply system in which the second computing power demand device is deployed, and determine the first application center set in each third computing power supply system;

[0029] Determine a second application center among the first application centers, where the second application center is the first application center with the largest service diversion ratio;

[0030] Determine the target resource utilization rate of the second computing power demand device according to a third ratio between the computing power resources of the second application center and the service diversion ratio of the second application center.

[0031] In some embodiments, determining the target resource utilization rate of the second computing power demand device according to a third ratio between the computing power resources of the second application center and the service diversion ratio of the second application center includes:

[0032] Determine the minimum number of first application centers that need to run normally to process the services of the second computing power demand device;

[0033] Determine the resource utilization rate of each first application center as the target resource utilization rate according to the third ratio and the minimum number.

[0034] In some embodiments, modifying the current computing power resources according to the target resource utilization rate to obtain the allocated computing power resources includes:

[0035] Obtain the utilization rate interval associated with the second computing power demand device;

[0036] In response to the target resource utilization rate being less than the lower limit value of the utilization rate interval, reduce the current computing power resources to obtain the first computing power resources;

[0037] In response to the target resource utilization rate being greater than the upper limit value of the utilization rate interval, increase the current computing power resources to obtain the first computing power resources.

[0038] In some embodiments, after determining the device identifier of the first computing power demand device according to the resource scheduling request, it further includes:

[0039] In response to the device identifier indicating that the first computing power demand device is a device with allocated computing power resources, determine a third computing power supply system designated by the first computing power demand device to provide computing power resources according to the resource scheduling request;

[0040] Obtain the allocated second computing power resources of the first computing power demand device;

[0041] According to the second computing power resources, send a resource scheduling instruction to the third computing power supply system for the third computing power supply system to provide the second computing power resources to the first computing power demand device.

[0042] In some embodiments, the obtaining the allocated second computing power resources of the first computing power demand device includes:

[0043] Determine the third computing power parameter of the single-core processing unit in the current computing power supply system, where the current computing power supply system is the computing power supply system currently providing computing power resources to the first computing power demand device;

[0044] Determine the second core number of the processing unit in the current computing power supply system that provides computing power resources to the first computing power demand device;

[0045] Determine the allocated second computing power resources of the first computing power demand device according to the product between the second core number and the third computing power parameter.

[0046] In some embodiments, the sending a resource scheduling instruction to the third computing power supply system according to the second computing power resources includes:

[0047] Obtain the fourth computing power parameter of the single-core processing unit in the third computing power supply system and determine the fourth ratio between the second computing power resources and the fourth computing power parameter;

[0048] According to the fourth ratio, determine the third core number of the processing unit required for the third computing power supply system to provide the second computing power resources, and generate a resource scheduling instruction according to the third core number;

[0049] Send the resource scheduling instruction to the third computing power supply system.

[0050] In some embodiments, the determining the first similarity between the first computing power demand device and each second computing power demand device includes:

[0051] Obtain the first portrait features of the first computing power demand device;

[0052] Input each of the first portrait features into a resource matching model to obtain the first similarity between the first computing power demand device and each second computing power demand device output by the resource matching model.

[0053] In some embodiments, before inputting each of the first portrait features into the resource matching model, the method further includes:

[0054] Obtaining second portrait features of each second computing power demand device, and constructing a training sample corresponding to the second computing power demand device according to the second portrait features corresponding to the second computing power demand device;

[0055] Training a preset model according to each of the training samples to obtain a resource matching model.

[0056] In a second aspect, the present application provides a computing power resource scheduling system, including:

[0057] A receiving module, configured to receive a resource scheduling request sent by a first computing power demand device, and determine a device identifier of the first computing power demand device according to the resource scheduling request;

[0058] A first determination module, configured to determine a first similarity between the first computing power demand device and each second computing power demand device in response to the device identifier indicating that the first computing power demand device is a device without allocated computing power resources, where the second computing power demand device is used to indicate a device with allocated computing power resources;

[0059] A second determination module, configured to determine a second similarity among the first similarities, and determine the allocated first computing power resources of the second computing power demand device corresponding to the second similarity as the target computing power resources of the first computing power demand device, where the second similarity is the first similarity greater than a preset similarity;

[0060] A sending module, configured to send a resource allocation instruction to a first computing power supply system according to the target computing power resources, so that the computing power supply system provides the target computing power resources to the first computing power demand device.

[0061] In a third aspect, the present application provides an electronic device, including: a processor, and a memory and a communication interface communicatively connected to the processor;

[0062] The communication interface is used for communicating with other communication devices;

[0063] The memory is used for storing computer execution instructions;

[0064] The processor is configured to execute the computer execution instructions stored in the memory to implement the computing power resource scheduling method provided in the first aspect.

[0065] Fourthly, the present application provides a computer-readable storage medium storing computer-executable instructions, which when executed by a processor, implement the computing power resource scheduling method provided in the first aspect.

[0066] Fifthly, the present application provides a computer program product including a computer program, which when executed by a processor, implement the computing power resource scheduling method provided in the first aspect.

[0067] For the computing power resource scheduling method and related devices provided in the present application, when the first computing power demand device applying for resource scheduling is a device without allocated computing power resources, a second computing power demand device similar to the first computing power demand device is determined among multiple second computing power demand devices with allocated computing power resources, so as to configure computing power resources for the first computing power demand device through the computing power resources of the similar second computing power demand device, without the need for manual calculation of the computing power resources required by the first computing power demand device, reducing the calculation duration of the computing power resources required by the first computing power demand device and improving the scheduling efficiency of the computing power resources of the computing power demand device. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings herein are incorporated into and constitute a part of this specification, showing embodiments consistent with the present application and, together with the specification, are used to explain the principles of the present application.

[0069] Figure 1 It is a schematic diagram of the scenario of the computing power resource scheduling method related to the present application;

[0070] Figure 2 It is a schematic flow chart of the steps of the computing power resource scheduling method provided in the embodiments of the present application Figure 1 ;

[0071] Figure 3 It is a schematic flow chart of the steps of the computing power resource scheduling method provided in the embodiments of the present application Figure 2 ;

[0072] Figure 4 It is a schematic flow chart of the steps of the computing power resource scheduling method provided in the embodiments of the present application Figure 3 ;

[0073] Figure 5 It is a schematic flow chart of the steps of the computing power resource scheduling method provided in the embodiments of the present application Figure 4 ;

[0074] Figure 6 It is a schematic flow chart of the steps of the computing power resource scheduling method provided in the embodiments of the present application Figure 5 ;

[0075] Figure 7Schematic diagram of the steps of the computing power resource scheduling method provided in the embodiments of the present application Figure 6 ;

[0076] Figure 8 Schematic diagram of program modules of a computing power resource scheduling system provided in the embodiments of the present application;

[0077] Figure 9 Schematic diagram of the hardware structure of an electronic device provided in the embodiments of the present application.

[0078] Through the above drawings, specific embodiments of the present application have been shown, and there will be more detailed descriptions hereinafter. These drawings and textual descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. Detailed implementation manners

[0079] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application. In addition, although the disclosure in the present application is introduced according to exemplary one or several examples, it should be understood that each aspect of these disclosures can also be separately constituted as a complete implementation manner.

[0080] It should be noted that the brief descriptions of the terms in the present application are only for the convenience of understanding the subsequent described implementation manners, rather than intending to limit the implementation manners of the present application. Unless otherwise specified, these terms should be understood in their ordinary and general meanings.

[0081] In addition, the terms "including" and "having" and any variations thereof are intended to cover but not exclude inclusion. For example, a product or device including a series of components does not necessarily have to be limited to those components clearly listed, but may include other components not clearly listed or inherent to these products or devices.

[0082] The term "module" used in the embodiments of the present application refers to any known or later developed combination of hardware, software, firmware, artificial intelligence, fuzzy logic, or hardware and / or software code that can perform functions related to that element.

[0083] 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 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. Moreover, the collection, use, and processing of relevant data need to comply with relevant laws, regulations, and standards, and corresponding operation entrances are provided for users to choose to authorize or refuse.

[0084] With the update, replacement, iteration, and upgrade of devices, it is necessary to configure computing power resources for these devices to ensure their normal operation after the update, replacement, iteration, and upgrade.

[0085] In exemplary technologies, the user calculates the computing power resources required for device operation based on relevant attributes in the device and then applies to the scheduling system for scheduling computing power resources to supply the device.

[0086] The inventors of this application found that the calculation time for the computing power resources required by the device manually is relatively long, resulting in low scheduling efficiency of the computing power resources of the device.

[0087] Therefore, the inventors of this application thought that when the first computing power demand device applying for resource scheduling is a device without allocated computing power resources, determine the second computing power demand device similar to the first computing power demand device among multiple second computing power demand devices with allocated computing power resources. Thus, configure computing power resources for the first computing power demand device through the computing power resources of the similar second computing power demand device, without the need for manual calculation of the computing power resources required by the first computing power demand device, reducing the calculation time of the computing power resources required by the first computing power demand device and improving the scheduling efficiency of the computing power resources of the computing power demand device.

[0088] Refer to Figure 1 , Figure 1This is a schematic diagram of the scenario of the computing power resource scheduling method of this application. The computing power resource scheduling system is communicatively connected to multiple computing power supply systems, and the multiple computing power supply systems are communicatively connected. The multiple computing power supply systems are, for example, computing power supply system A, computing power supply system B, and computing power supply system C. The computing power resource scheduling system is communicatively connected to multiple computing power demand devices, and the multiple computing power demand devices include computing power demand device D, computing power demand device E, and computing power demand device F. The computing power demand device F sends a resource scheduling request to the computing power supply system. If the computing power demand device F is a device that has not been allocated computing power resources, while the computing power demand device D and the computing power demand device E are devices that have been allocated computing power resources, the computing power resource scheduling system determines, among the computing power demand device D and the computing power demand device E, a computing power demand device similar to the computing power demand device F. For example, if the computing power demand device similar to the computing power demand device F is the computing power demand device E, the computing power resource scheduling system sends a resource allocation instruction to one of the computing power supply systems. For example, it sends a resource allocation instruction to the computing power supply system C, and then the computing power supply system C provides the target computing power resources to the computing power demand device F, and the target computing power resources are equal to the allocated computing power resources of the computing power demand device E. The above-mentioned computing power demand devices include but are not limited to D, E, and F, and the above-mentioned computing power supply systems include but are not limited to A, B, and C.

[0089] The following combines Figure 1 to elaborate in detail on the technical solution shown in this application. It should be noted that the following several embodiments can exist independently or be combined with each other. For the same or similar content, it will not be repeated in different embodiments.

[0090] Refer to Figure 2 , Figure 2 This is the process schematic Figure 1 of the computing power resource scheduling method in the embodiment of this application. The computing power resource scheduling method includes:

[0091] Step S201, receive a resource scheduling request sent by a first computing power demand device, and determine the device identifier of the first computing power demand device according to the resource scheduling request.

[0092] In this embodiment, the execution entity is the computing power resource scheduling system, and the computing power resource scheduling system can schedule the computing power resources of the system on the device supply side by the device on the user demand side. The device on the user demand side is defined as the computing power demand device, and the system on the device supply side is defined as the computing power supply system. For the sake of convenience of description, the following uses the scheduling system to refer to the computing power resource scheduling system.

[0093] The current production environment has chips of different brands, different models, and different architectures, so that the computing power and performance that can be provided by the computing power supply systems equipped with chips vary. The scheduling and management of these computing power supply systems are uniformly managed through the scheduling system.

[0094] When the first computing power demand device needs to perform computing power resource scheduling, the first computing power demand device sends a resource scheduling request to the scheduling system. The resource scheduling request contains the device identifier of the first computing power demand device, and the scheduling system parses the resource scheduling request to obtain the device identifier.

[0095] Step S202, in response to the device identifier indicating that the first computing power demand device is a device without allocated computing power resources, determine the first similarity between the first computing power demand device and each second computing power demand device, where the second computing power demand device is used to indicate a device with allocated computing power resources.

[0096] After obtaining the device identifier, the scheduling system determines whether the first computing power demand device is a device newly applying for resources or a device that has already applied for resources based on the device identifier.

[0097] Exemplarily, there is a whitelist stored in the scheduling system, and the computing power demand devices corresponding to the device identifiers stored in the whitelist are devices with allocated computing power resources. The scheduling system determines whether the device identifier is stored in the whitelist. When the device identifier is stored in the whitelist, it can be determined that the first computing power demand device is a device without allocated computing power resources, that is, a device newly applying for resources. The computing power resources include the computing power of the processing unit, and the processing unit includes a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), etc.

[0098] When the first computing power demand device is a device without allocated computing power resources, the scheduling system determines, from the computing power demand devices with allocated computing power resources, the computing power demand devices similar to the first computing power demand device. The computing power demand devices with allocated computing power resources are defined as second computing power demand devices, that is, the second computing power demand devices are used to indicate devices with allocated computing power resources.

[0099] Exemplarily, the scheduling system obtains the first portrait features of the first computing power demand device. The first portrait features include business attributes, technical features, and business volume indicators.

[0100] The business attributes include, but are not limited to, the scenario solution domain, service channel domain, unified access domain, acceptance service domain, content information domain, clearing and settlement domain, and business operation domain;

[0101] The technical attributes include, but are not limited to, online, batch, and management;

[0102] For online transaction systems, the business volume indicators mainly include the average daily transaction volume, peak transaction volume, concurrent transaction volume, number of transactions per second, etc.; for batch processing systems, the business volume indicators mainly include the data volume and processing duration; for management systems, the business volume indicators include the access frequency, number of users, and number of queries per second.

[0103] The scheduling system stores the second portrait features of each second computing power demand device. The second portrait features include business attributes, technical features, and traffic volume indicators. The scheduling system constructs a first feature vector based on the first portrait features of the first computing power demand device, constructs a second feature vector based on the second portrait features of the second computing power demand device, calculates the distance between the first feature vector and the second feature vector, and obtains the similarity through distance conversion. This similarity is defined as the first similarity, and the smaller the distance, the greater the first similarity.

[0104] Step S203: Determine the second similarity among the first similarities, and determine the allocated first computing power resources of the second computing power demand device corresponding to the second similarity as the target computing power resources of the first computing power demand device. The second similarity is the first similarity greater than the preset similarity.

[0105] After determining the first similarities, the scheduling system determines the second similarity among the first similarities, that is, determines the first similarity greater than the preset similarity among the first similarities. The first similarity greater than the preset similarity can be used as the second similarity. When there are multiple first similarities greater than the preset similarity, the largest first similarity is used as the second similarity.

[0106] After determining the second similarity, the scheduling system uses the allocated first computing power resources of the second computing power demand device corresponding to the second similarity as the target computing power resources of the first computing power demand device. Exemplarily, the scheduling system stores the allocation records of the allocated computing power resources of each second computing power demand device, and obtains the first computing power resources based on the allocation records.

[0107] Step S204: Send a resource allocation instruction to the first computing power supply system according to the target computing power resources, so that the computing power supply system provides the target computing power resources to the first computing power demand device.

[0108] After determining the target computing power resources, the scheduling system determines the computing power supply system with idle computing power resources as the first computing power supply system. The scheduling system generates a resource allocation instruction based on the target computing power resources and the device identifier, and then sends the resource allocation instruction to the first computing power supply system. The first supply system parses the resource allocation instruction to obtain the device identifier and the target computing power resources. The first computing power supply system provides the target computing power resources to the first computing power demand device corresponding to the device identifier.

[0109] In this embodiment, when the first computing power demand device for which resource scheduling is applied is a device without allocated computing power resources, a second computing power demand device similar to the first computing power demand device is determined among multiple second computing power demand devices with allocated computing power resources. Thus, the computing power resources of the similar second computing power demand device are used to configure the computing power resources for the first computing power demand device, eliminating the need for manual calculation of the computing power resources required by the first computing power demand device, reducing the calculation duration of the computing power resources required by the first computing power demand device, and improving the scheduling efficiency of the computing power resources of the computing power demand device.

[0110] Referring to Figure 3 , Figure 3 is a flowchart of the computing power resource scheduling method of this application Figure 2 , based on Figure 2 the embodiment shown, step S204 includes:

[0111] Step S301, determining the system where the first computing power demand device is deployed as the first computing power supply system.

[0112] In this embodiment, the types of processing units in different computing power supply systems are different, so the computing power resources of different computing power supply systems are different. Exemplarily, if the processing unit is a CPU, and the types of CPUs are different, then the computing power of a single-core CPU is different. Specifically, refer to the following table:

[0113] CPU Type Number of Logical Cores Total Computing Power Single-Core Computing Power A 48 327 6.81 B 64 270.7 4.23 C 32 377.8 11.81

[0114] As shown in the above table, if the CPU type of the computing power supply system is A, the logical core number of the A-type CPU is 48, and the computing power of each logical CPU core is the single-core computing power, and the single-core computing power is 6.81 units;

[0115] If the CPU type of the computing power supply system is B, the logical core number of the B-type CPU is 64, and the computing power of each logical CPU core is the single-core computing power, and the single-core computing power is 4.23 units;

[0116] If the CPU type of the computing power supply system is C, the logical core number of the C-type CPU is 32, and the computing power of each logical CPU core is the single-core computing power, and the single-core computing power is 11.81 units.

[0117] It should be noted that the processing unit in the computing power supply system can be a CPU, or other types of processing units such as a GPU, that is, a table with other types of processing units such as a GPU.

[0118] The scheduling system determines the computing power supply system where the first computing power demand device is deployed as the first computing power supply system.

[0119] Step S302: Obtain the first computing power parameter of the single-core processing unit in the first computing power supply system, and determine the first ratio between the target computing power resource and the first computing power parameter.

[0120] The above table is stored in the scheduling system. After determining the first computing power supply system, determine the type of the processing unit in the first computing power supply system, and determine the single-core computing power from the table based on this type. The single-core computing power is the first computing power parameter of the single-core processing unit in the first computing power supply system.

[0121] The scheduling system determines the first ratio between the target computing power resource and the first computing power parameter. For example, if the target computing power resource is the standardized computing power 1181 and the CPU type of the first computing power supply system is type A, then the first computing power parameter is 6.81. Therefore, the first ratio = 1181 / 6.81 = 173.4.

[0122] Step S303: According to the first ratio, determine the first number of cores of the processing unit required for the first computing power supply system to provide the target computing power resource, and generate a resource allocation instruction based on the first number of cores.

[0123] After determining the first ratio, determine the first number of cores of the processing unit required for the first computing power supply system to provide the target computing power resource based on the first ratio. Exemplarily, if the first ratio is 173.4, then the first number of cores is 173. The scheduling system generates a resource allocation instruction based on the first number of cores, that is, generates a resource allocation instruction based on the first number of cores and the device identifier.

[0124] Step S304: Send the resource allocation instruction to the first computing power supply system.

[0125] The scheduling system sends the resource allocation instruction to the first computing power supply system, that is, the first computing power supply system needs to provide the computing power of 173 logical CPU cores to the first computing power demand device.

[0126] In this embodiment, the scheduling system determines the computing power of the processing unit with the first number of cores that needs to be scheduled to the first computing power demand device based on the computing power parameter of the single-core processing unit in the first computing power supply system, so as to accurately provide accurate computing power resources for the first computing power demand device.

[0127] Refer to Figure 4 , Figure 4 For the flowchart illustration of the computing power resource scheduling method of this application Figure 3 , based on the embodiment shown in 3, before step S301, it further includes:

[0128] Step S401: Obtain the total number of cores of the processing units of each second computing power supply system, and the first computing power supply system is any one of the second computing power supply systems.

[0129] In this embodiment, the scheduling system can perform a standardized computing power evaluation on the heterogeneous computing power supply system. The heterogeneous computing power supply system refers to a computing system formed by combining different types of processing units, and the processing units include CPUs, GPUs, NPUs (Neural network Processing Unit), and FPGAs (Field Programmable Gate Array), etc. A system of one type of processing unit in the heterogeneous computing power supply system serves as the computing power supply system. In the exemplary technology, there is a lack of a computing power evaluation method for various types of processing units in the heterogeneous computing power supply system, resulting in an excessive or insufficient allocation of computing power resources

[0130] The scheduling system needs to determine the total number of cores of the processing units of each second computing power supply system. For example, the type of the CPU in a computing power supply system is A, and the logical number of cores of the CPU in the computing power supply system is 64, that is, the total number of cores is 64. The total number of cores of the processing units of the second computing power supply system can be obtained through the system attribute information of the second computing power supply system. In addition, the above-mentioned first computing power supply system is any second computing power supply system

[0131] Step S402: Use a variety of test cases to test the processing units of the second computing power supply system respectively, and obtain the computing power sub-parameters of the processing units of the second computing power supply system in each test case

[0132] The scheduling system uses a variety of test cases to test the processing units of the second computing power supply system respectively, and obtains the computing power sub-parameters of the processing units of the second computing power supply system in each test case. In one embodiment, when the processing unit is a CPU, test cases in four test sets, namely IntRate, IntSpeed, FpRate, and FpSpeed, can be used to test the processing unit, and the computing power sub-parameters of the processing unit in each test case are obtained, as shown in the following table

[0133]

[0134] where TC = TC IntRate + TC IntSpeed + TC FpRate + TC FpSpeed , TC is the standard total computing power (total computing power parameter) of the computing power supply system, TC IntRate is the value after normalization and standardization after the IntRate test, TC IntSpeed , TC FpRate , TC FpSpeed Similarly

[0135] TC IntRate , TCIntSpeed , TC FpRate , TC FpSpeed are all computing power sub-parameters.

[0136] In another example, the processing unit of the second computing power supply system is a graphics processing unit GPU. Determine the label of the GPU, which is used to indicate the use of the processing unit in the second computing power supply system. Exemplarily, the GPU can be used for training or for inference. The scheduling system tests the graphics processing unit respectively based on each test case of label matching. For example, when the GPU is used for training, various test cases include test cases for image classification, object detection, speech recognition, natural language processing, and reinforcement learning; when the GPU is used for inference, various test cases include test cases for image classification, object detection, speech-to-text conversion, language processing, etc.

[0137] Step S403: Determine the total computing power parameter of the second computing power supply system according to the sum of each computing power sub-parameter of the processing unit of the second computing power supply system.

[0138] After determining each computing power sub-parameter, the sum of each computing power sub-parameter is the total computing power parameter TC of the second computing power supply system.

[0139] Step S404: Determine the second computing power parameter of the single-core processing unit in the second computing power supply system according to the second ratio between the total computing power parameter of the second computing power supply system and the total number of cores, and associate and store the second computing power parameter with the system identifier of the second computing power supply system.

[0140] After the scheduling system determines the total computing power parameter, determine the second ratio between the total computing power parameter and the total number of cores, and then the second computing power parameter of the single-core processing unit in the second computing power supply system can be determined. The second computing power parameter SC is:

[0141] After determining the second computing power parameter, the scheduling system associates and stores the system identifier of the second computing power supply system and the second computing power parameter. When the scheduling system needs to determine the first computing power parameter, it obtains the computing power parameter associated with the system identifier of the first computing power supply system as the second computing power parameter.

[0142] In this embodiment, the scheduling system provides a standardized computing power evaluation for the computing power supply system, so as to conduct a unified computing power evaluation for the computing power of the computing power supply systems of different types of processing units, and then accurately configure computing power resources for the computing power demand devices, avoiding providing too much or too little computing power resources for the computing power demand devices.

[0143] Refer to Figure 5 , Figure 5 which is the flowchart of the computing power resource scheduling method of this application Figure 4, based on Figures 2 to 4 In any of the embodiments shown in

[0144] Step S501, obtain the target resource utilization rate of the second computing power demand device and determine the current computing power resources of the second computing power demand device.

[0145] In this embodiment, the resource utilization rate of the second computing power demand device, which is similar to the first computing power demand device, needs to be within the reasonable resource utilization rate range [a, b], and the computing power resources of the second computing power demand device also need to be appropriately adjusted based on the range.

[0146] For this, the scheduling system obtains the target resource utilization rate of the second computing power demand device. In one example, the target resource utilization rate can be sent by the second computing power demand device to the scheduling system.

[0147] In another example, the scheduling system determines each computing power supply system in which the second computing power demand device is deployed. The computing power supply system is defined as the third computing power supply system, and the scheduling system also determines the first application center set in each third computing power supply system. For example, the second computing power demand device is deployed in two computing power supply systems A and B. The computing power supply system A deploys M application centers. The application center is the logical center. The service traffic carried by each application center in A is J1, J2... J m , in B; the computing power supply system B deploys N application centers. The application center is the logical center. The service traffic carried by each application center in B is K1, K2... K n . The second computing power demand device has a direct correlation with the service traffic size. The larger the service traffic, the more fully the resources are utilized, and its utilization rate is more representative. Therefore, during the peak service period, the resource utilization rate of the application center with the highest service classification ratio can be considered as the benchmark to calculate the resource utilization rate of the second computing power demand device. The application centers set in each third computing power supply system are defined as the first application centers. The scheduling system determines the second application center among the first application centers. The second application center is the first application center with the largest service diversion ratio, and is based on the third ratio between the computing power resources of the second application center and the service classification ratio of the second application center.

[0148] Exemplarily, Among them, Umax represents the CPU and memory utilization rates of instances such as virtual machines, containers, databases, and caches in the application center with the highest diversion ratio among multiple first application centers. During the peak service period, the CPU or memory utilization rate of each instance can be weighted and averaged. That is, Umax is the computing power resources of the second application center. Max(J, K) represents the service diversion ratio of this application center. The scheduling system takes the third ratio as the target resource utilization rate.

[0149] In another example, the second computing power demand device needs to ensure the normal operation of the service, that is, it is necessary to set the service continuity guarantee index Q. That is, in an extreme scenario, the second computing power demand device allows M + N - Q application centers to be abnormal, and the remaining Q application centers undertake the service. Considering that in an extreme scenario, Q application centers undertake the guarantee function and carry all the service traffic of the second computing power demand device, and the service traffic ratio of each application center is 1 / Q. It can be understood that the scheduling system determines the minimum number of the first application centers required to operate normally for processing the service of the second computing power demand device, and the minimum number is Q. Based on the third ratio and the minimum number, the resource utilization rate of each first application center is determined. The resource utilization rate of each first application center is as follows:

[0150]

[0151] The resource utilization rate of the above-mentioned first application center can be used as the target resource utilization rate.

[0152] The scheduling system obtains the current computing power resources of the second computing power demand device, and the current computing power resources can also be sent by the second computing power demand device to the scheduling system.

[0153] Step S502, modify the current computing power resources according to the target resource utilization rate to obtain the first computing power resources.

[0154] After obtaining the target resource utilization rate, the scheduling system can modify the current computing power resources based on the target resource utilization rate to obtain the first computing power resources.

[0155] Exemplarily, the scheduling system obtains the utilization rate interval associated with the second computing power demand device, and compares the target resource utilization rate and the utilization rate interval.

[0156] When the target resource utilization rate is less than the lower limit value of the utilization rate interval, reduce the current computing power resources to obtain the first computing power resources, that is, downsize the computing power resources of the second computing power demand device, which is achieved by reducing the number of CPU cores provided by the computing power supply system to the second computing power demand device;

[0157] When the target resource utilization rate is greater than the upper limit value of the utilization rate interval, increase the current computing power resources to obtain the first computing power resources, that is, expand the computing power resources of the second computing power demand device, which is achieved by increasing the number of CPU cores provided by the computing power supply system to the second computing power demand device.

[0158] When the target resource utilization rate is within the utilization rate interval, there is no need to modify the current computing power resources, that is, regard the current computing power resources as the first computing power resources.

[0159] It should be noted that when the second computing power demand device needs to be expanded, the application center of the second computing power demand device also needs to be expanded proportionally; when the second computing power demand device needs to be scaled down, the application center of the second computing power demand device also needs to be scaled down proportionally.

[0160] In this embodiment, on the premise of ensuring the business continuity of the second computing power demand device, the second computing power demand device is expanded or scaled down to avoid resource redundancy, achieving the purpose of cost reduction and efficiency improvement.

[0161] Refer to Figure 6 , Figure 6 which is a schematic flowchart of the computing power resource scheduling method of this application Figure 5 Based on Figures 2 to 5 any of the embodiments shown in

[0162] Step S601, in response to the device identifier indicating that the first computing power demand device is a device with allocated computing power resources, determine the third computing power supply system designated by the first computing power demand device to provide computing power resources according to the resource scheduling request.

[0163] In this embodiment, resource scheduling can be resource scheduling for newly launched devices or computing power scheduling for devices that have already been allocated computing power resources. Computing power scheduling refers to switching from one computing power supply system to another computing power supply system to provide computing power. For example, the second computing power demand device is provided with computing power resources by computing power supply system A, and the computing power resources of the second computing power demand device are switched from computing power supply system A to computing power supply system B for supply.

[0164] In this regard, when the scheduling system obtains the device identifier of the first computing power demand device based on the resource scheduling request, if the device identifier is in the white list, it can be determined that the first computing power demand device is a device with allocated computing power resources. When the first computing power demand device is a device with allocated computing power resources, determine the third computing power supply system designated by the first computing power demand device to provide computing power resources according to the resource scheduling request. Exemplarily, the resource scheduling request contains the computing power supply system designated by the first computing power demand device, and the scheduling system parses the resource scheduling request to obtain the computing power supply system designated by the first computing power demand device as the third computing power supply system.

[0165] Step S602, obtain the second computing power resources already allocated to the first computing power demand device.

[0166] After the scheduling system determines the third computing power supply system, it obtains the second computing power resources already allocated to the first computing power demand device. The obtaining process of the second computing power resources is the same as the obtaining process of the first computing power resources. For specific reference, see the above description and details will not be repeated here.

[0167] Step S603: Send a resource scheduling instruction to the third computing power supply system according to the second computing power resource, so that the third computing power supply system provides the second computing power resource to the first computing power demand device.

[0168] After determining the second computing power resource, generate a resource scheduling instruction based on the second computing power resource, and send the resource scheduling instruction to the third computing power supply system. The third computing power supply system parses the resource scheduling instruction to obtain the second computing power resource, and the third computing power supply system provides the second computing power resource to the first computing power demand device. In addition, the scheduling system sends an instruction to the current computing power supply system of the first computing power demand device, so that the current computing power supply system stops providing computing power resources to the first computing power demand device.

[0169] In an embodiment, the scheduling system determines the third computing power parameter of the single-core processing unit in the current computing power supply system, where the current computing power supply system is the computing power supply system that currently provides computing power resources to the first computing power demand device.

[0170] Exemplarily, the types of processing units in different computing power supply systems are different, so the computing power resources of different computing power supply systems are different. The processing unit is a CPU, and if the types of CPUs are different, the computing power of a single-core CPU is different. Specifically, refer to the following table:

[0171] CPU Type Number of Logical Cores Total Computing Power Single-Core Computing Power A 48 327 6.81 B 64 270.7 4.23 C 32 377.8 11.81

[0172] As shown in the above table, if the CPU type of the computing power supply system is A, the number of logical cores of the A-type CPU is 48, and the computing power of each logical CPU core is the single-core computing power, and the single-core computing power is 6.81 units;

[0173] If the CPU type of the computing power supply system is B, the number of logical cores of the B-type CPU is 64, and the computing power of each logical CPU core is the single-core computing power, and the single-core computing power is 4.23 units;

[0174] If the CPU type of the computing power supply system is C, the number of logical cores of the C-type CPU is 32, and the computing power of each logical CPU core is the single-core computing power, and the single-core computing power is 11.81 units.

[0175] It should be noted that the processing unit in the computing power supply system can be a CPU, or other types of processing units such as GPUs, that is, a table with other types of processing units such as GPUs.

[0176] The scheduling system determines the computing power supply system where the first computing power demand device is deployed as the current computing power supply system. The above table is stored in the scheduling system. After determining the current computing power supply system, determine the type of the processing unit in the current computing power supply system, and determine the single-core computing power from the table based on this type. The single-core computing power is the third computing power parameter of the single-core processing unit in the current computing power supply system.

[0177] The scheduling system determines the second number of cores of the processing unit in the current computing power supply system that provides computing power resources to the first computing power demand device. Thus, based on the product between the second number of cores and the third computing power parameter, the second computing power resources allocated to the first computing power demand device can be obtained. Exemplarily, the second number of cores is 173, and the third computing power parameter is 6.81, then the second computing power resources are 173 × 6.81 = 1181.

[0178] Furthermore, the scheduling system obtains the fourth computing power parameter of the single-core processing unit in the third computing power supply system, and determines the fourth ratio between the second computing power resources and the fourth computing power parameter. Exemplarily, the processing unit of the third computing power supply system is of type B, the fourth computing power parameter is 4.23, and the second computing power resources are 1181, then the fourth ratio = 1181 / 4.23 = 279.2.

[0179] The scheduling system determines the third number of cores of the processing unit required for the second computing power resources provided by the third supply system based on the fourth ratio. For example, if the fourth ratio is 279.2, then the third number of cores is 279. The scheduling system generates a resource scheduling instruction based on the third number of cores, and sends the resource scheduling instruction to the third computing power supply system, that is, the third computing power supply system provides computing power resources of 279 CPU logical cores to the first computing power demand device.

[0180] In this embodiment, the scheduling system supports the first computing power demand device to perform switching of computing power resources.

[0181] Refer to Figure 7 , Figure 7 is the flow diagram of the computing power resource scheduling method of this application Figure 6 Based on Figures 2 to 6 In any of the embodiments shown, step S202 includes:

[0182] Step S701, obtain the first portrait features of the first computing power demand device.

[0183] Step S702, input each of the first portrait features into the resource matching model, and obtain the first similarity between the first computing power demand device and each of the second computing power demand devices output by the resource matching model.

[0184] In this embodiment, the scheduling system obtains the first portrait features of the first computing power demand device, each of the first portrait features. The first portrait features include business attributes, technical features, and business volume indicators.

[0185] The business attributes include but are not limited to the scenario solution domain, service channel domain, unified access domain, acceptance service domain, content information domain, clearing and settlement domain, and business operation domain; [[ID=3 June 2023]]

[0186] Technical attributes include, but are not limited to, online, batch, and management;

[0187] For online transaction systems, the business volume indicators mainly include the average daily transaction volume, peak transaction volume, concurrent transaction volume, number of transactions per second, etc.; for batch processing systems, the business volume indicators mainly include the data volume, processing duration, etc.; for management systems, the business volume indicators include the access frequency, number of users, and number of queries per second.

[0188] The scheduling system stores a resource matching model. When the scheduling system inputs each first portrait feature into the resource matching model, the first similarity between the first computing power demand devices output by the resource matching model and each second computing power demand device can be obtained.

[0189] In addition, the scheduling system obtains the second portrait features of the second computing power demand devices, constructs training samples corresponding to the second computing power demand devices based on the second portrait features of the second computing power demand devices, and trains a preset model based on each training sample to obtain a resource matching model. The second portrait features also include resource utilization rate, standardized computing power, and corrected standardized computing power. The acquisition method of the resource utilization rate is the same as the acquisition method of the above-mentioned target resource utilization rate, and will not be elaborated here. The standardized computing power is the total computing power of the second computing power demand device, and the corrected standardized computing power refers to the computing power obtained by scaling up or down based on the target resource utilization rate and the utilization rate interval.

[0190] It should be noted that if the second computing power demand device includes multiple second portrait features, determine the number of second portrait features. If the number is greater than the preset number, there are too many second portrait features. The scheduling system determines the importance parameters of each second portrait feature, deletes the second portrait features with importance parameters lower than the preset threshold, and constructs training samples from the remaining second portrait features. The training method of the preset model can train the second portrait features through gradient boosting regression or random forest method, and use cross-validation to adjust the hyperparameters to capture the non-linear relationship, and then the resource matching model can be trained.

[0191] It should be noted that after the first computing power demand device is allocated computing power resources, the first portrait features, resource utilization rate, etc. of the first computing power demand device are used as basic data to optimize the allocated computing power resources of the first computing power demand device. The optimization method is to scale up or down based on the target resource utilization rate of the above-mentioned second computing power demand device. The optimized computing power resources, first portrait features, and resource utilization rate are used as data to train the resource matching model, thereby optimizing the resource matching model.

[0192] In this embodiment, the similarity between the first computing power demand device and each second computing power demand device is quickly determined through the resource matching model, improving the resource scheduling efficiency.

[0193] Based on the content described in the above embodiments, an embodiment of the present application further provides a resource scheduling system. Referring to Figure 8 , Figure 8 which is a schematic diagram of program modules of a resource scheduling system provided in an embodiment of the present application. In some embodiments, the resource scheduling system 800 includes:

[0194] A receiving module 810, configured to receive a resource scheduling request sent by a first computing power demand device, and determine the device identifier of the first computing power demand device according to the resource scheduling request;

[0195] A first determination module 820, configured to determine a first similarity between the first computing power demand device and each second computing power demand device in response to the device identifier indicating that the first computing power demand device is a device without allocated computing power resources, where the second computing power demand device is used to indicate a device with allocated computing power resources;

[0196] A second determination module 830, configured to determine a second similarity among the first similarities, and determine the first computing power resources allocated to the second computing power demand device corresponding to the second similarity as the target computing power resources of the first computing power demand device, where the second similarity is a first similarity greater than a preset similarity;

[0197] A sending module 840, configured to send a resource allocation instruction to a first computing power supply system according to the target computing power resources, so that the computing power supply system provides the target computing power resources to the first computing power demand device.

[0198] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0199] Determine the system where the first computing power demand device is deployed as the first computing power supply system;

[0200] Obtain a first computing power parameter of a single-core processing unit in the first computing power supply system, and determine a first ratio between the target computing power resources and the first computing power parameter;

[0201] According to the first ratio, determine a first number of cores of the processing unit required for the first computing power supply system to provide the target computing power resources, and generate a resource allocation instruction according to the first number of cores;

[0202] Send the resource allocation instruction to the first computing power supply system.

[0203] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0204] Obtain the total number of cores of the processing units of each second computing power supply system, where the first computing power supply system is any one of the second computing power supply systems;

[0205] Use multiple test cases to test the processing unit of the second computing power supply system respectively, and obtain the computing power sub-parameters of the processing unit of the second computing power supply system in each test case;

[0206] Determine the total computing power parameter of the second computing power supply system according to the sum of the computing power sub-parameters of the processing unit of the second computing power supply system;

[0207] Determine the second computing power parameter of the single-core processing unit in the second computing power supply system according to the second ratio between the total computing power parameter of the second computing power supply system and the total number of cores, and store the second computing power parameter in association with the system identifier of the second computing power supply system.

[0208] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0209] In response to the processing unit of the second computing power supply system being an image processing unit, determine the label of the image processing unit, where the label is used to indicate the use of the image processing unit in the second computing power supply system;

[0210] Test the image processing unit respectively according to the test cases matched by the label.

[0211] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0212] Obtain the target resource utilization rate of the second computing power demand device and determine the current computing power resources of the second computing power demand device;

[0213] Modify the current computing power resources according to the target resource utilization rate to obtain the first computing power resources.

[0214] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0215] Determine each third computing power supply system deployed by the second computing power demand device, and determine the first application center set for each third computing power supply system;

[0216] Determine the second application center among the first application centers, where the second application center is the first application center with the largest service diversion ratio;

[0217] Determine the target resource utilization rate of the second computing power demand device according to the third ratio between the computing power resources of the second application center and the service diversion ratio of the second application center.

[0218] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0219] Determine the minimum number of first application centers that need to run normally for processing the services of the second computing power demand device;

[0220] Determine the resource utilization rate of each first application center as the target resource utilization rate according to the third ratio and the minimum quantity.

[0221] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0222] Obtain the utilization rate interval associated with the second computing power demand device;

[0223] In response to the target resource utilization rate being less than the lower limit value of the utilization rate interval, reduce the current computing power resource to obtain the first computing power resource;

[0224] In response to the target resource utilization rate being greater than the upper limit value of the utilization rate interval, increase the current computing power resource to obtain the first computing power resource.

[0225] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0226] In response to the device identifier indicating that the first computing power demand device is a device with allocated computing power resources, determine the third computing power supply system that the first computing power demand device is specified to provide computing power resources according to the resource scheduling request;

[0227] Obtain the allocated second computing power resource of the first computing power demand device;

[0228] According to the second computing power resource, send a resource scheduling instruction to the third computing power supply system for the third computing power supply system to provide the second computing power resource to the first computing power demand device.

[0229] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0230] Determine the third computing power parameter of the single-core processing unit in the current computing power supply system, where the current computing power supply system is the computing power supply system that currently provides computing power resources to the first computing power demand device;

[0231] Determine the second core number of the processing unit in the current computing power supply system that provides computing power resources to the first computing power demand device;

[0232] Determine the allocated second computing power resource of the first computing power demand device according to the product of the second core number and the third computing power parameter.

[0233] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0234] Obtain the fourth computing power parameter of the single-core processing unit in the third computing power supply system and determine the fourth ratio between the second computing power resource and the fourth computing power parameter;

[0235] According to the fourth ratio, determine the third core number of the processing unit required for the third computing power supply system to provide the second computing power resource, and generate a resource scheduling instruction according to the third core number;

[0236] Send the resource scheduling instruction to the third computing power supply system.

[0237] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0238] Obtain the first portrait features of the first computing power demand device;

[0239] Input each of the first portrait features into the resource matching model to obtain the first similarity between the first computing power demand device and each of the second computing power demand devices output by the resource matching model.

[0240] In some embodiments, the resource scheduling system 800 is specifically configured to:

[0241] Obtain the second portrait features of each second computing power demand device, and construct a training sample corresponding to the second computing power demand device according to the second portrait features corresponding to the second computing power demand device;

[0242] Train a preset model according to each training sample to obtain a resource matching model.

[0243] It should be noted that the specific steps in the computing power resource scheduling method executed by the computing power resource scheduling system are specifically referred to the above embodiments and will not be elaborated here.

[0244] Furthermore, based on the content described in the above embodiments, an electronic device is further provided in the embodiments of the present application. The electronic device includes at least one processor, and a communication interface and a memory communicatively connected to the processor; wherein, the communication interface is used to communicate with other communication devices, and the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory to implement each step in the computing power resource scheduling method described in the above embodiments.

[0245] For a better understanding of the embodiments of the present application, refer to Figure 9 , Figure 9 which is a schematic hardware structure diagram of an electronic device provided in the embodiments of the present application.

[0246] As Figure 9 shown, the electronic device 900 in this embodiment includes: a processor 901, a memory 902, and a communication interface 904; wherein:

[0247] The memory 902 is used to store computer execution instructions;

[0248] The communication interface 904 is used to communicate with other communication devices;

[0249] A processor 901 for executing computer-executable instructions stored in a memory to implement the various steps in the query optimization method described in the above embodiments.

[0250] Optionally, the memory 902 can be either independent or integrated with the processor 901.

[0251] When the memory 902 is independently provided, the device further includes a bus 903 for connecting the memory 902, the communication interface 904, and the processor 901.

[0252] An embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions, and when the processor executes the computer-executable instructions, the various steps in the computing power resource scheduling method described in the above embodiments are implemented.

[0253] An embodiment of the present application provides a computer program product including a computer program, and when the computer program is executed by the processor, the various steps in the computing power resource scheduling method described in the above embodiments are implemented.

[0254] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of devices or modules can be in electrical, mechanical or other forms.

[0255] The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0256] In addition, in each embodiment of the present application, the various functional modules can be integrated in a processing unit, or each module can exist physically alone, or two or more modules can be integrated in one unit. The units formed by the above modules can be implemented in the form of hardware or in the form of a hardware plus software functional unit.

[0257] The integrated module implemented in the form of software functional modules can be stored in a computer-readable storage medium. The above-mentioned software functional modules are stored in a storage medium and include several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to execute some steps of the methods of various embodiments of the present application.

[0258] It should be understood that the above-mentioned processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the application can be directly embodied as being executed by a hardware processor, or can be executed by a combination of hardware and software modules in the processor.

[0259] The memory may include high-speed memory and may also include non-volatile storage, such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a magnetic disk, or an optical disc, etc.

[0260] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the buses in the drawings of the present application are not limited to only one bus or one type of bus.

[0261] The above-mentioned storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory, an electrically erasable programmable read-only memory, an erasable programmable read-only memory, a programmable read-only memory, a read-only memory, a magnetic memory, a flash memory, a magnetic disk, or an optical disc. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0262] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A computing power resource scheduling method, characterized in that Including: Receiving a resource scheduling request sent by a first computing power demand device, and determining a device identifier of the first computing power demand device according to the resource scheduling request; In response to the device identifier indicating that the first computing power demand device is a device without allocated computing power resources, determining a first similarity between the first computing power demand device and each second computing power demand device, where the second computing power demand device is used to indicate a device with allocated computing power resources; Determining a second similarity among the first similarities, and determining the first computing power resources allocated to the second computing power demand device corresponding to the second similarity as the target computing power resources of the first computing power demand device, where the second similarity is the first similarity greater than a preset similarity; Sending a resource allocation instruction to a first computing power supply system according to the target computing power resources, so that the computing power supply system provides the target computing power resources to the first computing power demand device.

2. The method according to claim 1, characterized in that, The sending a first resource allocation instruction to a first computing power supply system according to the target computing power resources includes: Determining the system where the first computing power demand device is deployed as the first computing power supply system; Obtaining a first computing power parameter of a single-core processing unit in the first computing power supply system, and determining a first ratio between the target computing power resources and the first computing power parameter; Determining a first number of cores of the processing unit required for the first computing power supply system to provide the target computing power resources according to the first ratio, and generating a resource allocation instruction according to the first number of cores; Sending the resource allocation instruction to the first computing power supply system.

3. The method according to claim 2, wherein Before obtaining the first computing power parameter of the single-core processing unit in the first computing power supply system, it further includes: Obtaining the total number of cores of the processing units of each second computing power supply system, where the first computing power supply system is any one of the second computing power supply systems; Testing the processing units of the second computing power supply system respectively with a variety of test cases to obtain computing power sub-parameters of the processing units of the second computing power supply system in each test case; Determining the total computing power parameter of the second computing power supply system according to the sum of the computing power sub-parameters of the processing units of the second computing power supply system; Determining a second computing power parameter of the single-core processing unit in the second computing power supply system according to a second ratio between the total computing power parameter of the second computing power supply system and the total number of cores, and associatively storing the second computing power parameter with the system identifier of the second computing power supply system.

4. The method according to claim 3, characterized in that, The testing the processing units of the second computing power supply system respectively with a variety of test cases includes: In response to the processing unit of the second computing power supply system being an image processing unit, determining a label of the image processing unit, where the label is used to indicate the use of the image processing unit in the second computing power supply system; Testing the image processing unit respectively according to the test cases matched by the label.

5. The method according to claim 1, wherein The determining the first computing power resources allocated to the second computing power demand device corresponding to the second similarity includes: Obtain the target resource utilization rate of the second computing power demand device, and determine the current computing power resources of the second computing power demand device; Modify the current computing power resources according to the target resource utilization rate to obtain the first computing power resources.

6. The method according to claim 5, characterized in that, The obtaining the target resource utilization rate of the second computing power demand device includes: Determine each third computing power supply system deployed by the second computing power demand device, and determine the first application center set in each third computing power supply system; Determine a second application center among each of the first application centers, where the second application center is the first application center with the largest service diversion ratio; Determine the target resource utilization rate of the second computing power demand device according to a third ratio between the computing power resources of the second application center and the service diversion ratio of the second application center.

7. The method according to claim 6, characterized in that The determining the target resource utilization rate of the second computing power demand device according to a third ratio between the computing power resources of the second application center and the service diversion ratio of the second application center includes: Determine the minimum number of first application centers that need to operate normally to process the services of the second computing power demand device; Determine the resource utilization rate of each of the first application centers as the target resource utilization rate according to the third ratio and the minimum number.

8. The method according to claim 5, characterized in that The modifying the current computing power resources according to the target resource utilization rate to obtain the allocated computing power resources includes: Obtain the utilization rate interval associated with the second computing power demand device; In response to the target resource utilization rate being less than the lower limit value of the utilization rate interval, reduce the current computing power resources to obtain the first computing power resources; In response to the target resource utilization rate being greater than the upper limit value of the utilization rate interval, increase the current computing power resources to obtain the first computing power resources.

9. The method according to claim 1, characterized in that After determining the device identifier of the first computing power demand device according to the resource scheduling request, it further includes: In response to the device identifier indicating that the first computing power demand device is a device with allocated computing power resources, determine the third computing power supply system designated by the first computing power demand device to provide computing power resources according to the resource scheduling request; Obtain the allocated second computing power resources of the first computing power demand device; Send a resource scheduling instruction to the third computing power supply system according to the second computing power resources, so that the third computing power supply system provides the second computing power resources to the first computing power demand device.

10. The method according to claim 9, wherein The obtaining the allocated second computing power resources of the first computing power demand device includes: Determine the third computing power parameter of the single-core processing unit in the current computing power supply system, where the current computing power supply system is the computing power supply system currently providing computing power resources to the first computing power demand device; Determine the second core number of the processing unit in the current computing power supply system that provides computing power resources to the first computing power demand device; Determine the allocated second computing power resources of the first computing power demand device according to the product of the second core number and the third computing power parameter.

11. The method according to claim 9, wherein The sending a resource scheduling instruction to the third computing power supply system according to the second computing power resources includes: Obtain the fourth computing power parameter of the single-core processing unit in the third computing power supply system, and determine the fourth ratio between the second computing power resource and the fourth computing power parameter; According to the fourth ratio, determine the third core number of the processing unit required for the third computing power supply system to provide the second computing power resource, and generate a resource scheduling instruction according to the third core number; Send the resource scheduling instruction to the third computing power supply system.

12. The method according to any one of claims 1-11, characterized in that, The determining the first similarity between the first computing power demand device and each second computing power demand device includes: Obtain the first portrait feature of the first computing power demand device; Input each of the first portrait features into a resource matching model, and obtain the first similarity between the first computing power demand device and each second computing power demand device output by the resource matching model.

13. The method according to claim 12, characterized in that, Before inputting each of the first portrait features into the resource matching model, it further includes: Obtain the second portrait features of each second computing power demand device, and construct a training sample corresponding to the second computing power demand device according to the second portrait features corresponding to the second computing power demand device; Train a preset model according to each of the training samples to obtain a resource matching model.

14. A computing power resource scheduling system, characterized in that, It includes: A receiving module, configured to receive a resource scheduling request sent by a first computing power demand device, and determine the device identifier of the first computing power demand device according to the resource scheduling request; A first determining module, configured to, in response to the device identifier indicating that the first computing power demand device is a device without allocated computing power resources, determine the first similarity between the first computing power demand device and each second computing power demand device, where the second computing power demand device is used to indicate a device with allocated computing power resources; A second determining module, configured to determine a second similarity among each of the first similarities, and determine the allocated first computing power resource of the second computing power demand device corresponding to the second similarity as the target computing power resource of the first computing power demand device, where the second similarity is the first similarity greater than a preset similarity; A sending module, configured to send a resource allocation instruction to a first computing power supply system according to the target computing power resource, so that the computing power supply system provides the target computing power resource to the first computing power demand device.

15. An electronic device, characterized in that, It includes: A processor, and a memory and a communication interface communicatively connected to the processor; The communication interface is used to communicate with other communication devices; The memory is used to store computer execution instructions; The processor is used to execute the computer execution instructions stored in the memory to implement the computing power resource scheduling method according to any one of claims 1-13.

16. A computer-readable storage medium, characterized in that, Computer execution instructions are stored in the computer-readable storage medium, and when the computer execution instructions are executed by a processor, the computing power resource scheduling method according to any one of claims 1-13 is implemented.

17. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, the computing power resource scheduling method according to any one of claims 1-13 is implemented.

Images