Task execution method, electronic equipment, storage medium and program product

Abstract

The invention provides a task execution method which can be applied to the technical field of server energy efficiency improvement. The method comprises the following steps: processing current operation parameters of a server to obtain current operation data; according to the current operation data and first historical operation data of the server, predicting energy consumption and response duration of the server in a predetermined time period to obtain a prediction result; according to the prediction result, adjusting the initial task allocation scheme until an evaluation index based on the adjusted task allocation scheme meets a predetermined index condition, and obtaining a target task allocation scheme; and according to the expected heat load obtained based on the prediction result and the environmental data of the data center where the server is located, controlling cooling equipment associated with the data center to operate until the environmental data of the data center meets a predetermined environmental condition, so that the server executes the target task allocation scheme under the predetermined environmental condition. The invention further provides electronic equipment, a storage medium and a program product.

Description

Technical Field

[0001] The present invention relates to the technical field of improving server energy efficiency, and particularly to a task execution method, an electronic device, a storage medium, and a program product. Background Art

[0002] With the development of data centers, the high energy consumption problem of data centers has been continuously intensifying. By improving the energy efficiency of servers deployed in data centers, the energy utilization rate of data centers can be increased, and the high energy consumption of data centers can be reduced.

[0003] The current server energy efficiency improvement system is difficult to provide reliable and accurate future energy consumption predictions, and the cooling equipment associated with the data center is difficult to adjust its own parameters according to the energy consumption of the server and the environment of the data center, resulting in poor environmental control effects in the data center, shortening the lifespan of the servers deployed in the data center, and reducing the energy utilization rate. The existence of low energy consumption prediction accuracy and low energy utilization rate reduces the execution efficiency of the server energy efficiency improvement system. Summary of the Invention

[0004] In view of the above problems, the present invention provides a task execution method, an electronic device, a storage medium, and a program product.

[0005] One aspect of the present invention provides a task execution method, which includes: in response to an energy consumption prediction request for an initial task allocation scheme, processing the current operating parameters of the server to obtain current operating data; predicting the energy consumption and response duration of the server in a predetermined time period based on the current operating data and the first historical operating data of the server to obtain a prediction result; adjusting the initial task allocation scheme according to the prediction result until the evaluation index based on the adjusted task allocation scheme meets the predetermined index condition to obtain a target task allocation scheme; controlling the operation of the cooling equipment associated with the data center according to the expected heat load obtained based on the prediction result and the environmental data of the data center where the server is located until the environmental data of the data center meets the predetermined environmental condition, so that the server executes the target task allocation scheme under the predetermined environmental condition.

[0006] Another aspect of the present invention further provides an electronic device, including: one or more processors; a memory for storing one or more computer programs, and the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0007] Another aspect of the present invention further provides a computer-readable storage medium, on which a computer program or instruction is stored, and when the computer program or instruction is executed by a processor, the steps of the above method are implemented.

[0008] Another aspect of the present invention also provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps of the above method.

[0009] According to an embodiment of the present invention, in response to an energy consumption prediction request, the energy consumption and response duration of a server in a predetermined time period are predicted based on the current operating data and the first historical operating data of the server to obtain a prediction result; the initial task allocation scheme is adjusted according to the prediction result until a predetermined index condition is met; according to the predicted heat load and the environmental data of the data center, the operation of the cooling device is controlled so that the server executes the target task allocation scheme under predetermined environmental conditions. Since the energy consumption is predicted by combining the current operating data and the first historical operating data during the task execution process, it is beneficial to predict the energy consumption according to the distribution characteristics of the operating data, improve the prediction accuracy of the energy consumption, and provide a basis for adjusting the task allocation scheme under low energy consumption. When adjusting the task allocation scheme, the predetermined index condition can be set as the minimum energy consumption, and then the target task allocation scheme with the minimum energy consumption can be obtained. When controlling the operation of the cooling device, it can be controlled according to the heat load of the server and the environment of the data center, which can achieve precise adjustment, reduce energy waste, and improve energy utilization efficiency. The task execution method of the present invention can reduce energy consumption, improve energy utilization efficiency, and improve the execution efficiency of the energy-efficient server improvement system through the closed-loop control of predicting energy consumption, optimizing the task allocation scheme according to the minimum energy consumption, and collaborative cooling control. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above content and other objects, features, and advantages of the present invention will become clearer through the following description of the embodiments of the present invention with reference to the accompanying drawings.

[0011] Figure 1 The application scenario diagram of the task execution method according to an embodiment of the present invention is shown.

[0012] Figure 2 The flowchart of the task execution method according to an embodiment of the present invention is shown.

[0013] Figure 3 The schematic diagram of adjusting the task allocation scheme of the present invention is shown.

[0014] Figure 4 The flowchart of the task execution method according to another embodiment of the present invention is shown.

[0015] Figure 5 The structural block diagram of the task execution device according to an embodiment of the present invention is shown.

[0016] Figure 6 The block diagram of an electronic device suitable for implementing the task execution method according to an embodiment of the present invention is shown. Detailed Implementation Manner

[0017] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In the following detailed description, for the purpose of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present invention. However, obviously, one or more embodiments can also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present invention.

[0018] The terms used herein are merely for describing specific embodiments and are not intended to limit the present invention. The terms "including", "comprising", etc. used herein indicate the presence of the described features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0019] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0020] In the case of using expressions such as "at least one of A, B, and C, etc.", generally, it should be interpreted according to the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include, but is not limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.).

[0021] The improvement of the energy efficiency of a server refers to improving the overall energy usage efficiency of the server and its data center. With the increasing high energy consumption problem in data centers, improving the intelligent level and security of server energy efficiency has become one of the problems that need to be solved urgently.

[0022] In terms of improving server energy efficiency, existing server energy efficiency improvement systems are difficult to effectively capture the time series characteristics of server operating parameters, resulting in low accuracy of energy consumption prediction, and thus it is difficult to provide reliable future energy consumption prediction. Moreover, the existing cooling management systems in data centers are difficult to make dynamic adjustments according to real-time workloads and environmental conditions, resulting in energy waste or poor temperature control, affecting the equipment life. At the same time, existing server energy efficiency improvement systems lack an effective feedback adjustment mechanism for energy consumption prediction models, task allocation strategies, and cooling management strategies, and are difficult to adapt to changing workloads and environmental conditions.

[0023] In view of this, embodiments of the present invention provide a task execution method, an electronic device, a storage medium, and a program product, which are used to improve the accuracy of energy consumption prediction and energy utilization rate, as well as the execution efficiency of the server energy efficiency improvement system. The task execution method includes: in response to an energy consumption prediction request for an initial task allocation scheme, processing the current operating parameters of the server to obtain current operating data; predicting the energy consumption and response duration of the server in a predetermined time period according to the current operating data and the first historical operating data of the server to obtain a prediction result; adjusting the initial task allocation scheme according to the prediction result until the evaluation index based on the adjusted task allocation scheme meets the predetermined index condition to obtain a target task allocation scheme; controlling the operation of the cooling equipment associated with the data center according to the expected heat load obtained based on the prediction result and the environmental data of the data center where the server is located until the environmental data of the data center meets the predetermined environmental condition, so that the server executes the target task allocation scheme under the predetermined environmental condition.

[0024] Figure 1 FIG. shows an application scenario diagram of the task execution method according to an embodiment of the present invention.

[0025] As Figure 1 shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0026] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc., such as sending an energy consumption prediction request or receiving a task execution result, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as a task allocation application, an energy consumption prediction application, a shopping application, a web browser application, a search application, an instant messaging tool, an email client, a social platform software, etc. (only as examples). In some embodiments, the energy consumption prediction request may not need to be triggered by the user through the terminal device. The energy consumption prediction request may be automatically triggered when the initial task allocation scheme is generated, and the user can only view the task execution process of the embodiments of the present invention through the terminal device.

[0027] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices with a display screen, including but not limited to smartphones, tablets, laptop portable computers, and desktop computers, etc.

[0028] Server 105 may be a server that provides various services. For example, it may be a background management server (only for example) that supports requests sent by the user using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process data such as received requests, and feedback the processing results (such as task execution results, web pages, information, or data obtained or generated according to the requests) to the terminal devices.

[0029] It should be noted that the task execution method provided by the embodiments of the present invention can generally be executed by server 105. Correspondingly, the task execution device provided by the embodiments of the present invention can generally be set in server 105. The task execution method provided by the embodiments of the present invention can also be executed by a server or a server cluster different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the task execution device provided by the embodiments of the present invention can also be set in a server or a server cluster different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0030] It should be understood that Figure 1 the numbers of the terminal devices, networks, and servers in

[0031] are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers. Figure 1 Based on the Figures 2 to 3 scenario described below, the task execution method of the embodiments of the present invention will be described in detail through

[0032] Figure 2 FIG. shows a flowchart of the task execution method according to an embodiment of the present invention.

[0033] As Figure 2 shown, the task execution method of this embodiment includes operations S210 to S240.

[0034] In operation S210, in response to an energy consumption prediction request for the initial task allocation plan, the current operating parameters of the server are processed to obtain current operating data.

[0035] In operation S220, based on the current operating data and the first historical operating data of the server, the energy consumption and response duration of the server in a predetermined time period are predicted to obtain a prediction result.

[0036] In operation S230, according to the prediction result, the initial task allocation scheme is adjusted until the evaluation index based on the adjusted task allocation scheme meets the predetermined index condition, and the target task allocation scheme is obtained.

[0037] In operation S240, according to the expected heat load obtained based on the prediction result and the environmental data of the data center where the server is located, the cooling equipment associated with the data center is controlled to operate until the environmental data of the data center meets the predetermined environmental condition, so that the server executes the target task allocation scheme under the predetermined environmental condition.

[0038] In some embodiments, the initial task allocation scheme may be obtained by randomly allocating tasks to servers according to the relationship between tasks and services. In one example, a data center may have m1 servers, and these m1 servers need to execute m2 tasks in total. Both m1 and m2 are positive integers. When m1 and m2 are equal, the initial task allocation scheme may be that one server can execute one task; when m1 and m2 are not equal, the initial task allocation scheme may be to sequentially and cyclically allocate these m2 tasks to m1 servers in a polling manner.

[0039] In some embodiments, the server may be a device for executing tasks.

[0040] In some embodiments, the current operating parameters may be the operating parameters of the server obtained through the sensor network at the time point of processing the energy consumption prediction request. The operating parameters of the server may include various parameters such as the utilization rate of the central processing unit (CPU for short), memory usage, traffic, temperature, humidity, and power consumption. The current operating data may be obtained by processing the abnormal parameters in the operating parameters and performing a normalization process on the operating parameters after processing the abnormal parameters.

[0041] In some embodiments, the first historical operating data may be the historical operating data randomly extracted from the database storing the historical operating data of the server when predicting the energy consumption of the initial task allocation scheme, such as historical CPU utilization rate, historical memory utilization rate, historical traffic, historical temperature, historical humidity, and historical power consumption.

[0042] In some embodiments, the predetermined time period may be a future time period, for example, any time period after the time point when the energy consumption prediction request is processed. According to the data change trend and periodic characteristics reflected by the current operation data and the first historical operation data of the server, the energy consumption and response duration of the server in the future predetermined time period can be predicted, and the obtained prediction results may include an energy consumption prediction value and a response duration prediction value. In one embodiment, predicting the energy consumption and response duration of the server in the future predetermined time period can be achieved by means of an energy consumption prediction model. The energy consumption prediction model may be constructed based on a deep learning network, such as a long short-term memory neural network for processing and predicting long-term dependencies in time series data.

[0043] In some embodiments, according to the above energy consumption prediction value and response duration prediction value, the initial task allocation scheme can be continuously adjusted until the evaluation index of the adjusted task allocation scheme meets the predetermined index condition or reaches the predetermined number of adjustments. The predetermined index condition may be to ensure that the response duration does not exceed the threshold while minimizing the energy consumption, so as to achieve a balance between the energy consumption and the response duration.

[0044] Figure 3 A schematic diagram of adjusting the task allocation scheme according to the present invention is shown.

[0045] As Figure 3As shown, in one example, a task center may have three servers S1, S2, and S3. These three servers need to execute a total of five tasks T1 to T5. The initial task allocation scheme can be that S1 executes T1 and T2, S2 executes T3, and S3 executes T4 and T5. In this case, the prediction results obtained based on the current running data of the servers and the first historical running data of the servers can be an energy consumption prediction value 1 and a response duration prediction value 1. The evaluation index value obtained based on the energy consumption prediction value 1 and the response duration prediction value 1 is index value 1. Randomly adjust the initial task allocation scheme, for example, adjust it to S1 executes T1 and T3, S2 executes T2 and T4, and S3 executes T5. In this case, the prediction results obtained based on the current running data of the servers and the second historical running data randomly extracted from the database can be an energy consumption prediction value 2 and a response duration prediction value 2. The evaluation index value obtained based on the energy consumption prediction value 2 and the response duration prediction value 2 is index value 2. The process of continuously adjusting the task allocation scheme and calculating the evaluation index value can be looped until the number of loop adjustments reaches a predetermined number of adjustments; or the number of non-repeating task allocation schemes has reached the predetermined number of permutations and combinations between the servers and the tasks, then the process of adjusting the task scheme and calculating the evaluation index is ended, and index value z is obtained. z is a positive integer, which can be equal to the predetermined number or equal to the number of permutations and combinations between the servers and the tasks. Specifically, it is determined according to the end condition. If the end condition is reaching the predetermined number of adjustments, then z is the predetermined number. The target task allocation scheme can be the task allocation scheme corresponding to the minimum evaluation index value among these evaluation index values.

[0046] In one embodiment, when predicting the energy consumption and response duration of the updated task allocation scheme, the current running parameters of the server can be obtained again, and the time point of obtaining the current running parameters of the server again is later than the time point of processing the energy consumption prediction request for the initial task allocation scheme.

[0047] In some embodiments, the predicted heat load of the server can be the heat generated by the server due to task execution in a future predetermined time period. The predicted heat load can be proportional to the energy consumption prediction value, and the predicted heat load of the server can be obtained based on the prediction result.

[0048] In some embodiments, the environmental data of the data center can include temperature data and humidity data. The cooling equipment can include fans and air conditioners. The predetermined environmental condition is, for example, that the relative humidity of the data center is within a predetermined humidity range.

[0049] In some embodiments, according to the expected heat load, temperature data, and humidity data, a Proportion Integration Differentiation (PID) controller can be used to control the turning on, turning off, and wind speed of the air conditioner and the fan, so that the humidity in the data center is within a predetermined humidity range. Thus, the server can execute the target task allocation scheme within the predetermined humidity range, reducing energy consumption and improving energy utilization efficiency.

[0050] According to an embodiment of the present invention, in response to an energy consumption prediction request, the energy consumption and response duration of the server in a predetermined time period are predicted based on the current operation data and the first historical operation data of the server to obtain a prediction result; the initial task allocation scheme is adjusted according to the prediction result until a predetermined index condition is met; according to the predicted heat load and the environmental data of the data center, the operation of the cooling device is controlled so that the server executes the target task allocation scheme under predetermined environmental conditions. Since the energy consumption is predicted by combining the current operation data and the first historical operation data during the task execution process, it is beneficial to predict the energy consumption according to the distribution characteristics of the operation data, improving the prediction accuracy of the energy consumption and providing a basis for adjusting the task allocation scheme under low energy consumption. When adjusting the task allocation scheme, the predetermined index condition can be set as the minimum energy consumption, and thus a target task allocation scheme with the minimum energy consumption can be obtained. When controlling the operation of the cooling device, it can be controlled according to the heat load of the server and the environment of the data center, enabling precise adjustment, reducing energy waste, and improving energy utilization efficiency. The task execution method of the present invention can reduce energy consumption, improve energy utilization efficiency, and enhance the execution efficiency of the energy server energy efficiency improvement system through closed-loop control of predicting energy consumption, optimizing the task allocation scheme according to the minimum energy consumption, and collaborative cooling control.

[0051] In some embodiments, the process of processing the current operation parameters described in operation S210 above may include the following operations: abnormal parameters are removed from the current operation parameters to obtain intermediate operation parameters; based on the distribution characteristic values of the intermediate operation parameters, the intermediate operation parameters are standardized to obtain the current operation data.

[0052] In some embodiments, a high-precision sensor network can be deployed at each server node to collect various operation parameters including CPU utilization rate, memory usage rate, temperature, and power consumption to obtain the current operation parameters. The current operation parameters are initially cleaned to remove the error parameters or abnormal parameters in the current operation parameters to obtain the intermediate operation parameters.

[0053] For each type of parameter in the intermediate operating parameters, historical parameters of this type can be obtained from the database, and the mean and standard deviation of the historical parameters of this type can be determined to obtain the historical mean and historical standard deviation. The distribution characteristic values of the intermediate operating parameters can be the historical mean and historical standard deviation.

[0054] According to the historical mean and historical standard deviation, a standardization method can be used to standardize each parameter in the intermediate operating parameters to obtain the standardized current operating data.

[0055] According to an embodiment of the present invention, in the stage of collecting operating parameters, the high-precision sensor network not only ensures the comprehensiveness and accuracy of parameter collection, but also can detect changes in the server operating state in real time. By performing preliminary cleaning and standardization processing on the current operating parameters, noise and outliers can be effectively removed, making the data used for energy consumption prediction and the training of the energy consumption prediction model more reliable, thereby improving the stability and prediction accuracy of the server energy efficiency improvement system.

[0056] In some embodiments, the process of predicting the energy consumption and response duration of the server in a predetermined time period described in operation S220 above may include the following operations: using the energy consumption prediction model to process the current operating data and the first historical operating data of the server. The energy consumption prediction model can be trained in the following manner: obtaining a data sample set, where the data in the data sample set includes time points, operating data at the time points, and the actual energy consumption values at the time points; randomly dividing the data in the data sample set into K non-overlapping data sample subsets, where K is an integer greater than 1; using any one of the data sample subsets as the validation set, and using the remaining K - 1 data sample subsets as the training set. Using the operating data at at least two time points in the training set as input data, and using the actual energy consumption values at the time points after at least two time points as labels, training the initial energy consumption prediction model to obtain the performance index of the initial energy consumption prediction model; adjusting the model parameters of the initial energy consumption prediction model according to the performance index, and returning to the operation of training the initial energy consumption prediction model until the performance index meets the predetermined index value to obtain the energy consumption prediction model.

[0057] In some embodiments, the sample data set can be obtained from the database. The data sample set can include multiple data groups, and each data group can include a time point, the operating data of the server at the time point, and the actual energy consumption value of the server at the time point.

[0058] In some embodiments, for the collected data sample set, the processing method mentioned in operation S210 can also be adopted for preliminary cleaning and standardization processing to obtain a standardized data sample set.

[0059] In some embodiments, the data in the standardized data sample set can be randomly divided into K non-overlapping data sample subsets, where K is an integer greater than 1, and the initial energy consumption prediction model is trained by means of cross-validation. The initial energy consumption prediction model can be a constructed long short-term memory network.

[0060] In some embodiments, any one of the data sample subsets can be used as a validation set, and the remaining K - 1 data sample subsets can be used as a training set to train the initial energy consumption prediction model. When using the training set to train the initial energy consumption prediction model, the operating data at at least two time points in the training set can be used as input data, and the actual energy consumption value at the time point after at least two time points can be used as a label to train the initial energy consumption prediction model.

[0061] In some embodiments, the data in the training set can be combined into a feature vector . , is the standardized value of CPU utilization, is the standardized value of memory usage, is the standardized value of temperature, is the standardized value of power consumption.

[0062] In some embodiments, to enhance the learning ability of the initial energy consumption prediction model, lag terms, that is, operating parameters at at least two time points, can be introduced as input data. The input data

[0063] , where t represents any time point, such as the current time point, and t - 1 is the time point before t, is the lag term.

[0064] In some embodiments, the average actual energy consumption value within a future time period composed of at least two time points after time t can be used as a label , and the average actual energy consumption value can be obtained by averaging the actual energy consumption values at each time point within the future time period.

[0065] According to the embodiments of the present invention, in the actual application process of the long short-term memory network, directly inputting the current operating data may not be sufficient for the energy consumption prediction model to accurately predict the future energy consumption of the server. The lag term is essentially also a part of the time series data, representing the data at past time points. As additional information input to the energy consumption prediction model, the lag term can help the energy consumption prediction model better understand the relationship between the data at the current time point and the data at past time points, learn the change trend of energy consumption, so that the energy consumption prediction model can make predictions based on historical data points, thereby improving the prediction ability of the energy consumption prediction model.

[0066] In some embodiments, the input data constructed as described above is utilized and the label , and the weight parameters in the long short-term memory network are adjusted through the backpropagation algorithm so that the initial energy consumption prediction model outputs a value close to the label .

[0067] In some embodiments, the method of cross-validation can be adopted to evaluate the performance metrics of the initial energy consumption prediction model, such as prediction accuracy, and the hyperparameters of the initial energy consumption prediction model are adjusted according to the verification results until the predetermined prediction accuracy is achieved, and the energy consumption prediction model is obtained.

[0068] In some embodiments, the trained energy consumption prediction model can be expressed by formula (1).

[0069] (1)

[0070] Wherein is the functional expression of the energy consumption prediction model, are the trained weight parameters, is the input data, which can be the current running data and the first historical running data, can be the average energy consumption prediction value of the server within a predetermined time period, such as the energy consumption prediction value of the server.

[0071] In some embodiments, the long short-term memory network can process time series data and has a memory function, and can handle the correlation between different time periods. By introducing lag terms to enhance the learning ability of the energy consumption prediction model, the energy consumption prediction model can not only make predictions based on the current running data, but also combine the trends of historical data to improve the prediction accuracy, enhance the precision of energy consumption prediction, and provide a solid foundation for optimizing the workload allocation strategy.

[0072] In some embodiments, the predetermined evaluation metric described in the above operation S230 can be determined in the following manner: the energy consumption prediction value of the data center is determined according to the energy consumption prediction value of the server; the evaluation metric is determined according to the energy consumption prediction value of the data center and the energy consumption prediction weight, as well as the response duration prediction value and the response duration weight.

[0073] In some embodiments, the evaluation metric can be as shown in formula (2).

[0074] (2)

[0075] Wherein is the evaluation metric value, is the response duration weight, is the response duration prediction value, is the energy consumption prediction weight is the predicted energy consumption value of the data center. The predicted energy consumption value of the data center, that is, the total energy consumption of all servers in the data center, can be the sum of the predicted energy consumption values of the servers deployed in the data center. Based on formula (2), the optimization goal can be predetermined to minimize the total energy consumption while ensuring the quality of service and minimizing the evaluation index value.

[0076] According to an embodiment of the present invention, by setting the evaluation index, energy consumption minimization can be achieved on the premise of ensuring the response duration, thereby effectively balancing the relationship between resource utilization rate and response duration and improving the overall resource utilization rate of the data center.

[0077] In some embodiments, the process of obtaining the target task allocation scheme described in operation S230 above may include the following operations: determining an initial evaluation value for the initial task allocation scheme according to the predicted energy consumption value of the server and the predicted response duration value of the server; randomly adjusting the initial task allocation scheme to obtain an updated task allocation scheme; determining an updated evaluation value for the updated task allocation scheme according to the updated predicted energy consumption value and the updated predicted response duration value of the updated task allocation scheme; and determining the target task allocation scheme according to the relationship between the updated evaluation value and the initial evaluation value.

[0078] In some embodiments, for the initial task allocation scheme, the current operation data and the first historical operation data can be input into the energy consumption prediction model, and the predicted energy consumption value and the predicted response duration value can be output. The predicted energy consumption value and the predicted response duration value are input into the function shown in formula (2), and the output evaluation value is used as the initial evaluation value.

[0079] In some embodiments, the tasks can be randomly adjusted among multiple servers based on the reinforcement learning algorithm to obtain an updated task allocation scheme. The reinforcement learning algorithm can be as shown in formula (3).

[0080] (3)

[0081] Wherein, is the task allocation scheme, is the current server state, such as the CPU resource utilization rate, memory utilization rate, temperature, and humidity and other information in the server operation data described above, is the decision-making function of the reinforcement learning algorithm, can be the average predicted energy consumption value of the server within a predetermined time period obtained by formula (1). According to the task allocation scheme , adjust the task distribution on each server and feedback the actual operation effect to the reinforcement learning model as the basis for decision-making.

[0082] In one example, by using the current server state and the average energy consumption prediction value are jointly input into the decision function . By converting the current server state into a vector that can be processed by a neural network, tasks are assigned according to the processable vector, and the allocation scheme that minimizes formula (2) is selected to obtain the task allocation scheme . For example, in the current server state , if the current CPU usage rate of the server is as high as 90%, the reinforcement learning algorithm can reduce the probability of allocating new tasks to this server; for another example, if characterizes that the server enters the low energy consumption stage, the reinforcement learning algorithm can preferentially allocate tasks to this server.

[0083] In some embodiments, the process of randomly adjusting the task allocation scheme can stop when the number of adjustment times reaches a predetermined number of adjustment times; or when the non-repeating combination schemes of tasks reach a predetermined number of permutations and combinations between servers and tasks, the adjustment of the task scheme can stop.

[0084] In some embodiments, the updated energy consumption prediction value and the updated response duration prediction value for updating the task allocation scheme of the above operations can be obtained in the following manner: randomly extract the second historical operation data of the server from the database; according to the second historical operation data and the current operation data, predict the energy consumption and response duration of the server in a predetermined time period to obtain the updated energy consumption prediction value and the updated response duration prediction value for updating the task allocation scheme.

[0085] In some embodiments, when predicting the energy consumption of the updated task scheme for each random adjustment, the current operation data of the server can be obtained again, and historical data can be randomly extracted from the database again. The historical data extracted again can be used as the second historical operation data. The first historical operation data can be extracted from the database when predicting the initial task allocation scheme, and the second historical operation data can be extracted from the database when predicting the updated task allocation scheme. The first historical operation data and the second historical operation data can be different. The relationship between the second historical operation data and the first historical operation data includes one of the following: the second historical operation data includes the first historical operation data; the first historical operation data includes the second historical operation data; there is partial identity between the second historical operation data and the first historical operation data; the second historical operation data and the first historical operation data are completely different.

[0086] In some embodiments, the currently acquired operation data and the second historical data extracted again can be input into the trained energy consumption prediction model to output the energy consumption prediction result of the updated task allocation scheme. When predicting the energy consumption of the updated task allocation scheme, the input data of each updated task allocation scheme is different, that is, the historical operation data is different, and the currently acquired operation data can be the same or different. In the different case, when predicting the energy consumption of the current updated task allocation scheme, the time point for acquiring the currently acquired operation data is later than the time point for acquiring the operation data when predicting the energy consumption of the previous updated task allocation scheme.

[0087] In some embodiments, the updated energy consumption prediction value and the updated response duration prediction value obtained for each updated task allocation scheme can be input into the function shown in formula (2), and the output evaluation value is used as the updated evaluation value.

[0088] In some embodiments, by comparing the initial evaluation value and the updated evaluation value, the task allocation scheme corresponding to the minimum evaluation value can be used as the target task allocation scheme. In one example, when the updated evaluation value is less than the initial evaluation value, the target task allocation scheme can be obtained according to the updated task allocation scheme; when the updated evaluation value is greater than or equal to the initial evaluation value, the target task allocation scheme can be obtained according to the initial task allocation scheme.

[0089] According to the embodiments of the present invention, by using the trained energy consumption prediction model to dynamically analyze and predict the current and future workloads, potential high-energy consumption periods can be identified in advance. The task allocation adjustment based on the reinforcement learning algorithm can dynamically adjust the task allocation strategy according to the real-time changing workload situation, ensuring that the task volume of each server reaches the optimal state. By continuously feeding back the actual operation effect and optimizing the decision function accordingly, the system can maintain efficient operation in a complex and changeable environment, further reducing resource waste and at the same time improving the response speed and service quality.

[0090] In some embodiments, the expected heat load described in operation S240 above can be obtained in the following manner: According to the proportional relationship between the heat load of the server and the energy consumption prediction value, a heat load - energy consumption relationship is constructed, and the heat load - energy consumption relationship can be formula (4); based on the heat load - energy consumption relationship, the energy consumption prediction value of the server is processed to obtain the expected heat load of the server.

[0091] In some embodiments, a temperature and humidity sensor network can be deployed inside the data center to detect the temperature and relative humidity of each area inside the data center in real time. There can be at least one server in each area. According to the load distribution situation in the task allocation scheme, the expected heat load of each server node can be calculated.

[0092] There can be a direct proportional relationship between the heat load of the server and the predicted energy consumption value, that is, the heat load increases as the energy consumption increases. According to this direct proportional relationship, the constructed heat load - energy consumption relationship can be as shown in formula (4).

[0093] (4)

[0094] Wherein, represents the heat load of the th server node, is a proportionality coefficient, is the predicted energy consumption value of this node.

[0095] In some embodiments, by inputting the predicted energy consumption value of the target task allocation scheme, that is, the predicted energy consumption value of the initial task allocation scheme or the updated task allocation scheme, into formula (4), the expected heat load of the server can be obtained.

[0096] In some embodiments, based on the heat load determined by formula (4), the process of controlling the cooling equipment associated with the data center described in operation S240 above may include the following operations: constructing a heat distribution model of the data center based on the expected heat load and the environmental data of the data center; predicting the environmental change trend of the data center based on the heat distribution model; controlling the operating parameters of the cooling equipment according to the environmental change trend and the current temperature of the server until the environmental data of the data center meets the predetermined environmental conditions.

[0097] In some embodiments, the process of constructing a heat distribution model of the data center based on the expected heat load and the environmental data of the data center may include the following operations: using fluid mechanics to simulate the air flow and heat transfer in the data center to obtain the environmental data of the data center, where the environmental data includes temperature data and humidity data; determining the expected heat load of the data center according to the expected heat load of the server; constructing a heat distribution model according to the expected heat load, temperature data and humidity data of the data center.

[0098] In some embodiments, using Computational Fluid Dynamics (CFD for short) to simulate the air flow and heat transfer process inside the data center to construct a heat distribution model, the expression of the heat distribution model can be as shown in formula (5).

[0099] (5)

[0100] Wherein, represents the heat distribution condition of the entire data center, is a function expression obtained based on CFD analysis, is the temperature of each server node, is the relative humidity of each server node, and Q is the heat load vector of the data center, that is, the expected heat load of all server nodes inside the data center, which can be in Equation (4) summation.

[0101] In some embodiments, based on the heat distribution model, an adaptive cooling management system can be constructed to maintain ideal temperature and humidity conditions in the data center by adjusting the fan speed and the operating state of the air conditioning system. In one example, computational fluid dynamics (CFD) can be used to simulate the air flow and heat transfer process inside the data center to generate a heat distribution map of the heat distribution model; by analyzing the environmental data and operating data collected from the sensor network, and combining with the heat distribution model to predict the temperature change trend in a future predetermined time period; according to the expected heat load and real-time temperature feedback of each server node, use a PID controller to adjust the speed of the fan; according to the overall heat load of the data center and the external environmental conditions, dynamically adjust the temperature set point of the air conditioning system, and control the relative humidity to be between 40% and 60%; generate specific control instructions according to the optimization results, adjust the working state of the fan equipment, and turn on or off the air conditioning system.

[0102] According to the embodiments of the present invention, by real-time monitoring the temperature and humidity environmental conditions in the data center, and calculating the expected heat load of each server node in combination with the task allocation scheme, and through the prediction of the temperature change trend by the heat distribution model, the set point of the air conditioning system can be adjusted in advance to avoid unnecessary energy consumption; combined with the temperature change trend, the fan speed and the operating state of the air conditioning system can be controlled more precisely, so as to achieve the purpose of energy conservation and emission reduction, and further the working state of the cooling system can be accurately controlled. Using CFD to simulate the air flow and heat transfer process inside the data center to construct a heat distribution model can not only accurately predict the air flow and heat transfer process inside the data center, but also dynamically adjust the settings of the fan and the air conditioning according to the actual situation, thus avoiding over-cooling or under-cooling problems, achieving the best cooling effect configuration, and reducing energy consumption.

[0103] In some embodiments, the present invention can also establish a continuous monitoring mechanism to regularly evaluate the effectiveness of the best cooling effect configuration, the task allocation scheme, and the energy consumption prediction model, and perform feedback adjustment on the strategy and model according to the latest data analysis results to obtain optimized operating parameters. In one example, according to the current operating data, the performance index value of the server can be determined, where the performance index value includes at least one of the following: power usage efficiency, temperature control accuracy, relative humidity level, cooling equipment energy consumption, and server utilization rate; according to the deviation between the performance index value and the predetermined threshold, feedback and adjust the model parameters of the energy consumption prediction model to obtain an updated energy consumption prediction model.

[0104] In some embodiments, a closed-loop control system can be constructed to automatically collect the above-mentioned current operation data, calculate deviation information, and feed the deviation information back as input to the adaptive cooling management system, the task assignment adjustment algorithm, and the energy consumption prediction model to trigger corresponding adjustment actions. According to the feedback data, the weight parameters in the energy consumption prediction model can be adjusted to make the energy consumption prediction model better fit the data pattern. Optimizing the decision function in the task assignment adjustment algorithm, that is, the decision function in the reinforcement learning algorithm, can ensure that the optimal value.

[0105] In one embodiment, the closed-loop control system can include a data acquisition layer, a deviation analysis and feedback layer, and an execution adjustment layer.

[0106] The data acquisition layer can obtain the current operation data through sensors, monitoring tools, and log analysis. The current operation data can include at least one of energy consumption data, environmental condition data, server performance data, and task assignment scheme data.

[0107] The energy consumption data can include the actual power consumption values of server nodes at different time periods, the real-time monitoring results of the overall energy consumption of the data center, the energy consumption of information technology (IT) devices, the energy consumption records during each task execution, and the energy consumption data of the fan and air-conditioning systems, etc.

[0108] The environmental condition data can include the real-time measured values of temperature and relative humidity in each area of the data center, the temperature difference data between the server intake and exhaust ports, and the influence parameters of the external environmental temperature on the cooling efficiency of the data center.

[0109] The server performance data can include CPU utilization, memory usage, disk I / O rate, the load conditions generated during the operation of each application program, and the network traffic conditions and their impact on server performance, etc.

[0110] The task assignment scheme data can include the distribution of each task on different servers, the workload adjustment scheme obtained according to the energy consumption prediction model, and the task assignment scheme generated by the reinforcement learning algorithm, etc.

[0111] The deviation analysis and feedback layer can calculate the performance metric values of the server by using the data collected by the data acquisition layer. For example, the power usage effectiveness can be determined based on the ratio of the total energy consumption of the data center to the energy consumption of IT equipment. The temperature control accuracy can be determined based on the difference between the temperature of each area in the data center and the preset temperature. The relative humidity level can be obtained based on the real-time measured value of the relative humidity. The energy consumption of the cooling equipment can be obtained based on the energy consumption data of the fan and air conditioning systems. The server utilization rate can be obtained based on the CPU utilization rate, memory usage rate, and disk I / O rate.

[0112] In some embodiments, an index threshold can be set for each performance metric. The deviation analysis and feedback layer can calculate the deviation between the performance metric value and the index threshold. When the deviations of these metric values, such as the power usage effectiveness, temperature control accuracy, relative humidity level, energy consumption of the cooling equipment, and server utilization rate, are within the preset deviation, no feedback adjustment is required. When the deviation of at least one of the power usage effectiveness, temperature control accuracy, relative humidity level, energy consumption of the cooling equipment, and server utilization rate is not within the preset deviation, an operation for triggering feedback adjustment can be performed. By providing a fault tolerance space such as a preset deviation range, frequent triggering of feedback adjustment operations can be avoided, the energy consumption of the server caused by frequently executing unnecessary operations can be reduced, the energy utilization rate of the server can be improved, and the service life can be extended.

[0113] When the deviation analysis and feedback layer obtains that a feedback adjustment operation is required, the execution adjustment layer can screen out the abnormal performance metrics whose deviations are outside the preset deviation range from these performance metrics, such as the power usage effectiveness, temperature control accuracy, relative humidity level, energy consumption of the cooling equipment, and server utilization rate, and adjust the components related to the abnormal performance metrics. In one example, if the index of the power usage effectiveness is abnormal, since the power usage effectiveness is obtained based on the energy consumption of the data center, the energy consumption of the data center is obtained based on the energy consumption of each server node, and the energy consumption of the server node is obtained through an energy consumption prediction model, if the index of the power usage effectiveness is abnormal, it may indicate that there is a problem with the accuracy of the energy consumption prediction model. In such a case, real labels can be supplemented to the error samples, and the energy consumption prediction model can be continuously trained using the samples with supplemented real labels to achieve feedback adjustment of the energy consumption prediction model.

[0114] According to the embodiments of the present invention, the data acquisition layer, the deviation analysis and feedback layer, and the execution adjustment layer can regularly evaluate the energy consumption prediction model, task allocation scheme, and cooling management, timely locate the problems existing in the server energy efficiency improvement system, and perform feedback adjustment on the energy consumption prediction model, task allocation scheme, and cooling management to ensure the reliability of the server energy efficiency improvement system, improve the execution efficiency of the server energy efficiency improvement system, and reduce energy consumption.

[0115] Apply the updated energy consumption prediction model, task allocation scheme, and cooling management strategy, and closely observe the actual operation status of the data center; based on the updated energy consumption prediction model, task allocation scheme, and cooling management strategy, a detailed optimization report can be generated for the reference of management personnel.

[0116] According to an embodiment of the present invention, the design of the continuous detection mechanism enables the server energy efficiency improvement system to regularly evaluate the actual effects of each energy consumption prediction model, task allocation scheme, and cooling management strategy, and feedback-adjust the strategy and model according to the latest data analysis results. The closed-loop control system can not only timely discover and solve potential problems, but also continuously optimize the operation parameters to ensure that the system is always in the best operating state. By generating a detailed optimization report, management personnel can intuitively understand the operation status of the system and make more scientific decisions based on this, further improving the management efficiency and energy efficiency performance of the data center.

[0117] Figure 4 The flowchart of a task execution method according to another embodiment of the present invention is shown.

[0118] As Figure 4 shown, the task execution method of this embodiment may include operations S410 to S460.

[0119] In operation S410, the server operation parameters are collected and preprocessed by using a sensor network to obtain standardized server operation data. In one embodiment, operation S410 may refer to operation S210.

[0120] In operation S420, a machine learning algorithm is used to train an energy consumption prediction model for the standardized server operation data to obtain an energy consumption prediction model based on historical data.

[0121] In operation S430, the energy consumption prediction model is used to dynamically analyze and predict the current and future server workloads to obtain an optimized workload allocation strategy. In one embodiment, operations S420 to S430 may refer to operation S220. For the training of the energy consumption prediction model, it is not necessary to perform training before each prediction. The energy consumption prediction model may be a trained one, and when predicting energy consumption, the energy consumption prediction model can be directly called for prediction.

[0122] In operation S440, by applying the optimized workload allocation strategy to the intelligent workload allocation system, a reinforcement learning algorithm is used to dynamically adjust the allocation of tasks among multiple servers to obtain a task allocation scheme. In one embodiment, operation S440 may refer to operation S230.

[0123] In operation S450, by real-time monitoring of the temperature and humidity environment conditions in the data center and combining the task allocation scheme, an adaptive cooling management method is used to automatically adjust the working state of the fan device to obtain the best cooling effect configuration. In one embodiment, operation S450 can refer to operation S240.

[0124] In operation S460, by establishing a continuous monitoring mechanism, the effectiveness of the optimal cooling effect configuration, the task allocation scheme, and the energy consumption prediction model is regularly evaluated, and the strategy and model are adjusted based on the latest data analysis results to obtain optimized operating parameters. In one embodiment, operation S460 can refer to the above-described process of determining the performance index value of the server based on the current operating data, and based on the deviation between the performance index value and the predetermined threshold, feedback adjusting the model parameters of the energy consumption prediction model, the decision function of the task allocation scheme, and the cooling management system.

[0125] According to an embodiment of the present invention, by using a sensor network to collect and preprocess server operating parameters, standardized data processing is achieved, data quality and consistency are ensured, and a reliable data basis is provided for the training of energy consumption prediction models, thereby improving the accuracy and reliability of the model. By using the energy consumption prediction model to dynamically analyze and predict current and future server workloads, an optimized workload distribution strategy is implemented, which not only helps to reduce energy consumption, but also ensures that service quality is not affected. The effect of each solution is quantified through comprehensive evaluation indicators to ensure efficient use of resources and improve overall energy utilization efficiency. By applying the optimized workload distribution strategy to the intelligent workload distribution system, dynamic adjustment of tasks among multiple servers is achieved, ensuring that the task allocation of each server reaches the optimal state, further reducing unnecessary energy consumption and improving the system's response speed and service quality.

[0126] It should be noted that, unless it is explicitly stated that there is a sequence of execution between different operations shown in the flowchart in the embodiments of the present invention, or different operations have a sequence of execution in technical implementation, otherwise, the execution order of multiple operations may not be particular, and multiple operations may also be executed simultaneously.

[0127] Based on the above task execution method, the present invention also provides a task execution device. Figure 5 The device is described in detail.

[0128] Figure 5 A structural block diagram of a task execution device according to an embodiment of the present invention is shown.

[0129] like Figure 5 As shown, the task execution device 500 of this embodiment includes a processing module 510 , a prediction module 520 , an adjustment module 530 and a control module 540 .

[0130] A processing module 510, configured to process the current operating parameters of the server in response to an energy consumption prediction request for an initial task allocation scheme, so as to obtain current operating data.

[0131] A prediction module 520, configured to predict the energy consumption and response duration of the server in a predetermined time period according to the current operating data and the first historical operating data of the server, so as to obtain a prediction result.

[0132] An adjustment module 530, configured to adjust the initial task allocation scheme according to the prediction result until the evaluation index based on the adjusted task allocation scheme meets a predetermined index condition, so as to obtain a target task allocation scheme.

[0133] A control module 540, configured to control the operation of a cooling device associated with the data center according to the predicted heat load obtained based on the prediction result and the environmental data of the data center where the server is located, until the environmental data of the data center meets a predetermined environmental condition, so that the server executes the target task allocation scheme under the predetermined environmental condition.

[0134] According to an embodiment of the present invention, in response to an energy consumption prediction request, the energy consumption and response duration of the server in a predetermined time period are predicted according to the current operating data and the first historical operating data of the server, so as to obtain a prediction result; the initial task allocation scheme is adjusted according to the prediction result until a predetermined index condition is met; according to the predicted heat load and the environmental data of the data center, the operation of the cooling device is controlled, so that the server executes the target task allocation scheme under a predetermined environmental condition. Since in the process of task execution, the energy consumption is predicted by combining the current operating data and the first historical operating data, which is beneficial to predicting the energy consumption according to the distribution characteristics of the operating data, improving the prediction accuracy of the energy consumption, and providing a basis for adjusting the task allocation scheme under low energy consumption. When adjusting the task allocation scheme, the predetermined index condition can be set as the minimum energy consumption, and then a target task allocation scheme with the minimum energy consumption can be obtained. When controlling the operation of the cooling device, it can be controlled according to the heat load of the server and the environment of the data center, which can achieve precise adjustment, reduce energy waste, and improve energy utilization rate. The task execution method of the present invention can reduce energy consumption, improve energy utilization rate, and improve the execution efficiency of the energy efficiency improvement system of the server through the closed-loop control of predicting energy consumption, optimizing the task allocation scheme according to the minimum energy consumption, and collaborative cooling control.

[0135] Optionally, the task execution device may further include a first determination module and a second determination module.

[0136] The first determination module is configured to determine the energy consumption prediction value of the data center according to the energy consumption prediction value of the server.

[0137] A second determination module, configured to determine an evaluation index according to an energy consumption prediction value and an energy consumption prediction weight of a data center, as well as a response duration prediction value and a response duration weight.

[0138] Optionally, the adjustment module 530 may include a first determination sub-module, an adjustment sub-module, a second determination sub-module, and a third determination sub-module.

[0139] The first determination sub-module is configured to determine an initial evaluation value for an initial task allocation scheme according to an energy consumption prediction value of a server and a response duration prediction value of the server.

[0140] The adjustment sub-module is configured to randomly adjust the initial task allocation scheme to obtain an updated task allocation scheme.

[0141] The second determination sub-module is configured to determine an updated evaluation value for the updated task allocation scheme according to an updated energy consumption prediction value and an updated response duration prediction value of the updated task allocation scheme.

[0142] The third determination sub-module is configured to determine a target task allocation scheme according to the relationship between the updated evaluation value and the initial evaluation value.

[0143] Optionally, the third determination sub-module may include a first result unit and a second result unit.

[0144] The first result unit is configured to obtain a target task allocation scheme according to the updated task allocation scheme when the updated evaluation value is less than the initial evaluation value.

[0145] The second result unit is configured to obtain a target task allocation scheme according to the initial task allocation scheme when the updated evaluation value is greater than or equal to the initial evaluation value.

[0146] Optionally, the second determination sub-module may include an extraction unit and a prediction unit.

[0147] The extraction unit is configured to randomly extract second historical operation data of the server from a database.

[0148] The prediction unit is configured to predict the energy consumption and response duration of the server in a predetermined time period according to the second historical operation data and the current operation data, so as to obtain an updated energy consumption prediction value and an updated response duration prediction value of the updated task allocation scheme.

[0149] Optionally, the relationship between the second historical operation data and the first historical operation data includes one of the following: the second historical operation data includes the first historical operation data; the first historical operation data includes the second historical operation data; the second historical operation data and the first historical operation data are partially the same; the second historical operation data and the first historical operation data are completely different.

[0150] Optionally, the task execution device may further include a construction module and a result module.

[0151] The construction module is configured to construct a heat load - energy consumption relationship according to the proportional relationship between the heat load of the server and the predicted energy consumption value.

[0152] The result module is configured to process the predicted energy consumption value of the server based on the heat load - energy consumption relationship to obtain the expected heat load of the server.

[0153] Optionally, the control module 540 may include a construction sub - module, a prediction sub - module, and a control sub - module.

[0154] The construction sub - module is configured to construct a heat distribution model of the data center based on the expected heat load and the environmental data of the data center.

[0155] The prediction sub - module is configured to predict the environmental change trend of the data center based on the heat distribution model.

[0156] The control sub - module is configured to control the operating parameters of the cooling device according to the environmental change trend and the current temperature of the server until the environmental data of the data center meets the predetermined environmental conditions.

[0157] Optionally, the construction sub - module may include a simulation unit, a determination unit, and a construction unit.

[0158] The simulation unit is configured to use fluid mechanics to simulate the air flow and heat transfer in the data center to obtain the environmental data of the data center, where the environmental data includes temperature data and humidity data.

[0159] The determination unit is configured to determine the expected heat load of the data center according to the expected heat load of the server.

[0160] The construction unit is configured to construct a heat distribution model according to the expected heat load, temperature data, and humidity data of the data center.

[0161] Optionally, the prediction module 520 may include a processing sub - module, an acquisition sub - module, a division sub - module, a training sub - module, and an adjustment sub - module.

[0162] The processing sub - module is configured to process the current operating data and the first historical operating data of the server using the energy consumption prediction model.

[0163] The acquisition sub - module is configured to acquire a data sample set, where the data in the data sample set includes a time point, the operating data at the time point, and the actual energy consumption value at the time point.

[0164] The division sub - module is configured to randomly divide the data in the data sample set into K non - overlapping data sample subsets, where K is an integer greater than 1.

[0165] A training sub-module for using any subset of data samples as a validation set, and the remaining K-1 subsets of data samples as a training set. Using the operating data at at least two time points in the training set as input data, and the actual energy consumption values at the time points after at least two time points as labels, to train an initial energy consumption prediction model to obtain the performance metrics of the initial energy consumption prediction model.

[0166] An adjustment sub-module for adjusting the model parameters of the initial energy consumption prediction model according to the performance metrics, and returning the operation of training the initial energy consumption prediction model until the performance metrics meet the predetermined metric values to obtain an energy consumption prediction model.

[0167] Optionally, the task execution device may further include a third determination module and an adjustment module.

[0168] A third determination module for determining the performance metric values of the server according to the current operating data, where the performance metric values include at least one of the following: power usage effectiveness, temperature control accuracy, relative humidity level, cooling equipment energy consumption, and server utilization rate.

[0169] An adjustment module for feedback-adjusting the model parameters of the energy consumption prediction model according to the deviation between the performance metric values and the predetermined thresholds to obtain an updated energy consumption prediction model.

[0170] Optionally, the processing module 510 may include a rejection sub-module and a processing sub-module.

[0171] A rejection sub-module for rejecting abnormal parameters from the current operating parameters to obtain intermediate operating parameters.

[0172] A processing sub-module for standardizing the intermediate operating parameters based on the distribution characteristic values of the intermediate operating parameters to obtain the current operating data.

[0173] According to an embodiment of the present invention, any multiple of the processing module 510, the prediction module 520, the adjustment module 530, and the control module 540 may be combined and implemented in one module, or any one of them may be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules may be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present invention, at least one of the processing module 510, the prediction module 520, the adjustment module 530, and the control module 540 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on substrate, a system on package, an application specific integrated circuit (ASIC), or may be implemented by any other reasonable means such as hardware or firmware for integrating or packaging circuits, or may be implemented in any one of the three implementation manners of software, hardware, and firmware, or in any appropriate combination of several of them. Alternatively, at least one of the processing module 510, the prediction module 520, the adjustment module 530, and the control module 540 may be at least partially implemented as a computer program module, and when the computer program module is run, corresponding functions may be executed.

[0174] Figure 6 The block diagram of an electronic device suitable for implementing the task execution method according to an embodiment of the present invention is shown.

[0175] As Figure 6 shown, the electronic device 600 according to an embodiment of the present invention includes a processor 601, which may perform various appropriate actions and processes according to a program stored in a read only memory (ROM) 602 or a program loaded from a storage section 608 into a random access memory (RAM) 603. The processor 601 may include, for example, a general microprocessor (such as a CPU), an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application specific integrated circuit (ASIC)), and so on. The processor 601 may also include on-board memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0176] In the RAM 603, various programs and data required for the operation of the electronic device 600 are stored. The processor 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. The processor 601 performs various operations of the method flow according to the embodiments of the present invention by executing the programs in the ROM 602 and / or the RAM 603. It should be noted that the programs may also be stored in one or more memories other than the ROM 602 and the RAM 603. The processor 601 may also perform various operations of the method flow according to the embodiments of the present invention by executing the programs stored in the one or more memories.

[0177] According to an embodiment of the present invention, the electronic device 600 may further include an input / output (I / O) interface 605, and the input / output (I / O) interface 605 is also connected to the bus 604. The electronic device 600 may further include one or more of the following components connected to the input / output (I / O) interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed so that a computer program read from it can be installed into the storage section 608 as needed.

[0178] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist separately without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of the present invention is implemented.

[0179] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, device, or apparatus. For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 602 and / or RAM 603 described above and / or one or more memories other than the ROM 602 and RAM 603.

[0180] An embodiment of the present invention also includes a computer program product, which includes a computer program that contains program code for executing the method shown in the flowchart. When the computer program product runs in a computer system, the program code is used to enable the computer system to implement the task execution method provided by the embodiment of the present invention.

[0181] When the computer program is executed by the processor 601, it executes the above functions defined in the system / apparatus of the embodiment of the present invention. According to an embodiment of the present invention, the above-described systems, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0182] In one embodiment, the computer program can rely on tangible storage media such as optical storage devices and magnetic storage devices. In another embodiment, the computer program can also be transmitted and distributed in the form of a signal on a network medium, and is downloaded and installed through the communication part 609, and / or installed from the removable medium 611. The program code contained in the computer program can be transmitted by any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0183] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, it executes the above functions defined in the system of the embodiment of the present invention. According to an embodiment of the present invention, the above-described systems, devices, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0184] According to embodiments of the present invention, program code for executing the computer programs provided by the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, such as Java, C++, Python, the "C" language, or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (for example, by using an Internet service provider to connect through the Internet).

[0185] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the above-mentioned module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0186] Those skilled in the art can understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments of the present invention can be combined and / or combined in various ways. All such combinations and / or combinations fall within the scope of the present invention.

[0187] The above describes the embodiments of the present invention. However, these embodiments are only for illustrative purposes and not for limiting the scope of the present invention. Although the embodiments are described separately above, this does not mean that the measures in each embodiment cannot be used advantageously in combination. Without departing from the scope of the present invention, those skilled in the art can make various substitutions and modifications, and all such substitutions and modifications should fall within the scope of the present invention.

Claims

1. A task execution method, characterized in that: The method comprises: In response to the energy consumption prediction request for the initial task allocation plan, current operating parameters of the server are processed to obtain current operating data; Predicting the energy consumption and response time of the server in a predetermined time period according to the current operation data and the first historical operation data of the server to obtain a prediction result; According to the prediction result, the initial task allocation plan is adjusted until the evaluation index based on the adjusted task allocation plan meets the predetermined index condition, thereby obtaining the target task allocation plan; According to the expected heat load obtained based on the prediction result and the environmental data of the data center where the server is located, the operation of the cooling equipment associated with the data center is controlled until the environmental data of the data center meets the predetermined environmental conditions, so that the server executes the target task allocation plan under the predetermined environmental conditions.

2. The method according to claim 1, characterized in that The prediction result includes a predicted value of energy consumption of the server and a predicted value of response time of the server; The evaluation index is determined by the following method: Determining the predicted energy consumption value of the data center according to the predicted energy consumption value of the server; The evaluation index is determined according to the energy consumption prediction value and energy consumption prediction weight of the data center, and the response time prediction value and response time weight.

3. The method according to claim 2, characterized in that The step of adjusting the initial task allocation plan according to the prediction result until the evaluation index based on the adjusted task allocation plan meets the predetermined index condition to obtain the target task allocation plan includes: Determining an initial evaluation value for the initial task allocation scheme according to the predicted value of the energy consumption of the server and the predicted value of the response time of the server; Randomly adjusting the initial task allocation scheme to obtain an updated task allocation scheme; Determining an update evaluation value for the update task allocation scheme according to the update energy consumption prediction value and the update response time prediction value of the update task allocation scheme; The target task allocation scheme is determined according to the relationship between the updated evaluation value and the initial evaluation value.

4. The method according to claim 3, characterized in that Determining the target task allocation scheme according to the relationship between the updated evaluation value and the initial evaluation value includes: In the case where the updated evaluation value is less than the initial evaluation value, obtaining the target task allocation scheme according to the updated task allocation scheme; In a case where the updated evaluation value is greater than or equal to the initial evaluation value, the target task allocation scheme is obtained according to the initial task allocation scheme.

5. The method according to claim 3, characterized in that: The update energy consumption prediction value and update response time prediction value of the update task allocation scheme are obtained by: Randomly extracting second historical operation data of the server from the database; The energy consumption and response time of the server in the predetermined time period are predicted based on the second historical operation data and the current operation data to obtain an updated energy consumption prediction value and an updated response time prediction value of the updated task allocation scheme.

6. The method according to claim 5, characterized in that The relationship between the second historical operation data and the first historical operation data includes one of the following: The second historical operation data includes the first historical operation data; The first historical operation data includes the second historical operation data; The second historical operation data is partially identical to the first historical operation data; The second historical operation data is completely different from the first historical operation data.

7. The method according to claim 2, characterized in that Also includes: Constructing a heat load-energy consumption relationship according to a proportional relationship between the heat load of the server and the predicted energy consumption value; The predicted energy consumption value of the server is processed based on the heat load-energy consumption relationship to obtain the expected heat load of the server.

8. The method according to claim 7, characterized in that The step of controlling the operation of cooling equipment associated with the data center according to the expected heat load obtained based on the prediction result and the environmental data of the data center where the server is located, until the environmental data of the data center meets a predetermined environmental condition, includes: constructing a heat distribution model of the data center based on the expected heat load and environmental data of the data center; Predicting the environmental change trend of the data center based on the heat distribution model; According to the environmental change trend and the current temperature of the server, the operating parameters of the cooling device are controlled until the environmental data of the data center meets the predetermined environmental conditions.

9. The method according to claim 8, characterized in that The step of constructing a heat distribution model of the data center based on the expected heat load and environmental data of the data center includes: Using fluid mechanics to simulate air flow and heat transfer in the data center to obtain environmental data of the data center, wherein the environmental data includes temperature data and humidity data; Determining an expected heat load of the data center according to the expected heat load of the server; The heat distribution model is constructed according to the expected heat load of the data center, the temperature data and the humidity data.

10. The method according to claim 1, characterized in that The predicting of the energy consumption and response time of the server in a predetermined time period according to the current operation data and the first historical operation data of the server includes: processing the current operation data and the first historical operation data of the server by using an energy consumption prediction model; The energy consumption prediction model is trained in the following way: Acquire a data sample set, wherein the data in the data sample set includes a time point, operation data at the time point, and an actual energy consumption value at the time point; Randomly divide the data in the data sample set into K non-overlapping data sample subsets, where K is an integer greater than 1; Taking any data sample subset as a validation set, taking the remaining K-1 data sample subsets as training sets, using the operating data of at least two time points in the training set as input data, taking the actual energy consumption values ​​at the time points after the at least two time points as labels, training the initial energy consumption prediction model, and obtaining the performance index of the initial energy consumption prediction model; The model parameters of the initial energy consumption prediction model are adjusted according to the performance index, and the operation of training the initial energy consumption prediction model is returned until the performance index meets the predetermined index value to obtain the energy consumption prediction model.

11. The method according to claim 10, characterized in that The method further comprises: Determining a performance indicator value of the server according to the current operation data, wherein the performance indicator value includes at least one of the following: power usage efficiency, temperature control accuracy, relative humidity level, cooling equipment energy consumption, and server utilization; According to the deviation between the performance indicator value and the predetermined threshold, the model parameters of the energy consumption prediction model are feedback-adjusted to obtain an updated energy consumption prediction model.

12. The method according to claim 1, characterized in that The processing of the current operating parameters of the server to obtain the current operating data includes: Eliminating abnormal parameters from the current operating parameters to obtain intermediate operating parameters; Based on the distribution characteristic value of the intermediate operation parameter, the intermediate operation parameter is standardized to obtain the current operation data.

13. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 12.

14. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

15. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images