Calculation power evaluation method and device, equipment, storage medium and program product

Abstract

The invention discloses a computing power assessment method and device, equipment, a storage medium and a program product, and the method comprises the steps: obtaining first data, the first data indicating the demand of a first task for a single computing power in a to-be-assessed period, and the first task comprising one or more to-be-executed tasks; acquiring second data, wherein the second data indicates the demand quantity of the heterogeneous single computing power of the first task in the to-be-evaluated period; and based on the first data and the second data, determining the total computing power demand of the first task in the to-be-evaluated period.

Description

Technical Field

[0001] This application relates to, but is not limited to, the field of cloud computing technology, and in particular to a computing power evaluation method, apparatus, device, storage medium, and computer program product. Background Technology

[0002] Cloud service computing nodes provide computing resources to meet the computing power requirements of tasks or objects to be executed. Since the demand for computing power varies with the stage of computation or the computing environment, it is particularly important for cloud service computing power providers to accurately assess the changing trends in the computing power demands of tasks to be executed, and to provide appropriate computing resources to adapt to these changes.

[0003] In related technologies, the total demand for computing resources is predicted to assess the total computing power demand for the next cycle. However, these technologies neglect the type of computing power required by the tasks to be executed, making it impossible to accurately assess the demand for specific types of computing power when facing different types of computing power needs. This results in the problem of oversupply or undersupply of specific types of computing power resources. Summary of the Invention

[0004] This application provides a computing power evaluation method, apparatus, device, storage medium, and program product.

[0005] The technical solution of this application embodiment is implemented as follows:

[0006] A computing power evaluation method, the method comprising:

[0007] Acquire first data, which indicates the single computing power requirement of a first task during the evaluation period, the first task including one or more tasks to be executed;

[0008] Obtain second data, which indicates the heterogeneous single computing power requirement of the first task during the evaluation period;

[0009] Based on the first data and the second data, the total computing power requirement of the first task in the evaluation period is determined.

[0010] In the above scheme, obtaining the first data includes:

[0011] Obtain third data, which indicates the historical computing power requirement of a single task for the second task, which includes one or more historically executed tasks;

[0012] Obtain fourth data, which indicates the historical average demand for a single computing power for the second task;

[0013] Obtain fifth data, which indicates the overall deviation of the historical demand for a single computing power of the second task from the historical average demand for that single computing power;

[0014] The first data is determined based on the third data, the fourth data, and the fifth data.

[0015] In the above scheme, obtaining the fourth data includes:

[0016] Obtain sixth data, which indicates the number of tasks in the second task having the single computing power;

[0017] The fourth data is determined based on the third data and the sixth data.

[0018] In the above scheme, obtaining the fifth data includes:

[0019] The historical average demand of the single computing power of the k second tasks is subtracted from the historical demand of the single computing power of the k second tasks, and the sum of the resulting k differences is used as the fifth data.

[0020] In the above scheme, obtaining the second data includes:

[0021] Obtain the seventh data, which indicates the historical demand for heterogeneous computing power for the second task;

[0022] Based on the first type, the computing power of the seventh data is equivalently evaluated to determine the second data, where the first type indicates a preset single computing power type.

[0023] In the above scheme, determining the second data includes:

[0024] Obtain the eighth data, which indicates the ratio of the historical demand for a single heterogeneous computing power to the historical demand for the total heterogeneous computing power in the historical demand of the heterogeneous computing power of the second task.

[0025] The second data is determined based on the seventh and eighth data.

[0026] In the above scheme, determining the total computing power requirement of the first task in the evaluation period based on the first data and the second data includes:

[0027] The sum of the first data and the second data is used as the total computing power requirement of the first task in the evaluation period.

[0028] This application embodiment also provides a computing power evaluation device, the computing power evaluation device comprising:

[0029] An acquisition unit is used to acquire first data, which indicates the single computing power requirement of a first task in the evaluation period, and the first task includes one or more tasks to be executed.

[0030] An acquisition unit is used to acquire second data, which indicates the heterogeneous single computing power requirement of the first task in the evaluation period.

[0031] The processing unit is configured to determine the total computing power requirement of the first task in the evaluation period based on the first data and the second data.

[0032] This application also provides a computing power evaluation device, including: a processor and a memory for storing computer programs capable of running on the processor; wherein,

[0033] The processor is used to execute the steps of the aforementioned computing power evaluation method when running the computer program.

[0034] This application also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the aforementioned computing power evaluation method.

[0035] This application also provides a computer program product, including a computer program that can be executed by a processor of an electronic device to complete the steps of the aforementioned computing power evaluation method.

[0036] This application provides a computing power assessment method, apparatus, device, storage medium, and program product. The method includes: acquiring first data, whereby the first data indicates the single computing power requirement of a first task during an assessment period, the first task including one or more tasks to be executed; acquiring second data, whereby the second data indicates the heterogeneous single computing power requirement of the first task during the assessment period; and determining the total computing power requirement of the first task during the assessment period based on the first data and the second data. Specifically, this application, based on the type of computing power requirement of the first task, predicts the single computing power requirement of tasks with single computing power needs for specific types during the assessment period, and predicts the heterogeneous single computing power requirement of tasks with heterogeneous computing power needs for specific types during the assessment period, thereby obtaining the total computing power requirement of the first task for specific types during the assessment period. This solves the problem in related technologies where it is impossible to accurately assess the demand of tasks to be executed for specific types of computing power, resulting in excessive or insufficient supply of specific types of computing power resources. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating a computing power evaluation method according to an embodiment of this application;

[0038] Figure 2 This is a flowchart illustrating a computing power evaluation method in a real-world scenario according to an embodiment of this application.

[0039] Figure 3 This is a schematic diagram of the structure of a computing power evaluation device according to an embodiment of this application;

[0040] Figure 4 This is a schematic diagram of the structure of a computing power evaluation device according to an embodiment of this application. Detailed Implementation

[0041] The present application will now be described in further detail with reference to the accompanying drawings and embodiments.

[0042] Computing power, or computing capability, refers to the ability to process information and data and output target results. Traditionally, computing power relies on single-point computing resources, which has limited capabilities. This limits the computational capacity of these single-point resources for tasks exceeding their capabilities, thus hindering the development of computer technology and the socio-economic landscape. To improve the computing power of traditional resources, the concept of "computing power" has emerged. As a distributed computing approach, computing power packages and aggregates numerous fragmented computing resources to achieve low-cost computing while improving computational performance and reliability.

[0043] Currently, technical personnel utilize cloud services provided by network devices, based on distributed cloud service computing power nodes, to provide different types of computing power resources for tasks to be executed, in order to meet the different types of computing power resource requirements of the tasks or objects to be executed. The demand for computing power resources can be referred to as computing power requirement or computing power demand quantity. However, because the computing power requirement of the tasks or objects to be executed changes at different stages of computation or under different computing environments, if cloud service computing power providers continue to use the traditional method of providing single, fixed computing power resources, they cannot respond to changes in computing power demand in a timely manner, resulting in wasted or insufficient computing power resources. Based on this, related technologies propose trend prediction of computing power demand changes for computing power assessment. However, when assessing cloud service computing power demand, these technologies are based on changes in the overall demand for computing power resources, ignoring the impact of the different computing power resource requirements of the tasks to be executed on the computing power assessment results. This leads to inaccurate computing power assessment results when facing different types of computing power demand.

[0044] Based on this, this application provides a computing power evaluation method, referring to Figure 1 As shown, the method includes the following steps:

[0045] Step 101: Obtain the first data.

[0046] The first data indicates the computing power requirement of a single task during the evaluation period, and the first task includes one or more tasks to be executed.

[0047] In this application embodiment, the executing entity is the computing power provider, that is, the cloud service computing power provider, including the cloud service computing power platform, which can be called a cloud service platform or cloud computing service platform, including but not limited to computing power, storage capacity, and network capacity, and capable of providing computing resources and services such as virtual machines, storage space, databases, and applications.

[0048] The period to be evaluated is the next predictive evaluation period, which can be represented by T.

[0049] Among them, computing power refers to the ability to perform data operations per second when using computing power equipment to provide computing resources. Single computing power refers to the computing power when using a single type of computing power equipment to provide computing resources, and can be called single-type computing power or single-feature computing power.

[0050] Among them, computing power devices include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), data processing units (DPUs), instruction processing units (IPUs), field-programmable gate arrays (FPGAs), and application-specific integrated circuits (ASICs).

[0051] The first task is the task to be executed, which is the task to be executed during the evaluation period.

[0052] The first task, i.e., the task to be executed, indicates the degree to which the task needs computing power resources provided by a single type of computing power device. Understandably, the greater the demand for single computing power, the higher the degree to which the task needs computing power resources provided by a single type of computing power device, and the higher the number of data operations that the computing power device can perform per second. Understandably, when a task to be executed requires only a single type of computing power within a cycle, the task can be called a task with only a single computing power requirement; in this case, the type of computing power required by the task to be executed is single-type.

[0053] In practical applications, the demand for a single computing power for the first task during the evaluation period is predicted to obtain the demand for a single computing power for the first task during the evaluation period.

[0054] It should be noted that the computing power type of a single computing power obtained in this application embodiment is a pre-selected or specific type, which can be called a specific type or a specific computing power type. The specific type can be a pre-set type or a type determined in real time according to the dynamic computing power requirements of the task during the actual operation. The specific type is set according to the actual needs, and this application embodiment does not make specific limitations on this.

[0055] Step 102: Obtain the second data.

[0056] The second data indicates the demand for heterogeneous single computing power for the first task during the evaluation period.

[0057] Heterogeneous single computing power refers to a specific type of computing power within heterogeneous computing power. Heterogeneous computing power refers to the computing capability that provides computing resources (i.e., computing power resources) by utilizing different types of computing power devices through parallel computing; it can be called heterogeneous type computing power or heterogeneous computing power type computing power. Heterogeneous computing power encompasses multiple types of computing power.

[0058] The first task, i.e., the task to be executed, indicates the degree to which the task needs a specific type of computing power resource provided in parallel by different types of computing power devices during the evaluation period. Understandably, the greater the demand for heterogeneous single computing power, the higher the degree of demand for computing power resources provided by a single type of computing power device within the heterogeneous computing power, and the higher the number of data operations that computing power device can perform per second. Understandably, when a task to be executed requires different types of computing power within a period, the task can be called a task with heterogeneous computing power requirements; in this case, the type of computing power required by the task to be executed within a period is heterogeneous.

[0059] In practical applications, the demand for heterogeneous single computing power of the first task during the evaluation period is predicted to obtain the demand for heterogeneous single computing power of the first task during the evaluation period.

[0060] It should be noted that the computing power type of the heterogeneous single computing power obtained in the embodiments of this application is a pre-selected or specific type, which can be called a specific type. The specific type can be a pre-set type or a type determined in real time according to the dynamic computing power requirements of the task during the actual operation. The specific type is set according to the actual needs, and the embodiments of this application do not make specific limitations on this.

[0061] It should be noted that since heterogeneous computing power includes various types of computing power, it is possible to obtain the computing power requirements of different types of heterogeneous computing power for the first task during the evaluation period, and select a specific type of computing power that is consistent with the computing power type of the aforementioned single computing power to determine the heterogeneous single computing power; or it is possible to directly obtain the computing power requirements of a specific type of heterogeneous computing power for the first task during the evaluation period based on the specific type of computing power, thereby determining the heterogeneous single computing power. It can be set according to actual needs, and this application does not make specific limitations on this.

[0062] Step 103: Based on the first data and the second data, determine the total computing power requirement of the first task in the evaluation period.

[0063] In practical applications, the computational power requirements of one or more first tasks to be executed during the evaluation period are obtained, along with the computational power requirements of heterogeneous single computing power. The heterogeneous single computing power is of the same type as the single computing power, thereby determining the total computational power requirement of the first task for each type of computing power during the evaluation period. In other words, for different types of computing power within the heterogeneous computing power, the computational power requirements of each type and the heterogeneous single computing power requirements are determined separately, thereby determining the total computational power requirement of the first task for each type of computing power during the evaluation period.

[0064] In practical applications, based on a specific type (e.g., type A), we can determine the demand for type A computing power for tasks with only a single type of computing power, and the demand for type A computing power among different types of computing power under heterogeneous computing power for tasks with heterogeneous computing power, thereby determining the total computing power demand of the first task for that specific type of computing power in the evaluation period.

[0065] In practical applications, based on another specific type (e.g., type B), we can determine the demand for type B computing power for tasks with only a single type of computing power, and the demand for type B computing power among different types of computing power under heterogeneous computing power for tasks with heterogeneous computing power, thereby determining the total computing power demand of the first task for the other specific type of computing power in the evaluation period.

[0066] This application provides a computing power assessment method, comprising: acquiring first data, the first data indicating the single computing power requirement of a first task in an assessment period, the first task including one or more tasks to be executed; acquiring second data, the second data indicating the heterogeneous single computing power requirement of the first task in the assessment period; and determining the total computing power requirement of the first task in the assessment period based on the first data and the second data. Specifically, this application, based on the type of computing power requirement of the first task, predicts the single computing power requirement of a specific type for tasks with single computing power requirements in the assessment period for different types of computing power, and predicts the heterogeneous single computing power requirement of a specific type for tasks with heterogeneous computing power requirements in the assessment period, thereby obtaining the total computing power requirement of the first task for a specific type in the assessment period. This solves the problem in related technologies that it is impossible to accurately assess the requirement of a task to be executed for a specific type of computing power, resulting in an oversupply or undersupply of specific types of computing power resources.

[0067] In one embodiment, the method for obtaining the first data in step 101 further includes:

[0068] Obtain third data, which indicates the historical computing power requirement of a single task in the second task, which includes one or more historically executed tasks.

[0069] Obtain the fourth data point, which indicates the historical average computing power requirement for the second task.

[0070] Obtain the fifth data point, which indicates the overall deviation of the historical demand for a single computing power for the second task from the historical average demand for a single computing power.

[0071] Based on the third, fourth, and fifth data, the first data is determined.

[0072] The second task is the task that has been executed in the historical evaluation period, which can be called the executed task. The historical evaluation period can be the previous n evaluation periods T.

[0073] Among them, the historical demand for single computing power of the executed task corresponding to the third data refers to the demand for single computing power of the executed task in the first n evaluation periods T in history. The historical demand for single computing power reflects the changes in the demand for single computing power of the executed task in the historical period.

[0074] Among them, the historical average demand of single computing power for the executed tasks corresponding to the fourth data refers to the average demand of single computing power for the executed tasks in the first n evaluation periods T in history. This historical average demand of single computing power reflects the average demand of executed tasks for single computing power in the historical period.

[0075] The overall deviation refers to the degree of deviation between the computing power requirement of a task with a single computing power need and the average computing power requirement of multiple different tasks with the same single computing power need for that single computing power. A task with a single computing power need can also be called a task with a single type of computing power need, or a single-type task, corresponding to the single-type task computing power requirement. The average computing power requirement of multiple different tasks with a single computing power need for that single type can be called the single-type task average computing power requirement or the task average computing power requirement. The overall deviation can take the values ​​of negative, positive, and zero. A negative overall deviation indicates that the demand for a single computing power deviates downward from the average demand for that single computing power. In this case, the actual computing power demand for a task with a specific type of single computing power should be less than the average computing power demand. A positive overall deviation indicates that the demand for a single computing power deviates upward from the average demand for that single computing power. In this case, the actual computing power demand for a task with a specific type of single computing power should be greater than the average computing power demand. A zero overall deviation indicates that the demand for a single computing power does not deviate from the average demand for that single computing power. In this case, the actual computing power demand for a task with a specific type of single computing power should be equal to the average computing power demand.

[0076] In practical applications, the demand for a single computing power for one or more historically executed tasks in the previous n evaluation periods T is obtained. Based on the changing trend of the historical demand for that single computing power, trend analysis is performed to determine the estimated value of the single computing power demand for the task to be executed in the evaluation period. The estimated value can also be called the evaluation quantity.

[0077] In practical applications, the average demand of one or more historically executed tasks for a single computing power over the previous n evaluation periods T is obtained. Based on the changing trend of the historical average demand of the single computing power, trend analysis is performed to determine the evaluation amount of the average demand of the single computing power for the task to be executed in the evaluation period.

[0078] In practical applications, the trend of the overall deviation of the demand for a single computing power for one or more historically executed tasks relative to the historical average demand is obtained in the first n evaluation periods T. Trend analysis is performed, and based on the overall deviation, the overall deviation of the single computing power corresponding to the task to be executed in the evaluation period is determined.

[0079] In practical applications, the assessment of the single computing power requirement of the task to be executed during the assessment period is determined by comprehensively considering the assessment of the single computing power requirement of the task to be executed during the assessment period, the assessment of the average single computing power requirement, and the corresponding overall deviation.

[0080] In this embodiment, by analyzing the changing trend of the demand for a single computing power in historical evaluation periods, the demand for a single computing power in the period to be evaluated is predicted. Furthermore, by analyzing the changing trend of the average computing power demand for a single computing power and the overall deviation based on the average computing power demand, the deviation trend and degree of the demand for a single computing power in the period to be executed relative to the average computing power demand are predicted. This allows for a more refined adjustment of the predicted single computing power demand based on the predicted average computing power demand when evaluating the demand for a specific type of single computing power for a task to be executed, thereby enabling adjustments to the demand for a specific type of computing power based on the overall deviation and improving the accuracy of the assessment of the demand for a single computing power.

[0081] Furthermore, in one embodiment, obtaining the fourth data includes:

[0082] Obtain the sixth data point, which indicates the number of tasks in the second task with a single computing power;

[0083] Based on the third and sixth data, the fourth data is determined.

[0084] In practical applications, the second task refers to one or more tasks that have been executed in the historical evaluation period, and the second task is a task with only a single computing power. The number of one or more tasks that have been executed in the past is obtained, along with the demand for single computing power for the executed tasks in the previous n evaluation periods T. Based on the historical demand for single computing power and the number of tasks for the executed tasks, the historical average demand for single computing power for the executed tasks is determined. This historical average demand reflects the average demand for single computing power by the executed tasks in the historical period.

[0085] In one scenario, the fourth data point can be obtained by dividing the historical demand for a single computing power from the executed tasks by the number of tasks. Understandably, when the historical demand for a single computing power from executed tasks is the same, the more executed tasks there are, the less the average demand for a single computing power each task can be allocated; conversely, when the historical demand for a single computing power from executed tasks is the same, the fewer executed tasks there are, the more the average demand for a single computing power each task can be allocated.

[0086] Because the computational power requirements of tasks in actual demand are uncertain, the tasks requiring a single computational power are also discontinuous, resulting in an uncertainty in the number of tasks. In this embodiment, the number of executed tasks with a single computational power requirement is statistically analyzed. Based on this number, the computational power resources that each task can be allocated under the influence of the number of tasks are analyzed. This yields the historical average computational power requirement of executed tasks. This avoids the uncertainty of the number of tasks affecting the assessment of the average computational power requirement of tasks to be executed, allowing for a more realistic trend analysis of the average computational power requirement based on the number of tasks, thus improving the accuracy of the assessment of the average computational power requirement.

[0087] Furthermore, in one embodiment, obtaining the fifth data includes:

[0088] Subtract the historical average demand of the single computing power of the k second tasks from the historical demand of the single computing power of the k second tasks, and use the sum of the k differences as the fifth data.

[0089] Where k indicates the number of tasks that have been executed in the past, and k is a positive integer greater than or equal to 1.

[0090] In practical applications, since the second task includes a historically executed task, the historical demand of a single computing power for an executed task is subtracted from the historical average demand of that single computing power for that executed task. This yields the deviation difference between the historical demand of a single computing power for that executed task and the historical average demand of that single computing power. This deviation difference indicates the degree of deviation of the demand of an executed task for a single computing power from the average demand. Furthermore, the second task also includes multiple historically executed tasks. Therefore, by subtracting the historical average demand of the single computing power of the k executed tasks from the historical demand of the single computing power of the k executed tasks, multiple deviations of the historical demand of the single computing power of each of the k executed tasks relative to the historical average demand are obtained. These multiple deviations represent the degree of deviation of the demand of the single computing power of each of the multiple executed tasks from the average demand. In order to further obtain the degree of deviation of the historical demand of the single computing power of all the k executed tasks relative to the historical average demand, the obtained k differences are statistically analyzed and summed. The cumulative sum is taken as the degree of deviation of the historical demand of the single computing power of all the k executed tasks relative to the historical average demand, which is called the overall deviation, thus obtaining the aforementioned fifth data.

[0091] In this embodiment of the application, the overall deviation is determined by calculating the sum of the differences between the historical demand of a single computing power of an executed task and the historical average demand of the single computing power of the executed task. This overall deviation can more accurately describe the overall deviation of the historical demand of the single computing power of multiple executed tasks relative to the historical average demand when there are multiple executed tasks, thereby improving the accuracy of the overall deviation.

[0092] Furthermore, in one embodiment, obtaining the second data in step 102 includes:

[0093] Obtain the seventh data point, which indicates the historical demand for heterogeneous computing power for the second task;

[0094] Based on the first type, the computing power of the seventh data is equivalent to determine the second data. The first type indicates a preset single computing power type.

[0095] Heterogeneous computing power refers to the computing capability that provides computing resources (i.e., computing power resources) by utilizing different types of computing power devices through parallel computing. It can be called heterogeneous type computing power or heterogeneous computing power type computing power. Heterogeneous computing power includes multiple types of single computing power.

[0096] The first type corresponds to a single computing power type, which is a pre-selected or predefined specific type. It serves as the benchmark type for computing power equivalence. Computing power equivalence refers to the standardization of different computing power types within heterogeneous computing power based on this first type, achieving unified planning for different types of computing power in heterogeneous computing power. The first type can be called the planned computing power type. Understandably, after computing power equivalence, different types of computing power in heterogeneous computing power can be uniformly represented based on the first type, ensuring that different types have the same data statistical units.

[0097] It should be noted that different types of computing power can be used as the basis for computing power statistics. The type of computing power to be planned can be determined according to actual needs. This application does not make any specific restrictions on this.

[0098] In practical applications, heterogeneous computing power refers to the computing power requirements of various types of computing power for tasks to be executed within the same time period, and these types of computing power exhibit certain correlations. Therefore, this embodiment first determines a first type, obtains historical computing power demand data for heterogeneous computing power for the first n evaluation and prediction periods T, and then uses the determined first type as the baseline type to perform computing power equivalence on the historical demand of different computing power types under heterogeneous computing power, determining the heterogeneous computing power based on the first type equivalence, thereby obtaining the historical demand of heterogeneous computing power. Based on the changes in the historical demand of heterogeneous computing power, the demand of heterogeneous computing power in the evaluation period is trend-predicted to obtain the demand of heterogeneous single computing power for the task to be executed in the evaluation period, thereby obtaining the aforementioned second data.

[0099] This application embodiment comprehensively considers the correlation and heterogeneity between various types of heterogeneous computing power, determines a first type as the computing power equivalence benchmark, and unifies the historical computing power demand of heterogeneous computing power based on the first type, so that the data statistical units of the historical demand of different types of heterogeneous computing power are the same, and has the data statistical basis for heterogeneous single type computing power in heterogeneous types.

[0100] Further, in one embodiment, determining the second data includes:

[0101] Obtain the eighth data, which indicates the ratio of the historical demand for a single heterogeneous computing power to the historical demand for the total heterogeneous computing power in the historical demand for heterogeneous computing power of the second task.

[0102] Based on the seventh and eighth data, the second data is determined.

[0103] Among them, the historical demand for heterogeneous total computing power refers to the sum of the demand for different types of computing power in heterogeneous computing power for executed tasks, which can be called the total demand for different types of computing power in heterogeneous computing power.

[0104] In practical applications, the total demand for different types of computing power from heterogeneous computing power by executed tasks is obtained, along with the historical demand for a specific type of computing power within the heterogeneous computing power. The ratio of this historical demand to the total demand for all types of computing power in the heterogeneous computing power is then calculated, resulting in the eighth data point. The ratio corresponding to this eighth data point represents the proportion of each type of computing power demanded by the executed tasks within the total heterogeneous computing power demand. This yields the assessed value of the ratio of each type of computing power demanded by the task to be executed within the evaluation period. Understandably, a higher ratio for a specific type of computing power indicates a higher demand for that specific type of computing power from the heterogeneous computing power by the executed tasks.

[0105] In practical applications, after obtaining historical computing power demand data for the first n evaluation and prediction periods T, the historical demand of different computing power types under heterogeneous computing power is equivalentized using the determined first type as the baseline type. This determines the heterogeneous computing power based on the first type equivalence, thus obtaining the historical demand of heterogeneous computing power corresponding to the seventh data point. Based on the changes in the historical demand of heterogeneous computing power, the total demand of different types of heterogeneous computing power in the evaluation period is predicted to obtain the estimated demand of the total heterogeneous computing power for the task to be executed in the evaluation period. Furthermore, based on the estimated demand of the task to be executed for the total heterogeneous computing power in the evaluation period and the estimated value of the ratio of the demand for each type of computing power in the heterogeneous computing power to the total heterogeneous computing power demand, the estimated demand of the task to be executed for each type of heterogeneous computing power based on the first type equivalence is determined, thus obtaining the demand of the task to be executed for the heterogeneous single computing power in the evaluation period corresponding to the aforementioned second data point.

[0106] This application's embodiments calculate the ratio of different types of computing power to the total heterogeneous computing power within a historical evaluation period, obtaining ratios for multiple historical evaluation periods. By analyzing the changing trends of these historical ratios, the evaluation value of that ratio in the current evaluation period is predicted. In this way, the correlation between different types of computing power within heterogeneous computing power is quantitatively reflected through these ratios, achieving decoupling between different types of computing power and improving the evaluation accuracy of specific computing power types within heterogeneous computing power.

[0107] Further, in one embodiment, step 103, based on the first data and the second data, determines the total computing power requirement of the first task in the evaluation period, including:

[0108] Add the first data to the second data, and use the sum as the total computing power requirement of the first task in the evaluation period.

[0109] In practical applications, the estimated demand for a specific type of computing power for the task to be executed during the evaluation period is added to the estimated demand for the same type of heterogeneous computing power within the heterogeneous computing power pool. The sum of these two values ​​is taken as the total computing power demand for that specific type of computing power for the task to be executed during the evaluation period. In this way, the total computing power demand for a specific type of computing power for the task to be executed can be accurately assessed.

[0110] In practical applications, the estimated computing power requirement for each individual task under different types in the evaluation period is added to the estimated computing power requirement for heterogeneous individual computing power of the same type as the individual task. The sum is then used as the total computing power requirement for the task under different types in the evaluation period. In this way, the total computing power requirement can be accurately estimated for each type of computing power required by the task under different types.

[0111] In this embodiment, based on the computing power requirements of the task to be executed, a single computing power trend analysis is performed on the historical demand of a single type of computing power for the executed task. Based on the single computing power trend analysis results, the demand for a single computing power is predicted, thereby determining the estimated demand for a single computing power by the task to be executed, thus achieving a reasonable prediction of the demand for a single computing power. Furthermore, a trend analysis of the historical demand of heterogeneous computing power for the executed task is performed on the total heterogeneous computing power, determining the trend analysis results. Combined with the ratios of the demand for different types of computing power under heterogeneous computing power to the total demand for heterogeneous computing power, the estimated demand for a specific type of heterogeneous single computing power within the heterogeneous computing power is determined based on the trend analysis results and ratios of the total heterogeneous computing power. Therefore, for different computing power types, the estimated demand for different types of heterogeneous single computing power within the heterogeneous computing power can be determined, thus achieving a reasonable prediction of the demand for different types of computing power within the heterogeneous computing power type.

[0112] In this embodiment, considering both single and heterogeneous computing power, and based on the specific types of computing power requirements of the tasks to be executed, the estimated quantities of single computing power and heterogeneous single computing power under specific types are statistically analyzed to reasonably predict the total demand for specific types of computing power in cloud services. Furthermore, based on different types of computing power requirements of the tasks to be executed, the estimated quantities of single and heterogeneous single computing power requirements under different types are statistically analyzed to reasonably predict the total demand for different types of computing power in cloud services. This enables cloud service computing power providers to rationally allocate and adjust computing power resources when facing different types of computing power demands. Moreover, the trend analysis of computing power demand in this embodiment can adapt to changes in the computing power demands of different types of tasks, providing accurate and reliable data support for cloud service computing power providers.

[0113] The following section provides a more detailed description of this application with reference to application examples.

[0114] In a real-world scenario, referencing Figure 2 As shown, a computing power evaluation method is proposed, and the details are as follows:

[0115] S201. Obtain historical demand data for a single type of computing power for cloud service computing power tasks.

[0116] Among them, computing power can be referred to as cloud service computing power; a single type indicates a type of computing power, which can be referred to as a single computing power type; a single type of computing power indicates computing power resources of a single computing power type, and computing power resources can also be referred to as computing power provision or computing power supply; the historical demand data obtained in the embodiments of this application is the demand for computing power resources of a single computing power type.

[0117] In practical applications, an evaluation and prediction period T is set. For a single computing power type to be evaluated (corresponding to the specific computing power type mentioned above, the same below), the computing power demand of the single computing power type in the first n evaluation and prediction periods T in the historical time is obtained. For different computing power types, the computing power demand of each computing power type in the first n evaluation and prediction periods T in the historical time is obtained.

[0118] This application embodiment analyzes the overall demand trend of cloud service computing power by setting an assessment and prediction period T, which can better align with the regularly changing computing power market demand. Correspondingly, cloud service computing power providers also periodically adjust their computing power supply based on this assessment and prediction period T, which can better adapt to the cyclical changes in computing power market demand, enabling the computing power supply to meet computing power demand while maximizing the rational utilization of computing power resources.

[0119] In this embodiment, the evaluation and prediction period T can be a preset value, which can be set in combination with the changing patterns of computing power market demand and the adjustment period of computing power supply to computing power demand by computing power providers. The specific setting is based on actual needs, and this application does not limit it.

[0120] In this embodiment of the application, as much historical demand data as possible is obtained within the historical assessment and prediction period T, so as to further improve the accuracy of the data analysis results through a large sample size.

[0121] S202. Determine the curve data of computing power demand for a single type of computing power.

[0122] In practical applications, the curve data of computing power demand for a single type of computing power is determined based on historical demand data of a single type of computing power.

[0123] In practical applications, for a single target computing power type, historical data on the computing power demand for that type of computing power within the first n evaluation and prediction periods T in the historical timeframe is obtained. Then, a curve is established with the period T as the unit of time, with time as the horizontal axis and computing power demand as the vertical axis. This curve represents the change in the computing power demand of cloud service computing power tasks for that type of computing power over time as the period progresses. For different computing power types, historical data on the computing power demand for each corresponding type of computing power within the first n evaluation and prediction periods T in the historical timeframe is obtained separately. Then, curve data corresponding to each computing power type is established separately. These different curve data represent the change in the computing power demand of cloud service computing power tasks for different types of computing power over time as the period progresses. Thus, curve data on the computing power demand of a single type of cloud service computing power task is obtained. This curve data is used to describe the trend change of a single type of cloud service computing power over time.

[0124] S203. Perform trend analysis on the curve data of computing power demand for a single type of computing power to determine the computing power demand function for a single type of computing power.

[0125] In practical applications, based on the curve data of computing power demand for a single type of computing power, an overall demand trend analysis of the single type of computing power is conducted to determine the computing power demand function for the single type of computing power.

[0126] In practical applications, based on the historical demand curves of different types of computing power, numerical simulation analysis is performed on the curves showing how the demand for each single type of computing power changes over periodic time, in order to determine the computing power demand function for each single type of computing power. Where t represents time, and i represents the number of the different computing power types provided by the cloud service platform. It can also be called a single-type computing power demand function.

[0127] S204. Determine the curve data for the number of tasks with a single type of computing power requirement.

[0128] In practical applications, curve data of the number of tasks with computing power requirements of a single type are determined based on historical demand data of a single type of computing power.

[0129] Among them, the computing power requirement of a single type of computing power is called the single-type computing power requirement. Since there are one or more tasks with a single type of computing power requirement, in order to determine the overall usage of computing power resources of the target single type of computing power over a historical period, the number of different tasks with a single type of computing power requirement is obtained.

[0130] In practical applications, for historical demand data of a single computing power type, the number of tasks with a single type of computing power demand is obtained within the first n evaluation and prediction periods T in the historical timeframe. Then, a curve is established with T as the unit of time, with time as the horizontal axis and the number of tasks as the vertical axis. This curve represents the change in the number of tasks calling upon the target single computing power type's computing power resources over time as the periodic time progresses. For historical demand data of different computing power types, the number of tasks corresponding to each type of computing power demand is obtained within the first n evaluation and prediction periods T in the historical timeframe. Then, curves are established for the number of tasks corresponding to each type of computing power demand. Each curve represents the change in the number of tasks corresponding to each computing power type when different cloud service computing power tasks call upon different types of computing power resources over time as the periodic time progresses. This yields curve data on the number of tasks with a single type of computing power demand, which describes the trend of the number of tasks with a single type of computing power demand over time.

[0131] S205. Perform trend analysis on the curve data of the number of tasks to determine the task quantity function for a single type of computing power.

[0132] In practical applications, based on the curve data of the number of tasks with a single type of computing power requirement, a trend analysis of the number of tasks with a single type of computing power is performed to determine the task quantity function of a single type of computing power.

[0133] In practical applications, for the established curve data of the number of tasks with computing power requirements of a single type of computing power, numerical simulation analysis is performed on the curve data of the number of tasks for each single type of computing power changing with periodic time, to determine the task quantity function for each single type of computing power. Where t represents time, and i represents the number of the different computing power types provided by the cloud service platform.

[0134] The embodiments of this application can determine the number of tasks in the next prediction and evaluation period based on the task quantity function of a single type of computing power, realize the prediction of the number of tasks, and further adjust the prediction and evaluation results of the total computing power demand based on the computing power demand that may be required under the predicted number of tasks, thereby reducing the impact of the number of tasks on the prediction and evaluation results and improving the accuracy of the prediction and evaluation.

[0135] S206. Determine the average demand function for a single type of computing power.

[0136] In practical applications, the average demand function for a single type of computing power is determined based on the computing power demand function and the task quantity function of a single type of computing power.

[0137] In practical applications, under different tasks, the average demand function of a single type of computing power that varies periodically with time can be determined by referring to formula (1). This can also be referred to as the average computing power requirement function for a single type of computing power demand:

[0138]

[0139] Where t represents time, in milliseconds;

[0140] i represents the numbering of different computing power types provided by the cloud service platform;

[0141] It is a function of the average demand of different tasks for a single type of computing power numbered i, used to describe the changing trend of the average demand of different tasks for a single type of cloud service computing power over time.

[0142] Let i be the computing power demand function for a single type of computing power, used to describe the changing trend of computing power demand for a single type of cloud service computing power by different tasks over time;

[0143] Let i be the number of tasks with a single type of computing power, which is used to describe the changing trend of the number of tasks with a single type of cloud service computing power in different tasks over time.

[0144] S207. Determine the deviation function of computing power demand for a single type.

[0145] In practical applications, the deviation function of single-type computing power demand is determined based on the computing power demand of tasks with a single type of computing power requirement and the average demand of different tasks for a single type of computing power.

[0146] Among them, the deviation is the overall deviation, which indicates the degree of deviation of the computing power requirement of a task with a single type of computing power requirement from the average requirement of different tasks for a single type of computing power.

[0147] In practical applications, the overall deviation data of tasks with a single type of computing power demand relative to the average computing power demand of tasks within the first n evaluation and prediction periods T in historical time is obtained. Based on the overall deviation data, a curve is established with time as the horizontal axis and the deviation of the single type of computing power demand as the vertical axis. Numerical simulation analysis is performed on this curve data to determine the deviation function of the single type of computing power demand.

[0148] Specifically, for the first n evaluation and prediction periods T, the following overall deviation is determined in each evaluation and prediction period T:

[0149] First, obtain the number of tasks k with single-type computing power requirements within the assessment and prediction period T, and the computing power requirement of each task corresponding to a single type of computing power. These can be referred to as the single-type task computing power requirement or the task computing power requirement.

[0150] Secondly, based on the average computing power requirement function for a single type of task... The average computing power requirement for a given assessment and forecast period T can be referred to as the average computing power requirement for a single type of task or the average computing power requirement for the entire task.

[0151] The third step is to determine the computing power requirements of the task. and average computing power requirement per task Determine the overall deviation d within the assessment and prediction period T using formula (2). i :

[0152]

[0153] Where i represents the number of the different computing power types provided by the cloud service platform;

[0154] K represents the number of tasks, in units.

[0155] d i The deviation of the computing power requirement of a single type of task with number i within the historical assessment and prediction period T is relative to the average computing power requirement of different tasks with the same single type. It can also be called the deviation of the computing power requirement of a single type of task relative to the average computing power requirement of a single type of task.

[0156] For historical assessment and prediction of computing power requirements for a single type of task within a period T;

[0157] This represents the average computing power requirement for a single type of task within a historical assessment and prediction period T.

[0158] The fourth step is to obtain multiple overall deviations d corresponding to the first n evaluation and prediction periods T in historical time. i The curve data for determining the overall deviation of a single type of computing power is obtained.

[0159] Based on the multiple overall deviations d obtained from the previous n evaluation and prediction periods T in historical time. i Establish a time dimension as the horizontal axis and the overall deviation d as the vertical axis. iThe curve data, with y as the vertical axis, represents the degree of deviation of the computing power requirement of a single type of task (numbered i) from the average computing power requirement of that single type of task over a historical period, as the period progresses.

[0160] The fifth step is to conduct a trend analysis of the overall deviation of a single type of computing power based on the curve data of the deviation, and determine the deviation function of the computing power demand of a single type of computing power.

[0161] For the established curve data of the overall deviation of a single type of computing power, numerical simulation analysis is performed on the curve data of the overall deviation of each single type changing with periodic time to determine the overall deviation function of the computing power demand for each single type of computing power. Where t represents time, and i represents the number of the different computing power types provided by the cloud service platform. It can also be called the deviation function of single-type computing power demand.

[0162] This application embodiment determines the overall deviation based on the average computing power requirement of a task. This includes determining the difference between the actual computing power requirement of each task with only a single type of computing power and the average computing power requirement of the task within multiple evaluation and prediction periods in historical time. These differences are then summarized and combined. Within the evaluation and prediction period, using the task as the unit of measurement, the deviation trend of the actual computing power requirement of each task with a single type of computing power relative to the average computing power requirement of the task is determined. This deviation trend is used as the deviation degree for trend analysis, thereby determining the corresponding deviation degree function.

[0163] This application's embodiment performs trend analysis on deviation, which means analyzing the changes in the actual computing power demand of a task for a single computing power type based on the trend of deviation. Then, based on the calculated average computing power demand of the task, it dynamically adjusts the demand according to the determined changes in the actual computing power demand of the task for a single computing power type, to obtain a reasonable computing power demand for a single type of task and an average computing power demand for a single type of task under actual conditions. It can be understood that when the deviation trend is negative, it indicates that the actual computing power demand of each task with a single computing power type, measured in tasks, should be less than the calculated average computing power demand of the task; when the deviation trend is positive, it indicates that the actual computing power demand of each task with a single computing power type, measured in tasks, should be greater than the calculated average computing power demand of the task.

[0164] The deviation analysis in this application is essentially a trend analysis, which has a certain range of fluctuations. Therefore, by setting a deviation threshold, it can be further determined whether the calculated average computing power requirement of the task needs to be adjusted. It can be understood that when the deviation threshold is reached, the average computing power requirement of the task is adjusted; otherwise, it is not adjusted. This avoids the waste of additional computing resources caused by frequent threshold adjustments and obtains an accurate average computing power requirement of the task. This makes the prediction and evaluation of computing power based on the average computing power requirement of the task more reasonable, thereby obtaining a more accurate evaluation result.

[0165] S208. Determine the computing power adjustment assessment amount for a single computing power type.

[0166] In practical applications, the computing power adjustment assessment quantity for a single type of computing power is determined based on the computing power demand function, the average demand function, and the deviation function of the demand for a single type of computing power.

[0167] In practical applications, the computing power demand function of a single type of computing power that changes periodically over time is obtained. Deviation function of single type computing power demand and the average demand function for a single type of computing power Determine the adjustment assessment amount for each individual computing power type under different computing power types, including:

[0168] First, according to periodicity Determine the total computing power demand for different single types of computing power in the next assessment forecast period T to be predicted.

[0169] The second step, according to Determine the overall deviation of different single types of computing power in the next assessment and prediction period T.

[0170] The third step, according to Determine the average computing power requirement for different single types of computing power in the next assessment and forecast period T.

[0171] The fourth step is to determine the total computing power demand for different single types of computing power in the next assessment and prediction period T. Overall deviation and average computing power requirement per task Refer to formula (3) to determine the single computing power type adjustment assessment amount for different single types of computing power.

[0172]

[0173] Where i represents the number of the different computing power types provided by the cloud service platform;

[0174] Adjust the assessment amount for each single type of computing power in the next assessment and prediction period T;

[0175] The average computing power requirement for different single types of computing power tasks in the next assessment and prediction period T;

[0176] The overall deviation of different single types of computing power in the next assessment and prediction period T;

[0177] This represents the total computing power demand for different single types of computing power in the next assessment and prediction period T.

[0178] in, Indicates to The calculation results are then converted to mean.

[0179] In practical applications, based on The overall trend change predicts the total computing power demand for different single types of computing power in the next assessment forecast period T, and based on... Predict the average computing power requirement for different single types of computing power in the next assessment forecast period T, based on After predicting the overall deviation of different single-type computing power in the next assessment prediction period T, and determining the single-type computing power adjustment assessment amount for different single-type computing power in the next assessment prediction period T, the cloud service platform uses the tasks to be executed as the publishing objects of computing power demand, and allocates or publishes computing power resources to the tasks to be executed based on the adjustment assessment amount.

[0180] In practical applications, since computing power resources are published in units of tasks to be executed, when allocating computing power resources, the non-integer computing power resources can be supplemented and adjusted in combination with the task volume when dividing the total computing power resources, so as to further improve the accuracy of the supply of computing power.

[0181] In one scenario, when the computing power requirement of the task to be executed is continuous, the computing power requirement of each task with only a single type of computing power requirement is clear in the overall trend analysis of computing power requirement for a single type of computing power. Based on this clear computing power requirement, the computing power requirement in the next evaluation and prediction period can be determined.

[0182] In another scenario, when the computing power demand of the tasks to be executed is not continuous but rather presents as discrete computing power demand and its combination embodied in the tasks, when predicting the computing power demand in the next assessment and prediction period, it is necessary to further consider the impact of the characteristics of the computing power demand of the tasks to be executed on the amount of computing power demand, in order to comprehensively determine the amount of computing power demand in the next assessment and prediction period.

[0183] In this embodiment, by using the average computing power requirement of a task as the evaluation basis, considering the deviation between the computing power requirement of different tasks for a single type of computing power and the average computing power requirement of the task, the corresponding deviation degree is calculated, and then the average computing power requirement of the task is adjusted based on the deviation degree to obtain a computing power requirement of a single type of task and an average computing power requirement of a single type of task that are more in line with the actual situation. This avoids ignoring the impact of the deviation on the determination of the total computing power by directly using the average computing power requirement of the task for computing power prediction, and improves the accuracy of the prediction results.

[0184] S209. Obtain historical demand data for heterogeneous computing power for cloud service computing power tasks.

[0185] Among them, heterogeneous computing power can be called heterogeneous computing power, which refers to a computing power system that combines multiple types of computing power for collaborative processing.

[0186] In practical applications, the type of computing power to be planned is first determined. After obtaining the historical computing power demand data of heterogeneous computing power for the first n evaluation and prediction periods T, the determined type of computing power to be planned is used as the benchmark type. Computing power equivalence is performed on different types of computing power under heterogeneous computing power to determine the total amount of heterogeneous computing power planning based on the equivalent type of computing power to be planned, thereby obtaining the historical demand data of heterogeneous computing power.

[0187] Before performing trend analysis on the total amount of heterogeneous computing power, this application embodiment comprehensively considers the correlation and heterogeneity between various types of heterogeneous computing power, determines the benchmark computing power type, and performs unified preprocessing on the historical computing power demand data of heterogeneous computing power based on the benchmark computing power type, so that the historical demand data of heterogeneous computing power is unified data, which has the data statistical basis for trend analysis on the total amount of heterogeneous computing power demand.

[0188] In this embodiment of the application, the planned computing power type is a pre-set benchmark computing power type, which is used to unify the computing power type of heterogeneous computing power. Different computing power types can be used as the basis for computing power statistics. The planned computing power type can be determined according to actual needs. This application does not make specific limitations on this.

[0189] S210, Determine the curve data of computing power demand for heterogeneous computing power.

[0190] In practical applications, the curve data of computing power demand for heterogeneous computing power is determined based on historical demand data of heterogeneous computing power.

[0191] In practical applications, for heterogeneous computing power types, historical demand data for heterogeneous computing power is obtained within the first n evaluation and prediction periods T in historical time. Corresponding curve data are then established for each heterogeneous computing power type. These different curve data represent the change in the demand for heterogeneous computing power by cloud service computing power tasks over time as the periodic time progresses. This yields curve data of the computing power demand for heterogeneous computing power in cloud service computing power tasks, which describes the trend of heterogeneous cloud service computing power changes over time. Understandably, since the historical demand data for heterogeneous computing power is based on data that has been uniformly processed according to the planned computing power type, the resulting curve data can also be called the total planned demand curve data for heterogeneous computing power or the trend change curve data for heterogeneous computing power.

[0192] S211. Perform trend analysis on the curve data of computing power demand for heterogeneous computing power to determine the computing power demand function for heterogeneous computing power.

[0193] In practical applications, based on the curve data of computing power demand for heterogeneous computing power, an overall demand trend analysis of heterogeneous computing power is conducted to determine the computing power demand function for heterogeneous computing power.

[0194] In practical applications, numerical simulation analysis is performed on the curve data of the computing power demand of heterogeneous computing power to determine the total computing power demand function of heterogeneous computing power that changes periodically with time. Where t represents time. It can also be called the total demand function for periodic heterogeneous computing power planning or the total demand function for heterogeneous computing power planning.

[0195] S212. Determine the average type ratio function for heterogeneous computing power tasks.

[0196] In practical applications, the average type ratio function of heterogeneous computing power tasks is determined based on historical demand data of heterogeneous computing power.

[0197] In practical applications, the computing power demand of different tasks for each single computing power type in heterogeneous computing power is first obtained from the historical demand data of heterogeneous computing power. Then, taking the task as the unit of measurement, the average type ratio of heterogeneous computing power tasks in the first n evaluation and prediction periods T is determined with reference to formula (4).

[0198]

[0199] Where i represents the number of the different computing power types provided by the cloud service platform;

[0200] m is the task number;

[0201] The heterogeneous computing power task type ratio refers to the ratio of the computing power demand of task number m for a single computing power type number i in the heterogeneous computing power type to the total computing power of the task corresponding to the heterogeneous computing power type under the evaluation and prediction period T.

[0202] The average type ratio of heterogeneous computing power tasks over the first n evaluation and prediction periods T.

[0203] Furthermore, based on the average type ratio of heterogeneous computing power tasks in the first n evaluation and prediction periods T... Curves showing the periodic variation of the average type ratio of heterogeneous computing power tasks with different computing power types over time were established, forming a function of the average type ratio of heterogeneous computing power tasks that varies periodically over time. It can also be called the average type ratio function of heterogeneous computing power tasks or the average type ratio function of periodic heterogeneous computing power tasks.

[0204] Among them, the total computing power of tasks corresponding to heterogeneous computing power refers to the total demand of tasks with heterogeneous computing power requirements for different types of computing power, which can also be called the total computing power of heterogeneous tasks. This total computing power is determined based on the planned computing power type. For the computing power demand of each single computing power type with different numbers in the heterogeneous computing power type, it is determined based on the aforementioned same planned computing power type.

[0205] In practical applications, when a task has only a single type of computing power requirement, only that single type of computing power requirement is considered for statistical analysis. In this embodiment, the ratio of the computing power requirement of different types of computing power in historical heterogeneous computing power to the total computing power of the corresponding task is statistically analyzed. This ratio quantifies the correlation between different types of computing power in heterogeneous computing power when a task has heterogeneous computing power requirements, improving the accuracy of evaluating a single type of computing power in heterogeneous computing power. Furthermore, when predicting the ratio of different types of computing power in heterogeneous computing power for corresponding tasks, the accuracy of trend prediction of the computing power requirement of different types of computing power in heterogeneous computing power can be improved throughout the entire prediction and evaluation period.

[0206] It is important to note that the total computing power of tasks in the heterogeneous computing power task type ratio is determined based on the planned computing power type to ensure the consistency of computing power demand statistics.

[0207] S213. Determine the single-type adjustment evaluation quantity for heterogeneous computing power.

[0208] In practical applications, the single-type adjustment evaluation quantity of heterogeneous computing power is determined based on the total computing power demand function of heterogeneous computing power and the average type ratio function of heterogeneous computing power tasks.

[0209] In practical applications, according to Q organize (t) and Determine the adjustment assessment quantity for a single type of heterogeneous computing power. This single-type adjustment assessment quantity refers to the assessment quantity of the computing power demand for a target single computing power type within the heterogeneous computing power in the next forecast assessment period, including:

[0210] First, based on the total demand function Q for periodic heterogeneous computing power planning. organize (t), determine the total heterogeneous computing power planning amount Q2 in the next assessment and prediction period T;

[0211] The second step is to use the average type ratio function of periodic heterogeneous computing power tasks of different computing power types under heterogeneous computing power. Determine the ratio of heterogeneous computing power types for different computing power types within the next assessment and prediction period T.

[0212] The third step is to plan the total heterogeneous computing power Q2 and the ratio of heterogeneous computing power types. Refer to formula (5) to determine the heterogeneous equivalent computing power demand for different computing power types under the planned computing power type.

[0213]

[0214] Where i represents the number of the different computing power types provided by the cloud service platform;

[0215] The heterogeneous equivalent computing power demand for different computing power types within the next assessment and prediction period T;

[0216] Q2 represents the total planned heterogeneous computing power within the next assessment and forecast period T.

[0217] The ratio of heterogeneous computing power types for different computing power types within the next assessment prediction period T;

[0218] It is important to note that the heterogeneous equivalent computing power requirement is also based on tasks; therefore, it is necessary to... The settlement result is finalized and adjusted based on the smallest computational unit required by the task for computing power, that is, through calculation. To conduct The calculation results are rounded down to the smallest computing unit.

[0219] The fourth step is to determine the heterogeneous equivalent computing power requirement based on the target single computing power type. The equivalent conversion of computing power yields the single-type adjustment assessment quantity for heterogeneous computing power.

[0220] For example, based on the estimated ratio of the computing power demand of Task 2 (numbered 2) for a single computing power type (numbered i) within the heterogeneous computing power type to the total computing power of the tasks corresponding to the heterogeneous computing power in the next evaluation and prediction period T, the estimated value of the heterogeneous equivalent computing power demand for Task 2 is determined. Then, the computing power demand of the target single computing power type (numbered i) is determined from this estimated value, thus obtaining the evaluation amount of the computing power demand of Task 2 for the target heterogeneous single computing power type (numbered i), i.e., the heterogeneous single computing power type adjustment evaluation amount. It can be understood that by estimating the computing power demand for single types with different numbers, the adjustment evaluation amounts corresponding to different types of computing power in the heterogeneous computing power can be obtained. It should be noted that the determination of the target single computing power type in this embodiment can be set according to actual needs, and the target single computing power type corresponding to the heterogeneous single computing power type has the same number as the aforementioned single computing power type.

[0221] This application embodiment analyzes the trend of the proportion of different types of computing power under heterogeneous computing power to determine the average type ratio function of heterogeneous computing power tasks corresponding to the target single computing power type, and analyzes the trend of computing power demand of heterogeneous computing power to determine the total planned demand function of heterogeneous computing power corresponding to the target single computing power type. Then, based on the determined average type ratio function of heterogeneous computing power tasks and the total planned demand function of heterogeneous computing power, the computing power demand of the task for the target single computing power type in heterogeneous computing power is determined, thereby obtaining the heterogeneous single type computing power demand.

[0222] S214. Determine the total cloud service computing power demand assessment based on the adjustment assessment quantity for single computing power type and the adjustment assessment quantity for single heterogeneous computing power type.

[0223] In practical applications, the evaluation quantity is adjusted according to the specific computing power type. Adjustment of evaluation quantity for single type of heterogeneous computing power Refer to formula (6) to comprehensively determine the total demand assessment for cloud service computing power types:

[0224]

[0225] Where i represents the number of the different computing power types provided by the cloud service platform;

[0226] The total demand for cloud service computing power types within the next assessment and forecast period T is estimated.

[0227] Adjust the assessment amount for each single type of computing power within the next assessment and prediction period T;

[0228] Adjust the assessment amount for heterogeneous computing power of different single types within the next assessment and prediction period T.

[0229] This application embodiment assesses the computing power requirement of a single type of target computing power for tasks with that single type of target computing power during the evaluation period, as well as the computing power requirement of tasks with heterogeneous types of computing power for the single type of target computing power in the heterogeneous computing power. Based on the single type of target computing power, the two are statistically integrated to obtain the total estimated computing power requirement of the task to be executed for the single type of target computing power in the cloud service. It can be understood that different types of computing power can be statistically integrated based on the above two to obtain the total estimated computing power requirement of different types of cloud service computing power, thus realizing the assessment and prediction of the total amount of cloud service computing power.

[0230] This application embodiment assesses the computing power demand of different types, enabling cloud service computing power providers to make reasonable allocations and adjustments to computing power resources according to changes in computing power demand, avoiding excessive or insufficient provision of computing power resources, and achieving rational utilization of computing power resources.

[0231] It should be noted that the application scenarios of this application embodiment include, but are not limited to, medical, education, and transportation scenarios. Among them, medical scenarios include scenarios that require cloud computing power in medical services such as inpatient information statistics, online consultation, and appointment registration.

[0232] To implement the computing power evaluation method of this application embodiment, this application embodiment also provides a computing power evaluation device, referring to... Figure 3 As shown, the computing power evaluation device 300 includes: an acquisition unit 301 and a processing unit 302; wherein,

[0233] The acquisition unit 301 is used to acquire first data, which indicates the single computing power requirement of the first task in the evaluation period. The first task includes one or more tasks to be executed.

[0234] Acquisition unit 301 is used to acquire second data, which indicates the heterogeneous single computing power requirement of the first task in the evaluation period;

[0235] The processing unit 302 is used to determine the total computing power requirement of the first task in the evaluation period based on the first data and the second data.

[0236] in,

[0237] In one embodiment, the acquisition unit 301 is specifically configured to: acquire third data, the third data indicating the historical demand of a single computing power for a second task, the second task including one or more historically executed tasks; acquire fourth data, the fourth data indicating the historical average demand of a single computing power for the second task; and acquire fifth data, the fifth data indicating the overall deviation of the historical demand of a single computing power for the second task from the historical average demand of a single computing power.

[0238] In one embodiment, the processing unit 302 is specifically configured to: determine the first data based on the third data, the fourth data, and the fifth data.

[0239] In one embodiment, the acquisition unit 301 is specifically used to: acquire sixth data, the sixth data indicating the number of tasks of the second task with a single computing power;

[0240] In one embodiment, the processing unit 302 is specifically configured to: determine the fourth data based on the third data and the sixth data.

[0241] In one embodiment, the processing unit 302 is specifically used to: subtract the historical average demand of the single computing power of the k second tasks from the historical demand of the single computing power of the k second tasks, and use the cumulative sum of the resulting k differences as the fifth data.

[0242] In one embodiment, the acquisition unit 301 is specifically used to: acquire seventh data, which indicates the historical demand for heterogeneous computing power of the second task;

[0243] In one embodiment, the processing unit 302 is specifically used to: perform computational power equivalence on the seventh data based on the first type, and determine the second data, wherein the first type indicates a preset single computational power type.

[0244] In one embodiment, the acquisition unit 301 is specifically used to: acquire eighth data, which indicates the ratio of the historical demand of a single heterogeneous computing power to the historical demand of the total heterogeneous computing power in the historical demand of the heterogeneous computing power of the second task.

[0245] In one embodiment, the processing unit 302 is specifically configured to: determine the second data based on the seventh data and the eighth data.

[0246] In one embodiment, the processing unit 302 is specifically configured to: add the first data to the second data, and use the sum as the total computing power requirement of the first task in the evaluation period.

[0247] The descriptions of the same steps and contents as in other embodiments in this example can be found in the descriptions of other embodiments, and will not be repeated here.

[0248] This system utilizes a simple system architecture consisting of a computing power data storage unit and a computing power evaluation and processing unit to perform trend analysis of cloud service computing power demand and prediction of computing power resource demand. It achieves the effect of efficient data analysis and processing with low-cost equipment, providing a reasonable and sufficient hardware foundation for the evaluation of cloud service computing power.

[0249] To implement the computing power assessment method of this application embodiment, this application embodiment also provides a computing power assessment device, referring to... Figure 4 As shown, the computing power evaluation device 400 includes: a processor 401, a memory 402, and a communication bus 403; wherein,

[0250] Processor 401 is used to execute the methods provided by one or more of the above-described technical solutions when running a computer program;

[0251] Memory 402 stores computer programs that can run on processor 401;

[0252] The communication bus 403 is used to implement the communication connection between the processor 401 and the memory 402.

[0253] It should be noted that the specific processing procedure of processor 401 can be understood by referring to the above method, and will not be repeated here.

[0254] Of course, in practical applications, the various components in the computing power evaluation device 400 are coupled together via a communication bus 403. It can be understood that the communication bus 403 is used to achieve communication between these components. In addition to a data bus, the communication bus 403 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 4 The general labeled all buses as communication bus 403.

[0255] The memory 402 in this embodiment is used to store various types of data to support the operation of the computing power evaluation device 400. Examples of such data include any computer program used to operate on the computing power evaluation device 400.

[0256] The methods disclosed in the embodiments of this application can be applied to processor 401, or implemented by processor 401. Processor 401 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 401 or by instructions in the form of software. Processor 401 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 401 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in memory 402. Processor 401 reads the information in memory 402 and combines its hardware to complete the steps of the aforementioned method.

[0257] In an exemplary embodiment, the computing power evaluation device 400 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0258] It is understood that the memory (memory 402) in this embodiment of the application can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Sync Link Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memories described in the embodiments of this application are intended to include, but are not limited to, these and any other suitable types of memories.

[0259] In an exemplary embodiment, this application also provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium, such as a memory 402 storing a computer program. The computer program in the memory 402 can be executed by the processor 401 of the computing power evaluation device 400 to complete the steps of the aforementioned method. The computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface memory, optical disk, or CD-ROM.

[0260] In an exemplary embodiment, this application also provides a computer program product, including a computer program that can be executed by the processor 701 of the computing power evaluation device 700 to complete the steps of the aforementioned method of the computing power evaluation device.

[0261] It should be noted that terms such as "first" and "second" are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0262] Furthermore, the technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0263] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. A computing power evaluation method, characterized in that, The method includes: Acquire first data, which indicates the single computing power requirement of a first task during the evaluation period, the first task including one or more tasks to be executed; Obtain second data, which indicates the heterogeneous single computing power requirement of the first task during the evaluation period; Based on the first data and the second data, the total computing power requirement of the first task in the evaluation period is determined.

2. The method according to claim 1, characterized in that, The acquisition of the first data includes: Obtain third data, which indicates the historical computing power requirement of a single task for the second task, which includes one or more historically executed tasks; Obtain fourth data, which indicates the historical average demand for a single computing power for the second task; Obtain fifth data, which indicates the overall deviation of the historical demand for a single computing power of the second task from the historical average demand for that single computing power. The first data is determined based on the third data, the fourth data, and the fifth data.

3. The method according to claim 2, characterized in that, The acquisition of the fourth data includes: Obtain sixth data, which indicates the number of tasks in the second task having the single computing power; The fourth data is determined based on the third data and the sixth data.

4. The method according to claim 2, characterized in that, The acquisition of the fifth data includes: The historical average demand of the single computing power of the k second tasks is subtracted from the historical demand of the single computing power of the k second tasks, and the sum of the resulting k differences is used as the fifth data.

5. The method according to claim 2, characterized in that, The acquisition of the second data includes: Obtain the seventh data, which indicates the historical demand for heterogeneous computing power for the second task; Based on the first type, the computing power of the seventh data is equivalently evaluated to determine the second data, where the first type indicates a preset single computing power type.

6. The method according to claim 5, characterized in that, Determining the second data includes: Obtain the eighth data, which indicates the ratio of the historical demand for a single heterogeneous computing power to the historical demand for the total heterogeneous computing power in the historical demand of the heterogeneous computing power of the second task. The second data is determined based on the seventh data and the eighth data.

7. The method according to claim 1, characterized in that, The step of determining the total computing power requirement of the first task in the evaluation period based on the first data and the second data includes: The sum of the first data and the second data is used as the total computing power requirement of the first task in the evaluation period.

8. A computing power evaluation device, characterized in that, The device includes: An acquisition unit is used to acquire first data, which indicates the single computing power requirement of a first task in the evaluation period, and the first task includes one or more tasks to be executed. An acquisition unit is used to acquire second data, which indicates the heterogeneous single computing power requirement of the first task in the evaluation period. The processing unit is configured to determine the total computing power requirement of the first task in the evaluation period based on the first data and the second data.

9. A computing power assessment device, characterized in that, The computing power assessment device includes: a processor and a memory for storing computer programs capable of running on the processor; wherein, The processor, when running the computer program, performs the steps of the method according to any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images