Power consumption profiling method, device, data center, and storage medium

Abstract

The embodiment of the present application provides a power consumption portrait method, device, data center and storage medium. In the embodiment of the present application, a reference model and a reference service of a new service model can be acquired; a historical power consumption portrait of the reference model running the reference service, a power consumption dictionary of the reference model and a power consumption dictionary of the new model are acquired; further, an initial power consumption portrait of a to-be-tested model running a to-be-tested service (the new service model) can be determined according to the historical power consumption portrait. Further, actual power consumption data generated by the to-be-tested model running the to-be-tested service (the new service model) deployed in a cabinet can be acquired; and the initial power consumption portrait is corrected by using the actual power consumption data, so that adaptive adjustment of the new service model portrait is realized, the obtained power consumption portrait of the new service model is as close as possible to the actual power consumption portrait, and the accuracy of the power consumption portrait is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, and particularly relates to a power consumption profiling method, device, data center and storage medium. BACKGROUND

[0002] As an information infrastructure, a data center is a powerful support for the development of digital economy. A rack (also referred to as a cabinet) as a basic infrastructure of the data center can be used as a carrier for storing servers and can provide the servers with resources such as power and communication. Since the power consumption supported by the rack is limited, in order to improve the utilization rate of the cabinet resources, the data center needs to reasonably set the number of servers in the cabinet.

[0003] In the existing planning scheme of the data center, the allocation scheme of the servers is usually determined according to the total power consumption supported by each rack and the power consumption of each server, which requires accurate estimation of the power consumption of the servers and pre-profiling of the power consumption of the servers. SUMMARY

[0004] Aspects of the present application provide a power consumption profiling method, device, data center and storage medium to improve the accuracy of server power consumption profiling.

[0005] The present application provides a power consumption profiling method, comprising:

[0006] obtaining a reference model of a to-be-tested model from models deployed in a cabinet;

[0007] obtaining a historical power consumption profile of a reference service of the reference model running the to-be-tested service;

[0008] determining an initial power consumption profile of the to-be-tested model running the to-be-tested service according to the historical power consumption profile;

[0009] obtaining actual power consumption data of the to-be-tested model running the to-be-tested service deployed in the cabinet;

[0010] correcting the initial power consumption profile by using the actual power consumption data to obtain a power consumption profile of the to-be-tested model running the to-be-tested service.

[0011] The present application also provides a data center, comprising: a plurality of cabinets, servers deployed in the cabinets and a management device; the models of the servers deployed in the cabinets comprise: a to-be-tested model;

[0012] The management device is configured to acquire a reference model of the to-be-tested model from models of servers deployed in the cabinet, acquire a historical power consumption profile of a reference service run by the reference model, determine an initial power consumption profile of the to-be-tested model running the to-be-tested service according to the historical power consumption profile, acquire actual power consumption data of the to-be-tested model running the to-be-tested service deployed in the data center, and correct the initial power consumption profile by using the actual power consumption data to obtain a power consumption profile of the to-be-tested model running the to-be-tested service.

[0013] The embodiment of the present application further provides a computing device, comprising a memory and a processor; wherein the memory is configured to store a computer program;

[0014] The processor is coupled to the memory and is configured to execute the computer program to execute the steps in the power consumption profile method.

[0015] The embodiment of the present application further provides a computer readable storage medium storing computer instructions, which, when executed by one or more processors, cause the one or more processors to execute the steps in the power consumption profile method.

[0016] The embodiment of the present application further provides a computer program product, comprising a computer program; when the computer program is executed by a processor, the processor is caused to execute the steps in the power consumption profile method.

[0017] In the embodiment of the present application, the reference model and the reference service of the new service model can be acquired, and the historical power consumption profile of the reference model running the reference service, the power consumption dictionary of the reference model and the power consumption dictionary of the new model can be acquired; further, the initial power consumption profile of the to-be-tested model running the to-be-tested service (the new service model) can be determined according to the historical power consumption profile. Further, the actual power consumption data of the to-be-tested model running the to-be-tested service (the new service model) deployed in the cabinet can be acquired, and the initial power consumption profile is corrected by using the actual power consumption data, so that the adaptive adjustment of the new service model profile is realized, and the obtained power consumption profile of the new service model is as close as possible to the actual power consumption profile, which helps to improve the accuracy of the power consumption profile. BRIEF DESCRIPTION OF DRAWINGS

[0018] The accompanying drawings, which are included to provide a further understanding of the present application, constitute a part of the present application and illustrate the illustrative embodiments of the present application and its description, which serve to explain the present application, and do not constitute improper limitations on the present application. In the drawings:

[0019] Figure 1 A flowchart of a power consumption profile method provided by the embodiment of the present application is shown in the figure;

[0020] Figure 2A power consumption dictionary curve diagram provided by an embodiment of the present application;

[0021] Figure 3a A structure diagram of a data center provided by an embodiment of the present application;

[0022] Figure 3b A structure diagram of a computing device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described below in detail with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0024] A data center is a computer room that can accommodate a large number of servers and related power distribution, network and cooling equipment, and can control application programs for storing, managing and processing data. Therefore, the data center as a whole can be regarded as a warehouse-sized computer, or simply as a server cluster.

[0025] In order to more conveniently expand or reduce the size of the data center, a modular data center component unit formed by standardized products conforming to industry standards can be used at present. In this way, it can be repeatedly increased or reduced according to the demand. There are mainly three types of applications for such modular data center component units, including container data centers, micro-module data centers and warehouse data centers. It should be noted that the method of configuring racks of the present application can be applied to various types of data centers, and is particularly suitable for such modular data center component units.

[0026] In addition, when building a data center, the total cost of the data center needs to be considered, including servers, cabinets, power consumption, network equipment and maintenance costs. Since the power consumption supported by the cabinet is limited, the power consumption needs to be estimated when deploying the server in front of the cabinet, and a decision needs to be made on how many servers are deployed on the cabinet. If the estimated server power consumption is too low, it may lead to too many servers being deployed when the server is deployed in the cabinet, resulting in an over-power risk; if the estimated server power consumption is too high, it may lead to too few servers being deployed when the server is deployed in the cabinet, resulting in low power utilization of the cabinet and high PUE (Power Usage Effectiveness) of the data center. Therefore, accurate power consumption estimation is crucial for safe and economical operation of the data center.

[0027] The server power consumption estimation generally includes: using the rated power consumption as the server on-rack power consumption, estimating the server on-rack power consumption by using the power consumption dictionary given by the server demand side under the load rate, and power consumption profiling of the server. In the embodiment of the present application, the server on-rack refers to the server deployed in the cabinet, and the server on-rack power consumption refers to the power consumed by the server deployed in the cabinet. The power consumption dictionary stores the relationship between the CPU load rate of the server and the power consumption of the server.

[0028] The power consumption dictionary of the server described above can be obtained by testing the relationship between the CPU load rate and the power consumption of the server under the given experimental environment, i.e., making the other components of the server except the CPU fully loaded. For example, the relationship curve between the CPU load rate and the power consumption of the server can be obtained by testing the power consumption of the server corresponding to the CPU load rate of 10%, 20%, …, 100% under the given experimental environment, i.e., making the other components of the server except the CPU fully loaded. Generally, the power consumption of the server is positively correlated with the CPU load rate, i.e., the greater the CPU load rate, the greater the power consumption of the server.

[0029] For the scheme of using the rated power consumption as the server on-rack power consumption, since the daily power consumption of the server is usually lower than the rated power consumption, using the rated power consumption as the server on-rack power consumption will result in an overestimation of the power consumption, which will lead to a lower number of servers deployed on the cabinet, and further lead to a lower power utilization rate of the cabinet and a higher PUE of the data center.

[0030] For the scheme of estimating the server on-rack power consumption by using the power consumption dictionary given by the server demand side under the load rate, since the daily load rate of the server is usually lower than the load rate given by the server demand side, the power consumption of the server estimated by using the power consumption dictionary given by the server demand side under the load rate will usually be overestimated, which will also lead to a lower number of servers deployed on the cabinet, and further lead to a lower power utilization rate of the cabinet and a higher PUE of the data center.

[0031] Power consumption image is closer to real power consumption than relative rated power consumption and power consumption dictionary. Therefore, estimating server on-rack power using power consumption image can improve power utilization. Existing power consumption image method is to statistically analyze server historical power consumption, which is mainly used for mature service models. The mature service model refers to a mature model running a mature service. The mature model refers to a model deployed in a cabinet and generating actual power consumption data reaching a set data amount threshold. The set data amount threshold can be flexibly set according to the accuracy requirement of the power consumption image. The higher the accuracy requirement of the power consumption image, the larger the data amount threshold. The mature service refers to a service or application whose actual power consumption data generated by the model deployed in the cabinet reaches the set data amount threshold. Accordingly, the mature service model refers to a model that has been deployed in the cabinet for a long enough time, and the model has run a certain service or services for a long enough time to generate enough actual power consumption data, which can accurately image the power consumption of the server running a certain service or services.

[0032] However, the method of imaging the power consumption of the server based on the historical power consumption data requires enough historical power consumption data, and the future power consumption distribution is consistent with the historical power consumption distribution. For new server models or new services (referred to as new service models), there is not enough historical power consumption data to image the power consumption of the server, or the new service model generates historical power consumption data for a short time, has a low load rate, etc., which may also lead to the historical power consumption distribution of the new service model not equal to the future power consumption distribution, and may also lead to inaccurate power consumption image.

[0033] The above-mentioned new service model refers to a service model that has not generated enough actual power consumption data to image the power consumption of the server, which can include a new model and / or a new service. Specifically, the new service model can include a new model running a new service, a new model running a mature service, and a mature model running a new service, etc. The new model refers to a model that has not been deployed in a cabinet and / or a model that has been deployed in a cabinet but has not generated enough actual power consumption data. Not generating enough actual power consumption data refers to the data amount of the generated actual power consumption data not reaching the set data amount threshold. The new service refers to a service or application that has not been run by a server deployed in a cabinet, and / or a service or application whose actual power consumption data generated by the model deployed in the cabinet has not reached the set data amount threshold.

[0034] In order to improve the accuracy of the server power consumption profile, especially the accuracy of the power consumption profile of a new service model, in some embodiments of the present application, a reference model and a reference service of the new service model can be obtained; and a historical power consumption profile of the reference model running the reference service, a power consumption dictionary of the reference model and a power consumption dictionary of the new model can be obtained; further, the initial power consumption profile of the to-be-tested model running the to-be-tested service (the new service model) can be determined according to the historical power consumption profile. Further, the actual power consumption data generated by the to-be-tested model running the to-be-tested service (the new service model) deployed in the cabinet can be obtained; and the initial power consumption profile is corrected by using the actual power consumption data, so as to realize the adaptive adjustment of the new service model profile, so that the obtained power consumption profile of the new service model is as close as possible to the actual power consumption profile, which helps to improve the accuracy of the power consumption profile.

[0035] The technical solutions provided by the embodiments of the present application will be described in detail below with reference to the drawings.

[0036] It should be noted that the same reference numerals represent the same objects in the following drawings and embodiments, so that once an object is defined in one drawing or embodiment, it does not need to be further discussed in subsequent drawings and embodiments.

[0037] In practical applications, the model of the server and the service software running on the server are two factors that affect the power consumption of the server. The power consumption of the server varies with different models; the power consumption generated by the same model running different service software is also different; of course, the power consumption generated by different models running the same service software can also be different. In the embodiments of the present application, the model running the service is taken as the granularity of the power consumption profile, and the power consumption profile of the to-be-tested model running the to-be-tested service is obtained. The main implementation manner is as follows:

[0038] Figure 1 The flowchart of the power consumption profiling method provided by the embodiments of the present application is shown in FIG. 1. Figure 1 As shown in FIG. 1, the power consumption profiling method comprises the following steps:

[0039] 101, obtaining a reference model of a to-be-tested model.

[0040] 102, obtaining a historical power consumption profile of a reference service of the reference model running the to-be-tested service.

[0041] 103, determining an initial power consumption profile of the to-be-tested model running the to-be-tested service according to the historical power consumption profile.

[0042] 104, obtaining actual power consumption data generated by the to-be-tested model running the to-be-tested service deployed in the cabinet.

[0043] 105, correcting the initial power consumption profile by using the actual power consumption data, so as to obtain the power consumption profile of the to-be-tested model running the to-be-tested service.

[0044] In the embodiments of the present application, the model running service is referred to as service model for convenience of description. For example, model A running service B can be referred to as service model AB, and the like. The power consumption profiling method provided in the embodiments of the present application is applicable to power consumption profiling of any service model running any service. Since there are sufficient actual power consumption data of mature model running mature service for power consumption profiling of mature service model (mature service model), the power consumption profiling method provided in the embodiments of the present application is mainly applied to new service model.

[0045] In the embodiments of the present application, for the to-be-tested service model, at least one of the to-be-tested model and the to-be-tested service is new, that is, the to-be-tested service model includes a new model and / or a new service. However, in the embodiments of the present application, the new model does not include the above-mentioned model without any server of the model deployed in the cabinet; and the new service does not include a service without any server deployed in the server deployed in the cabinet.

[0046] In the embodiments of the present application, since the data amount of the actual power consumption data generated by the to-be-tested model running the to-be-tested service has not reached the set data amount threshold, the power consumption profiling of the to-be-tested model running the to-be-tested service based on the actual power consumption data generated by the to-be-tested model running the to-be-tested service has low accuracy.

[0047] Therefore, in order to realize the power consumption profiling of the to-be-tested service model, in step 101, a reference model of the to-be-tested model can be acquired. Specifically, the reference model of the to-be-tested model can be acquired from the models deployed in the cabinet. The cabinet is a cabinet in a data center or a computer room. The reference model generates actual power consumption data with a data amount greater than or equal to a set data amount threshold. The set data amount threshold refers to a data amount of actual power consumption data with which the accuracy of power consumption profiling of the server can meet the set accuracy requirement, which can be flexibly set according to the accuracy requirement of power consumption profiling. The higher the accuracy requirement of power consumption profiling is, the greater the data amount threshold is.

[0048] Further, in step 102, a historical power consumption profile of a reference service of the to-be-tested service running by the reference model can be acquired. Alternatively, a service running by the reference model can be acquired; and the reference service of the to-be-tested service can be acquired from the service running by the reference model; further, the historical power consumption profile of the reference service running by the reference model can be acquired. The actual power consumption data generated by the reference model running the reference service has a data amount greater than or equal to the set data amount threshold.

[0049] In the embodiments of the present application, the specific implementation of the reference model of the to-be-tested model and the reference service of the to-be-tested service is not limited. In some embodiments, the reference model of each model can be preset by the technician or the maintenance personnel of the data center, and the correspondence between the model and the reference model is pre-stored in the device, the equipment or the module for power consumption profiling. Further, step 101 can be implemented by matching the to-be-tested model in the pre-set correspondence between the model and the reference model according to the identification of the to-be-tested model to obtain the reference model corresponding to the to-be-tested model.

[0050] Of course, the reference service of each service can also be preset by the technician or the maintenance personnel of the data center. Specifically, the technician and the maintenance personnel of the data center can obtain a service similar to the service from other services deployed in the data center as the reference service according to the service attribute of the service provided by the server demand side and the CPU load rate required by the service.

[0051] In the embodiments of the present application, the reference model of the to-be-tested model can be the to-be-tested model deployed in the cabinet. Alternatively, the to-be-tested service is the target service, which refers to the service whose actual power consumption data generated by the model deployed in the cabinet is greater than or equal to the set data amount threshold. Accordingly, when the to-be-tested service is the target service, the reference model is the model deployed in the cabinet that runs the to-be-tested service to meet the set requirement. For example, when the to-be-tested service is the target service, the reference model is the model deployed in the cabinet that runs the to-be-tested service the most, and the like. Alternatively, the reference model is the model deployed in the cabinet that meets the set condition in terms of the similarity of the power consumption dictionary and the power consumption dictionary of the to-be-tested model. For example, the reference model is the model deployed in the cabinet that has a similarity greater than or equal to the set similarity threshold in terms of the power consumption dictionary and the power consumption dictionary of the to-be-tested model; or the reference model is the model deployed in the cabinet that has the greatest similarity in terms of the power consumption dictionary and the power consumption dictionary of the to-be-tested model, and the like. Wherein, the power consumption dictionary refers to the relationship between the CPU load rate and the power consumption corresponding to the model. The method for obtaining the power consumption dictionary can refer to the related content of the above embodiments, which will not be repeated here.

[0052] Based on the above screening principle, in the embodiments of the present application, the reference model can also be automatically obtained by the computing device. Optionally, the number of target services running on the to-be-tested model deployed in the cabinet can be obtained; the target service refers to a service whose actual power consumption data generated by running the service has a data volume greater than or equal to a set data volume threshold; if the number of target services running on the to-be-tested model deployed in the cabinet is greater than or equal to a set number threshold, the to-be-tested model can be determined as the reference model. Correspondingly, if the number of target services running on the to-be-tested model deployed in the cabinet is less than the set number threshold, it can be determined whether the to-be-tested service is a target service; if the to-be-tested service is a target service, the model running the to-be-tested service that meets the set requirement can be determined as the reference model from the models deployed in the cabinet. For example, the model running the to-be-tested service the most can be determined as the reference model from the models deployed in the cabinet.

[0053] Further, if the number of target services running on the to-be-tested model deployed in the cabinet is less than the set number threshold and the to-be-tested service is not a target service, the reference model can be determined according to the power consumption dictionary of the model deployed in the cabinet and the power consumption dictionary of the to-be-tested model. Optionally, the similarity between the power consumption dictionary of the model deployed in the cabinet and the power consumption dictionary of the to-be-tested model can be determined according to the power consumption dictionary of the model deployed in the cabinet and the power consumption dictionary of the to-be-tested model; and the model deployed in the cabinet whose similarity meets the set condition can be selected as the reference model. For example, the model whose similarity between the power consumption dictionary and the power consumption dictionary of the to-be-tested model is greater than or equal to a set similarity threshold can be selected as the reference model from the models deployed in the cabinet; or the model whose similarity between the power consumption field and the power consumption field of the to-be-tested model is the greatest can be selected as the reference model from the models deployed in the cabinet.

[0054] The embodiments of obtaining the reference model of the to-be-tested model provided by the above embodiments are only exemplary and do not constitute a limitation.

[0055] After the reference model of the to-be-tested model is determined, the reference service of the to-be-tested service can be obtained from the services run by the reference model. The number of the reference services can be one or more. In some embodiments, all the services run by the reference model can be determined as the reference services of the to-be-tested service. In other embodiments, the service attribute of the to-be-tested service can be obtained. The service attribute refers to information that can represent the service property of the service. For example, for an online shopping service, the service attribute can be online shopping; for example, for a network car-hailing service, the service attribute can be network car-hailing, and the like. Further, the service with the same service attribute as the to-be-tested service can be obtained as the reference service according to the service attribute of the service run by the reference model, and the like. Alternatively, the service attribute and the CPU load rate of the to-be-tested service can be obtained. The CPU load rate of the to-be-tested service refers to the CPU load rate of the server running the service. Further, the service with the same service attribute as the to-be-tested service and the difference between the CPU load rates satisfying the set requirement can be obtained as the reference service from the services run by the reference model according to the service attribute and the CPU load rate of the service run by the reference model. For example, the service with the same service attribute as the to-be-tested service and the difference between the CPU load rates less than the set difference threshold can be obtained as the reference service from the services run by the reference model, and the like.

[0056] After the reference model of the to-be-tested model and the reference service of the to-be-tested service are determined, the historical power consumption profile of the reference model running the reference service can be obtained. In the embodiments of the present application, the power consumption profile of the reference model running the reference service can be real-time profiled, or the historical power consumption profile of the reference model running the reference service can be obtained from the pre-stored historical power consumption profiles of the reference model running various services, and the like.

[0057] In the embodiments of the present application, the specific expression form of the power consumption profile is not limited. In some embodiments, the power consumption profile can be represented by a mathematical model, such as a distribution function, a probability density function, and the like. Of course, the power consumption profile can also be represented by physical parameters, such as the mean power consumption and the standard deviation of power consumption, or the mean power consumption and the variance of power consumption, and the like. The mean power consumption can represent the mean power consumption of the model running the service, and the standard deviation of power consumption or the variance of power consumption can represent the fluctuation of the power consumption of the model running the service. Therefore, the mean power consumption and the standard deviation of power consumption (or the variance of power consumption) of the model can provide a basis for determining the maximum power consumption of the model. Therefore, the mean power consumption and the standard deviation of power consumption (or the variance of power consumption) can represent the power consumption profile.

[0058] For the embodiment of representing the power consumption image by the power consumption mean value and the power consumption standard deviation, the step 102 can be implemented as: calculating the power consumption mean value and the power consumption standard deviation of the reference model running the reference service according to the actual power consumption data generated by the reference model running the reference service, to obtain the historical power consumption image of the reference model running the reference service. The historical power consumption image of the reference model running the reference service is represented by the power consumption mean value and the power consumption standard deviation of the reference model running the reference service.

[0059] Since the historical power consumption image of the reference model running the reference service has a reference effect on the power consumption image of the test model running the test service to a certain extent, in the step 103, the initial power consumption image of the test model running the test service can be determined according to the historical power consumption image of the reference model running the reference service.

[0060] In the embodiments of the present application, the specific implementation of determining the initial power consumption image of the test model running the test service is not limited. In some embodiments, the historical power consumption image of the reference model running the reference service can be directly taken as the initial power consumption image of the test model running the test service. In other embodiments, the reference service is multiple, which means 2 or more. Correspondingly, the mean image of the historical power consumption images of the reference model running the multiple reference services can be calculated as the initial power consumption image of the test model running the test service, and so on.

[0061] In yet other embodiments, considering that the CPU load rate also affects the server power consumption, generally, the greater the CPU load rate of a server, the greater the power consumption, therefore, when determining the power consumption image of the test model running the test service according to the historical power consumption image of the reference model running the reference service, the power consumption image of the test model running the test service can be inferred from the historical power consumption image of the reference model running the reference service according to the equal CPU load rate principle. Based on this, the power consumption dictionary of the reference model and the power consumption dictionary of the test model can be obtained. The power consumption dictionary of the reference model can reflect the relationship between the CPU load rate and the power consumption of the reference model; the power consumption dictionary of the test model can reflect the relationship between the CPU load rate and the power consumption of the test model. The power consumption of the model is positively correlated with the CPU load rate, that is, the greater the CPU load rate, the greater the power consumption of the model. For the obtaining method of the power consumption dictionary of the reference model and the power consumption dictionary of the test model, please refer to the related content of the above embodiments, which will not be repeated here.

[0062] Further, the initial power consumption profile of the to-be-tested model running the to-be-tested service can be determined according to the power consumption dictionary of the reference model, the power consumption dictionary of the to-be-tested model, and the historical power consumption profile of the reference model running the reference service. Wherein, the representation of the power consumption profile is different, and the implementation of determining the initial power consumption profile of the to-be-tested model running the to-be-tested service is different. Hereinafter, taking the power consumption mean and the power consumption standard deviation representing the power consumption profile as an example, the specific implementation of determining the initial power consumption profile of the to-be-tested model running the to-be-tested service is exemplarily described.

[0063] For the embodiment that the historical power consumption profile of the reference model running the reference service is represented by the power consumption mean and the power consumption standard deviation, the historical power consumption profile of the reference model running the reference service is denoted as P0(i) = (μ i , σ i ). Wherein, i = 1, 2, …, n. n represents the number of the reference services, and n is a positive integer. μ i represents the power consumption mean generated by the reference model running the reference service i; and σ i represents the power consumption standard deviation generated by the reference model running the reference service i.

[0064] Correspondingly, the initial power consumption profile of the to-be-tested model running the to-be-tested service is denoted as represents the initial power consumption mean generated by the to-be-tested model running the to-be-tested service; represents the power consumption standard deviation of the initial power consumption generated by the to-be-tested model running the to-be-tested service.

[0065] Correspondingly, when determining the initial power consumption profile of the to-be-tested model running the to-be-tested service, the average value μ0 of the power consumption mean corresponding to the historical power consumption profile of the reference model running the reference service can be calculated. Wherein, the average value μ0 of the power consumption mean corresponding to the historical power consumption profile of the reference model running the reference service can be represented as:

[0066]

[0067] Further, the power consumption mean corresponding to the initial power consumption profile can be determined according to the power consumption dictionary of the reference model, the power consumption dictionary of the to-be-tested model, and the average value of the power consumption mean corresponding to the historical power consumption profile.

[0068] Specifically, the average value of the power consumption mean corresponding to the historical power consumption profile in the power consumption dictionary of the reference model can be determined according to the power consumption dictionary of the reference model, and the power consumption corresponding to the target CPU load rate in the power consumption dictionary of the to-be-tested model can be determined according to the power consumption dictionary corresponding to the to-be-tested model, as the power consumption mean corresponding to the initial power consumption profile of the to-be-tested model running the to-be-tested service.

[0069] The present application inventors found that the power consumption of a model has a linear relationship with the CPU load rate in the power consumption dictionary of the model, and the power consumption of the model increases with the increase of the CPU load rate. As shown in Figure 2 , it is assumed that the power consumption dictionary of a reference model is represented as: D0=k0x0+b0; where x0 represents the CPU load rate of the reference model; D0 represents the power consumption of the reference model, k0 represents the slope of the straight line of the power consumption of the reference model changing with the CPU load rate; and b0 represents the intercept of the straight line of the power consumption of the reference model changing with the CPU load rate. The power consumption dictionary of a to-be-tested model is represented as: D * =k * x * +b * . Where x * represents the CPU load rate of the to-be-tested model; D * represents the power consumption of the to-be-tested model, k * represents the slope of the straight line of the power consumption of the to-be-tested model changing with the CPU load rate; and b * represents the intercept of the straight line of the power consumption of the to-be-tested model changing with the CPU load rate.

[0070] Correspondingly, based on the linear relationship between the power consumption and the CPU load rate of a model, the average value μ0 of the power consumption mean value corresponding to the historical power consumption profile can be determined according to the power consumption dictionary of a reference model, and the target CPU load rate x(μ0) corresponding to the average value μ0 of the power consumption mean value in the power consumption dictionary of the reference model is:

[0071]

[0072] In formula (2), x1 and x2 represent two CPU load rates, where x1 01 represents the power consumption of the reference model corresponding to the CPU load rate x1 in the power consumption dictionary of the reference model; and D 02 represents the power consumption of the reference model corresponding to the CPU load rate x2 in the power consumption dictionary of the reference model. x(μ0) represents the CPU load rate corresponding to the average value μ0 of the power consumption mean value in the power consumption dictionary of the reference model, i.e. the target CPU load rate.

[0073] Further, the power consumption corresponding to the target CPU load rate in the power consumption dictionary of the to-be-tested model can be determined according to the power consumption dictionary corresponding to the to-be-tested model, and the power consumption corresponding to the target CPU load rate in the power consumption dictionary of the to-be-tested model is: the initial power consumption profile of the to-be-tested model running the to-be-tested service , where the power consumption corresponding to the target CPU load rate in the power consumption dictionary of the to-be-tested model is:

[0074]

[0075] ​In formula (3), D *1 represents the power consumption of the reference model corresponding to the CPU load rate x1 in the power consumption dictionary of the to-be-tested model. *2 represents the power consumption of the to-be-tested model corresponding to the CPU load rate x2 in the power consumption dictionary of the to-be-tested model.

[0076] Further, formula (2) is brought into formula (3), and the following formula (4) can be obtained:

[0077]

[0078] In the embodiments of the present application, the specific values of D 02 and D 01 are not limited. Alternatively, D 02 and D 01 may respectively take the power consumptions corresponding to the CPU load rates of 100% and 10% in the power consumption dictionary of the reference model. Correspondingly, the power consumption average value of the initial power consumption profile may be represented as:

[0079]

[0080] In formula (5), D0(100%) and D0(10%) respectively refer to the power consumptions corresponding to the CPU load rates of 100% and 10% in the power consumption dictionary of the reference model.

[0081] In the embodiments of the present application, when determining the initial power consumption profile of the to-be-tested model running the to-be-tested service, the standard deviation of the initial power consumption profile also needs to be determined. In the embodiments of the present application, for the power consumption standard deviation of the initial power consumption profile, the average value of the power consumption standard deviations of the historical power consumption profiles of the reference model running the reference service can be calculated as the power consumption standard deviation of the initial power consumption profile. The specific formula can be represented as:

[0082]

[0083] After the power consumption average value and the power consumption standard deviation of the initial power consumption profile of the to-be-tested model running the to-be-tested service are determined, the power consumption average value and the power consumption standard deviation can be used to represent the initial power consumption profile of the to-be-tested model running the to-be-tested service.

[0084] Since the accuracy of the initial power consumption profile of the to-be-tested model running the to-be-tested service is low, in order to improve the accuracy of the determined power consumption profile of the to-be-tested model running the to-be-tested service, in step 104, actual power consumption data generated by the to-be-tested model running the to-be-tested service deployed in the cabinet can also be acquired.

[0085] In the embodiment of the present application, the data center has a to-be-tested machine type deployed in the cabinet, and the to-be-tested machine type runs a to-be-tested service. However, when the power consumption profile of the to-be-tested machine type running the to-be-tested service is performed, the actual power consumption data generated by the to-be-tested machine type running the to-be-tested service has a data quantity less than the set data quantity threshold, and the actual power consumption data is insufficient to accurately profile the to-be-tested machine type running the to-be-tested service.

[0086] In the embodiment of the present application, in order to improve the accuracy of the power consumption profile of the to-be-tested machine type running the to-be-tested service, the initial power consumption profile obtained in step 103 can be corrected by using the actual power consumption data generated by the to-be-tested machine type running the to-be-tested service, so as to obtain the power consumption profile of the to-be-tested machine type running the to-be-tested service.

[0087] In the embodiment, since the initial power consumption profile of the to-be-tested machine type running the to-be-tested service is corrected by using the actual power consumption data generated by the to-be-tested machine type running the to-be-tested service, the adaptive adjustment of the power consumption profile of the to-be-tested machine type running the to-be-tested service is realized, which can make the obtained power consumption profile of the to-be-tested machine type running the to-be-tested service as close as possible to the actual power consumption profile, and helps to improve the accuracy of the power consumption profile of the to-be-tested machine type running the to-be-tested service.

[0088] In the embodiment of the present application, the specific implementation of correcting the initial power consumption profile obtained in step 103 by using the actual power consumption data generated by the to-be-tested machine type running the to-be-tested service is not limited. In some embodiments, the actual power consumption profile of the to-be-tested machine type running the to-be-tested service deployed in the cabinet can be determined according to the actual power consumption data generated by the to-be-tested machine type running the to-be-tested service, and the initial power consumption profile can be corrected by using the actual power consumption profile of the to-be-tested machine type running the to-be-tested service, so as to obtain the power consumption profile of the to-be-tested machine type running the to-be-tested service.

[0089] Specifically, the actual power consumption data generated by the to-be-tested machine type running the to-be-tested service corresponding to each profiling period (briefly described as the actual power consumption data corresponding to each profiling period) can be obtained. Further, the actual power consumption profile of the to-be-tested machine type running the to-be-tested service corresponding to each profiling period deployed in the cabinet can be determined according to the actual power consumption data corresponding to each profiling period. In the embodiment of the present application, the specific value of the profiling period is not limited. Alternatively, the profiling period can be 1 hour, 3 hours, 1 day, 1 week, etc.

[0090] For the embodiment in which the power consumption profile is represented by the power consumption mean and the power consumption standard deviation, the power consumption mean and the power consumption standard deviation of the to-be-tested machine type running the to-be-tested service deployed in the cabinet in each profiling period can be calculated according to the actual power consumption data corresponding to each profiling period, and the actual power consumption profile of the to-be-tested machine type running the to-be-tested service corresponding to each profiling period can be represented by the power consumption mean and the power consumption standard deviation of the profiling period, denoted as:

[0091] P * (tj ) = (μ * (t j ), σ * (t j )) (7).

[0092] In formula (7), t j represents the jth image cycle; j = 1, 2, …, m. m represents a set image cycle threshold. The image cycle threshold can be determined according to the data amount of the actual power consumption data generated by the to-be-tested service running on the to-be-tested machine in each image cycle. When the mth image cycle is reached, the actual power consumption data generated by the to-be-tested service running on the to-be-tested machine can reach the set data amount threshold, that is, when the mth image cycle is reached, the actual power consumption data generated by the to-be-tested service running on the to-be-tested machine can be accurately imaged. For example, 1 day is an image cycle, and m can represent that the actual power consumption data generated by the to-be-tested service running on the to-be-tested machine on the mth day can be accurately imaged. In the embodiments of the present application, the case that the actual power consumption data generated by the to-be-tested service running on the to-be-tested machine reaches the set data amount threshold is defined as the maturation of the to-be-tested service running on the to-be-tested machine.

[0093] The closer the actual power consumption data to the maturation time of the to-be-tested service running on the to-be-tested machine, the higher the similarity of the power consumption data generated at the maturation time of the to-be-tested service running on the to-be-tested machine, and the greater the value of the power consumption image of the to-be-tested service running on the to-be-tested machine. Based on this, when the initial power consumption image is corrected by using the actual power consumption image of the to-be-tested service running on the to-be-tested machine deployed in the cabinet, the weights of the actual power consumption images of the multiple image cycles can be determined according to the time sequence of the multiple image cycles. The closer the image cycle to the current time, the greater the weight of the actual power consumption image of the image cycle.

[0094] In actual application, there can be a case that the actual power consumption data generated by the to-be-tested service running on the to-be-tested machine is suddenly changed due to a sudden event on a certain day. For example, for a take-out service, the access amount on weekdays is large, and the access amount on holidays is small, so that the actual power consumption data generated by the to-be-tested service running on holidays is lower than that on weekdays. For example, for an online shopping service, the access amount on a certain preferential day is larger than that on other ordinary days, so that the actual power consumption generated by the online shopping service on the preferential day is suddenly increased compared with that on ordinary days. In the embodiments of the present application, in order to reduce the influence of the actual power consumption mutation caused by accidental factors, a weight update period can be set, and the weights of the image cycles in the same weight update period are the same. One weight update period can include multiple image cycles. For example, the image cycle is 1 day, and the weight update period can be one week.

[0095] Correspondingly, in determining the weights of the actual power consumption profiles of the plurality of profiling periods, the actual power consumption profile corresponding to each weight update period can be determined according to the time of the plurality of profiling periods; one weight update period includes a plurality of profiling periods; and the same weight coefficient is set for the actual power consumption profile in each weight update period according to the time sequence of the plurality of weight update periods; wherein the closer the weight update period is to the current time, the greater the weight coefficient of the actual power consumption profile of the weight update period.

[0096] Further, the weight of the actual power consumption profile of each profiling period can be determined according to the number and the weight coefficient of the actual power consumption profile of each profiling period. Alternatively, the product of the number and the weight coefficient of the actual power consumption profile of each profiling period is calculated as the weight of the actual power consumption profile of the profiling period. Correspondingly, the specific calculation formula can be represented as:

[0097] w(t j )=n(t j )×c(t j ) (8)。

[0098] In formula (8), w(t j ) represents the weight of the actual power consumption profile of the jth profiling period; n(t j ) represents the number of the actual power consumption profile of the jth profiling period; and c(t j ) represents the weight coefficient of the actual power consumption profile of the jth profiling period. Wherein the weight coefficients in the same weight update period are the same, and the closer the weight update period is to the current time, the greater the weight coefficient of the actual power consumption profile of the weight update period.

[0099] After obtaining the weights of the actual power consumption profiles of the plurality of profiling periods, the initial power consumption profile can be corrected according to the weights of the actual power consumption profiles of the plurality of profiling periods and the actual power consumption profiles of the plurality of profiling periods, to obtain the power consumption profile of the to-be-tested machine type running the to-be-tested service.

[0100] Specifically, the weight of the initial power consumption profile can be determined according to the weights of the actual power consumption profiles of the plurality of profiling periods. In the embodiments of the present application, the weights of the actual power consumption profiles of the plurality of profiling periods can be brought into the preset weight calculation formula of the initial power consumption profile to obtain the weight of the initial power consumption profile. Wherein the weight of the initial power consumption profile can be represented as:

[0101]

[0102] In formula (9), q represents the profiling period currently in; 1≤q≤m, and k is an integer. W max represents the maximum weight of the preset initial power consumption profile, which can be equal to the sum of the weights of the actual power consumption profiles of the preset m profiling periods, and can be represented as:

[0103]

[0104] Thus, after the cumulative period reaches the mth image period, the weight of the initial power consumption image is 0, and the initial power consumption image no longer plays a role in subsequent determination of the power consumption image of the to-be-tested machine type running the to-be-tested service, which can be determined by the actual power consumption image of the to-be-tested machine type running the to-be-tested service.

[0105] After the weight of the initial power consumption image and the weight of the actual power consumption image of the plurality of image periods are determined, the actual power consumption image of the plurality of image periods and the initial power consumption image can be weighted according to the weight of the initial power consumption image and the weight of the actual power consumption image of the plurality of image periods, to obtain the power consumption image of the to-be-tested machine type running the to-be-tested service.

[0106] In the following, the power consumption image is taken as an example of the power consumption mean and the power consumption standard deviation. Correspondingly, when the actual power consumption image of the plurality of image periods and the initial power consumption image are weighted, the power consumption mean corresponding to the actual power consumption image of the plurality of image periods and the power consumption mean corresponding to the initial power consumption image can be weighted according to the weight of the initial power consumption image and the weight of the actual power consumption image of the plurality of image periods, to obtain the power consumption mean corresponding to the power consumption image of the to-be-tested machine type running the to-be-tested service; and the power consumption standard deviation corresponding to the actual power consumption image of the plurality of image periods and the power consumption standard deviation corresponding to the initial power consumption image can be weighted according to the weight of the initial power consumption image and the weight of the actual power consumption image of the plurality of image periods, to obtain the power consumption standard deviation corresponding to the power consumption image of the to-be-tested machine type running the to-be-tested service.

[0107] Further, the power consumption image of the to-be-tested machine type running the to-be-tested service can be represented by the power consumption mean and the power consumption standard deviation corresponding to the power consumption image.

[0108] In the following, the implementation of the above-mentioned weighting of the actual power consumption image of the plurality of image periods and the initial power consumption image is exemplarily described in combination with specific calculation formulas. Specifically, the power consumption image of the to-be-tested machine type running the to-be-tested service can be calculated according to the following formula:

[0109]

[0110] In formula (11), T={init, t1, t2, …, t q}. Wherein, w(init) is the weight of the initial power consumption image. w(t k ) represents the weight of the actual power consumption image of the kth image period, k=1, 2, …, q.

[0111] Further, the power consumption mean of the power consumption image of the to-be-tested machine type running the to-be-tested service can be represented as:

[0112]

[0113] In equation (12), T = {init, t1, t2, ..., t} q}. μ * (init) represents the average power consumption of the initial power consumption profile; μ * (tk) represents the average power consumption of the actual image in the k-th image cycle.

[0114] Accordingly, the power consumption variance and standard deviation of the power consumption profile of the device under test running the service under test are expressed as follows:

[0115]

[0116]

[0117] In equation (13), s 2 This represents the power variance of the power consumption profile of the device under test running the service under test. Accordingly, in equation (14), This represents the standard deviation of the power consumption profile of the device under test running the service under test. In equation (13), T = init, t1, t2, ..., t q Correspondingly, σ * (init) represents the standard deviation of power consumption in the initial power consumption profile; σ * (t k ) represents the standard deviation of the power consumption of the actual power consumption of the k-th image period.

[0118] Based on the above equations (11)-(14), when weighting the average power consumption of the actual power consumption portraits of multiple portrait periods and the average power consumption of the initial power consumption portrait according to the weights of the actual power consumption portraits of multiple portrait periods and the weights of the initial power consumption portraits, the weights of the actual power consumption portraits of multiple portrait periods, the weights of the initial power consumption portraits, the average power consumption of the actual power consumption portraits of multiple portrait periods and the average power consumption of the initial power consumption portrait can be substituted into the above equations (11) and (12) to obtain the average power consumption of the power consumption portraits of the device under test running the service under test.

[0119] Further, in the weighting processing of the power consumption standard deviations corresponding to the actual power consumption images of the multiple image periods and the initial power consumption image, the weight of the actual power consumption images of the multiple image periods, the weight of the initial power consumption image, the power consumption standard deviations corresponding to the actual power consumption images of the multiple image periods and the initial power consumption image, and the power consumption mean corresponding to the power consumption image of the to-be-tested service running on the to-be-tested machine type are brought into formula (13), so as to calculate the power consumption variance corresponding to the power consumption image of the to-be-tested service running on the to-be-tested machine type; further, the power consumption variance corresponding to the power consumption image of the to-be-tested service running on the to-be-tested machine type is square root processed by using formula (14), so as to obtain the power consumption standard deviation corresponding to the power consumption image of the to-be-tested service running on the to-be-tested machine type.

[0120] Compared with directly using the power consumption mean and the power consumption standard deviation corresponding to the actual power consumption images of the multiple image periods and the initial power consumption image, and using the above formula (11)-(14) deduced by the embodiment of the present application to calculate the power consumption mean and the power consumption standard deviation corresponding to the to-be-tested service running on the to-be-tested machine type, the calculation efficiency can be improved.

[0121] The embodiment of the present application modifies the initial power consumption image by using the actual power consumption image of each image period on the basis of the initial power consumption image, and the weight of the initial power consumption image is gradually decreased and the weight of the actual power consumption image is gradually increased, so as to avoid the situation that the power consumption image of the to-be-tested service running on the to-be-tested machine type deviates from the actual situation due to improper selection of the reference service and / or the reference machine type, and to realize smooth transition from the initial power consumption image to the power consumption image of the mature service machine type along with the online running of the new service machine type, which is helpful to improve the accuracy of the power consumption image of the to-be-tested service running on the to-be-tested machine type.

[0122] In the embodiment of the present application, after the power consumption image of the to-be-tested service running on the to-be-tested machine type is determined, the servers of the to-be-tested machine type running the to-be-tested service can be deployed in the cabinet according to the power consumption image of the to-be-tested service running on the to-be-tested machine type. Alternatively, the maximum value of the sum of the power consumptions of the multiple to-be-tested machine types running the to-be-tested service in a future period of time can be predicted according to the power consumption image of the to-be-tested service running on the to-be-tested machine type; further, the number of the servers of the to-be-tested machine type running the to-be-tested service deployed in the cabinet can be determined according to the maximum value of the sum of the power consumptions of the multiple to-be-tested machine types running the to-be-tested service in the future period of time and the power consumption supported by the cabinet; and the servers of the to-be-tested machine type running the to-be-tested service are deployed in the cabinet according to the number. In the embodiment of the present application, the future period of time refers to a set prediction time period, and is not limited to a specific value, which can be one week in the future, one month in the future, or 15 days in the future, etc.

[0123] Since the power consumption portrait of the to-be-tested server model running the to-be-tested service determined by the embodiment of the present application has high accuracy, the accuracy of predicting the maximum power consumption of the to-be-tested server model running the to-be-tested service according to the power consumption portrait of the to-be-tested server model running the to-be-tested service is high, so that the quantity of the servers of the to-be-tested server model running the to-be-tested service deployed in the cabinet is determined with high accuracy. On the one hand, as many servers of the to-be-tested server model running the to-be-tested service as possible can be deployed in the cabinet to improve the power utilization rate of the cabinet; on the other hand, the quantity of as many servers of the to-be-tested server model running the to-be-tested service as possible deployed in the cabinet is reasonable, which can reduce the over-power risk.

[0124] It should be noted that the execution subject of each step of the method provided in the above embodiments can be the same device, or the method can also be executed by different devices as the execution subject. For example, the execution subject of steps 101 and 102 can be device A; for another example, the execution subject of step 101 can be device A, and the execution subject of step 102 can be device B; and the like.

[0125] In addition, in some of the processes described in the above embodiments and the accompanying drawings, a plurality of operations appearing in a specific order are included, but it should be clearly understood that these operations can be executed in the order appearing in the text or in parallel, and the serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel.

[0126] Correspondingly, the embodiment of the present application also provides a computer readable storage medium storing computer instructions, when the computer instructions are executed by one or more processors, causing the one or more processors to execute the steps in the power consumption portrait method described above.

[0127] The embodiment of the present application also provides a computer program product, comprising: a computer program. When the computer program is executed by one or more processors, causing the one or more processors to execute the steps in the power consumption portrait method described above. In the embodiment of the present application, the specific implementation form of the computer program product is not limited. In some embodiments, the computer program product can be realized as a plug-in, a software function module, etc. The computer program product can be embedded in a server on-shelf decision system to plan or manage the process of deploying servers in a cabinet, etc.

[0128] The power consumption portrait method provided by the embodiment of the present application can be adapted to places such as computer rooms and data centers that can accommodate a large number of servers. The power consumption portrait method provided by the embodiment of the present application will be exemplarily described below taking a data center as an example.

[0129] Figure 3a The structure of the data center provided by the embodiment of the present application is shown in the schematic diagram. As shown in FIG. 1, the data center comprises a plurality of cabinets 101, and each cabinet 101 comprises a plurality of servers 102.Figure 3a As shown, the data center 30 includes: multiple server racks 31, management equipment 32, and servers deployed in the racks. Figure 3a (Not shown in the image).

[0130] Server rack 31 serves as a container for servers and also provides them with power and communication resources. Management device 32 refers to the hardware device deployed with the aforementioned computer software products, capable of planning and managing the deployment of servers within the rack. Management device 32 can maintain information about rack 31 and the servers deployed on it. For example, management device 32 can store and maintain information such as the power consumption supported by rack 31. Management device 32 can also store and maintain historical actual power consumption data, historical power consumption profiles, and power consumption fields of the servers deployed on rack 31.

[0131] In this embodiment, the connection between the cabinet 31 and the management device 32 can be wireless or wired. Optionally, the cabinet 31 can communicate with the corresponding management device 32 via a mobile network. Accordingly, the mobile network standard can be any one of 2G (GSM), 2.5G (GPRS), 3G (WCDMA, TD-SCDMA, CDMA2000, UTMS), 4G (LTE), 4G+ (LTE+), 5G, WiMax, etc. Optionally, the cabinet 31 can also communicate with its corresponding management device 32 via Bluetooth, WiFi, infrared, etc.

[0132] In this embodiment, the model of the server deployed in the rack is not limited. The models deployed in rack 31 can be the same model or different models. In this embodiment, the model of the server deployed in the rack may include the model under test.

[0133] In this embodiment of the application, for the device under test, since the amount of actual power consumption data generated by the device under test running the service under test has not reached the set data volume threshold, the accuracy of the power consumption profile obtained by creating a power consumption profile based on the actual power consumption data generated by the device under test running the service under test is low.

[0134] Therefore, to create a power consumption profile for the server model under test, management device 32 can obtain a reference model from the server models deployed in rack 31. The amount of actual power consumption data generated by the reference model is greater than or equal to a set data volume threshold. The set data volume threshold refers to the amount of actual power consumption data that, when used to create a power consumption profile, meets the set accuracy requirements. This threshold can be flexibly set according to the accuracy requirements of the power consumption profile. The higher the accuracy requirement of the power consumption profile, the larger the data volume threshold.

[0135] Further, the management device 32 can acquire a historical power consumption profile of a reference service run by the reference model of the to-be-tested model. Alternatively, the reference service run by the reference model can be acquired; and the reference service of the to-be-tested service can be acquired from the reference service run by the reference model; further, the historical power consumption profile of the reference service run by the reference model can be acquired. In this case, the actual power consumption data generated by the reference service run by the reference model has a data quantity greater than or equal to a set data quantity threshold.

[0136] For specific implementation of the management device 32 acquiring the reference model of the to-be-tested model and the reference service of the to-be-tested service, refer to the related content of the method embodiment described above, which will not be repeated here.

[0137] In the embodiment of the present application, the reference model of the to-be-tested model can be the to-be-tested model deployed in the cabinet. Alternatively, the to-be-tested service is a target service; the target service refers to a service run by the model deployed in the cabinet, and the actual power consumption data generated by the service has a data quantity greater than or equal to a set data quantity threshold. Accordingly, when the to-be-tested service is a target service, the reference model is a model deployed in the cabinet and running the to-be-tested service that meets the set requirements. Alternatively, the reference model is a model deployed in the cabinet and having a similarity degree between the power consumption dictionary and the power consumption dictionary of the to-be-tested model that meets the set condition.

[0138] Based on the above screening principle, in the embodiment of the present application, the reference model can also be acquired automatically by the management device 32. For specific implementation of the management device acquiring the reference model, refer to the related content of acquiring the reference model of the to-be-tested model in the method embodiment described above, which will not be repeated here.

[0139] After determining the reference model of the to-be-tested model, the management device 32 can acquire the reference service of the to-be-tested service from the service run by the reference model. In this case, the number of reference services can be one or more. In some embodiments, all services run by the reference model can be determined as the reference service of the to-be-tested service. In other embodiments, the service attribute of the to-be-tested service can be acquired.

[0140] Alternatively, the management device 32 can acquire the service attribute and CPU load rate of the to-be-tested service. The CPU load rate of the to-be-tested service refers to the CPU load rate of the server running the service. Further, according to the service attribute and CPU load rate of the reference model, the service run by the reference model and having the same service attribute as the to-be-tested service and the difference between the CPU load rates meeting the set requirements can be acquired as the reference service.

[0141] After determining the reference model of the to-be-tested model and the reference service of the to-be-tested service, the management device 32 can acquire the historical power consumption profile of the reference model running the reference service. In the embodiments of the present application, the power consumption profile of the reference model running the reference service can be acquired in real time, or the historical power consumption profile of the reference model running the reference service can be acquired from the pre-stored historical power consumption profiles of the reference model running various services, and the like.

[0142] In the embodiments of the present application, the specific expression form of the power consumption profile is not limited. In some embodiments, the power consumption profile can be represented by the power consumption mean value and the power consumption standard deviation (or power consumption variance). For the embodiments in which the power consumption profile is represented by the power consumption mean value and the power consumption standard deviation, the management device 32 can calculate the power consumption mean value and the power consumption standard deviation of the reference model running the reference service according to the actual power consumption data generated by the reference model running the reference service, to obtain the historical power consumption profile of the reference model running the reference service. The historical power consumption profile of the reference model running the reference service is represented by the power consumption mean value and the power consumption standard deviation of the reference model running the reference service.

[0143] Since the historical power consumption profile of the reference model running the reference service has a reference effect on the power consumption profile of the to-be-tested model running the to-be-tested service to a certain extent, the management device 32 can determine the initial power consumption profile of the to-be-tested model running the to-be-tested service according to the historical power consumption profile of the reference model running the reference service.

[0144] In the embodiments of the present application, the specific implementation manner of determining the initial power consumption profile of the to-be-tested model running the to-be-tested service is not limited. In some embodiments, the historical power consumption profile of the reference model running the reference service can be directly used as the initial power consumption profile of the to-be-tested model running the to-be-tested service. In other embodiments, the reference service is multiple, which means 2 or more. Correspondingly, the management device 32 can calculate the mean value profile of the historical power consumption profiles of the reference model running multiple reference services, to use the mean value profile as the initial power consumption profile of the to-be-tested model running the to-be-tested service, and the like.

[0145] In still other embodiments, the management device 32 can acquire the power consumption dictionary of the reference model and the power consumption dictionary of the to-be-tested model. The description of the power consumption dictionary can be referred to the related content of the above-mentioned embodiments, which will not be repeated here.

[0146] Further, the management device 32 can determine the initial power consumption profile of the to-be-tested model running the to-be-tested service according to the power consumption dictionary of the reference model, the power consumption dictionary of the to-be-tested model, and the historical power consumption profile of the reference model running the reference service. The implementation manner of the management device 32 determining the initial power consumption profile of the to-be-tested model running the to-be-tested service can be referred to the related content of the above-mentioned method embodiments, which will not be repeated here.

[0147] Since the accuracy of the initial power consumption profile of the to-be-tested service running on the to-be-tested server is low, in order to improve the accuracy of the determined power consumption profile of the to-be-tested service running on the to-be-tested server, the management device 32 can further acquire actual power consumption data generated by the to-be-tested server running the to-be-tested service deployed in the rack.

[0148] In the embodiment of the present application, the data center has to-be-tested servers deployed in the rack, and the to-be-tested servers run to-be-tested services. However, when the power consumption profile of the to-be-tested service running on the to-be-tested server is determined, the amount of actual power consumption data generated by the to-be-tested service running on the to-be-tested server is less than the set data amount threshold, and the actual power consumption data is insufficient to accurately profile the to-be-tested service running on the to-be-tested server.

[0149] In the embodiment of the present application, in order to improve the accuracy of the power consumption profile of the to-be-tested service running on the to-be-tested server, the management device 32 can correct the initial power consumption profile by using the actual power consumption data generated by the to-be-tested service running on the to-be-tested server, and further obtain the power consumption profile of the to-be-tested service running on the to-be-tested server. For specific implementation of correcting the initial power consumption profile by using the actual power consumption data generated by the to-be-tested service running on the to-be-tested server, please refer to the related content of the method embodiment described above, which will not be repeated here.

[0150] In the embodiment, since the initial power consumption profile of the to-be-tested service running on the to-be-tested server is corrected by using the actual power consumption data generated by the to-be-tested service running on the to-be-tested server, the adaptive adjustment of the power consumption profile of the to-be-tested service running on the to-be-tested server is realized, which can make the obtained power consumption profile of the to-be-tested service running on the to-be-tested server as close as possible to the actual power consumption profile, and help to improve the accuracy of the power consumption profile of the to-be-tested service running on the to-be-tested server.

[0151] In the embodiment of the present application, after the power consumption profile of the to-be-tested service running on the to-be-tested server is determined, the management device 32 can deploy servers of the to-be-tested server running the to-be-tested service in the rack according to the power consumption profile of the to-be-tested service running on the to-be-tested server. Alternatively, the management device 32 can predict the maximum value of the sum of power consumptions of multiple to-be-tested servers running the to-be-tested service in a future period of time according to the power consumption profile of the to-be-tested service running on the to-be-tested server; further, the number of servers of the to-be-tested server running the to-be-tested service deployed in the rack can be determined according to the maximum value of the sum of power consumptions of multiple to-be-tested servers running the to-be-tested service in a future period of time and the power consumption supported by the rack; and the servers of the to-be-tested server running the to-be-tested service can be deployed in the rack according to the number, and so on.

[0152] Since the power consumption portrait of the to-be-tested machine type running the to-be-tested service determined by the embodiments of the present application has high accuracy, the accuracy of predicting the maximum power consumption of the to-be-tested machine type running the to-be-tested service according to the power consumption portrait of the to-be-tested machine type running the to-be-tested service is high, so that the number of servers of the to-be-tested machine type running the to-be-tested service deployed in the cabinet is determined with high accuracy. On the one hand, as many servers of the to-be-tested machine type running the to-be-tested service as possible can be deployed in the cabinet to improve the power utilization rate of the cabinet; on the other hand, the number of as many servers of the to-be-tested machine type running the to-be-tested service as possible deployed in the cabinet is reasonable, which can reduce the over-power risk.

[0153] Figure 3b A structural schematic diagram of a computing device is provided for the embodiments of the present application. In the embodiments of the present application, the implementation form of the computing device is not limited. The computing device can be a single server device, a clouded server array, or a virtual machine (VM) running in the clouded server array. In addition, the computing device can also refer to other computing devices with corresponding service capabilities, such as terminal devices (running service programs) such as computers, etc.

[0154] As shown in Figure 3b , the computing device includes a memory 30a and a processor 30b. The memory 30a is configured to store a computer program.

[0155] The processor 30b is coupled to the memory 30a and is configured to execute the computer program to: obtain a reference machine type of a to-be-tested machine type; obtain historical power consumption portraits of a reference service of the reference machine type running a to-be-tested service; determine an initial power consumption portrait of the to-be-tested machine type running the to-be-tested service according to the historical power consumption portraits; obtain actual power consumption data generated by the to-be-tested machine type running the to-be-tested service deployed in a cabinet; and correct the initial power consumption portrait by using the actual power consumption data to obtain the power consumption portrait of the to-be-tested machine type running the to-be-tested service.

[0156] In some embodiments, when the processor 30b corrects the initial power consumption portrait by using the actual power consumption data, it is specifically configured to: determine an actual power consumption portrait of the to-be-tested machine type running the to-be-tested service deployed in the cabinet according to the actual power consumption data; and correct the initial power consumption portrait by using the actual power consumption portrait to obtain the power consumption portrait of the to-be-tested machine type running the to-be-tested service.

[0157] Optionally, when the processor 30b determines the actual power consumption portrait of the to-be-tested machine type running the to-be-tested service deployed in the cabinet, it is specifically configured to: obtain actual power consumption data corresponding to a plurality of portrait periods respectively; and determine an actual power consumption portrait of each portrait period corresponding to the to-be-tested machine type running the to-be-tested service deployed in the cabinet according to the actual power consumption data corresponding to each portrait period.

[0158] Correspondingly, the processor 30b is specifically configured to determine the weights of the actual power consumption images of the plurality of image periods according to the time sequence of the plurality of image periods when correcting the initial power consumption image by using the actual power consumption images; and correct the initial power consumption image according to the weights of the actual power consumption images of the plurality of image periods and the actual power consumption image of each image period, to obtain the power consumption image of the to-be-tested machine type running the to-be-tested service.

[0159] Optionally, the processor 30b is specifically configured to determine the weight of the initial power consumption image according to the weights of the actual power consumption images of the plurality of image periods when correcting the initial power consumption image; and perform weighted processing on the actual power consumption images of the plurality of image periods and the initial power consumption image according to the weights of the actual power consumption images of the plurality of image periods and the weight of the initial power consumption image, to obtain the power consumption image of the to-be-tested machine type running the to-be-tested service.

[0160] Optionally, the power consumption image can be represented by a power consumption mean value and a power consumption standard deviation. Correspondingly, the processor 30b is specifically configured to perform weighted processing on the power consumption mean values corresponding to the actual power consumption images of the plurality of image periods and the initial power consumption image according to the weights of the actual power consumption images of the plurality of image periods and the weight of the initial power consumption image, to obtain the power consumption mean value corresponding to the power consumption image of the to-be-tested machine type running the to-be-tested service; and perform weighted processing on the power consumption standard deviations corresponding to the actual power consumption images of the plurality of image periods and the initial power consumption image according to the weights of the actual power consumption images of the plurality of image periods and the weight of the initial power consumption image, to obtain the power consumption standard deviation corresponding to the power consumption image of the to-be-tested machine type running the to-be-tested service.

[0161] In some other embodiments, the processor 30b is specifically configured to determine the actual power consumption image corresponding to each weight update period according to the time of the plurality of image periods; one weight update period includes a plurality of image periods; set the same weight coefficient for the actual power consumption images in each weight update period according to the time sequence of the plurality of weight update periods; determine the weight of the actual power consumption image of each image period according to the number of the actual power consumption images of the image period and the weight coefficient; and the weight coefficient of the actual power consumption image of the weight update period is greater when the weight update period is closer to the current time.

[0162] In some other embodiments, the processor 30b is specifically configured to determine the initial power consumption image of the to-be-tested machine type running the to-be-tested service according to the historical power consumption image, the power consumption dictionary of the reference machine type and the power consumption dictionary of the to-be-tested machine type; and the power consumption dictionary stores the relationship between the CPU load rate and the power consumption of the corresponding machine type.

[0163] Optionally, the power consumption profile can be represented by a power consumption mean value and a power consumption standard deviation. Accordingly, the processor 30b, in determining the initial power consumption profile of the to-be-tested machine type running the to-be-tested service, is specifically configured to: calculate an average of power consumption mean values corresponding to the historical power consumption profiles; determine, according to the power consumption dictionary of the reference machine type, the power consumption dictionary of the to-be-tested machine type, and the average of the power consumption mean values corresponding to the historical power consumption profiles, a power consumption mean value corresponding to the initial power consumption profile; and calculate an average of power consumption standard deviations corresponding to the historical power consumption profiles as a power consumption standard deviation corresponding to the initial power consumption profile.

[0164] Further, the processor 30b, in determining the power consumption mean value corresponding to the initial power consumption profile, is specifically configured to: determine, according to the power consumption dictionary of the reference machine type, a target CPU load rate corresponding to the average of the power consumption mean values corresponding to the historical power consumption profiles in the power consumption dictionary of the reference machine type; and determine, according to the power consumption dictionary corresponding to the to-be-tested machine type, a power consumption corresponding to the target CPU load rate in the power consumption dictionary corresponding to the to-be-tested machine type as the power consumption mean value corresponding to the initial power consumption profile.

[0165] In the embodiments of the present application, the processor 30b, in obtaining the reference machine type of the to-be-tested machine type, is specifically configured to: obtain, from the machine types deployed in the cabinet, the reference machine type of the to-be-tested machine type. The cabinet can be a cabinet in a data center or a computer room.

[0166] The reference machine type is the to-be-tested machine type deployed in the cabinet; the actual power consumption data generated by the reference machine type running the reference service has a data quantity greater than or equal to a set threshold; or the to-be-tested service is a target service; the reference machine type is a machine type deployed in the cabinet and running the to-be-tested service that meets a set requirement; the target service refers to a service running which generates actual power consumption data having a data quantity greater than or equal to a set data quantity threshold; or the reference machine type is a machine type deployed in the cabinet and having a similarity degree between the power consumption dictionary and the power consumption dictionary of the to-be-tested machine type meeting a set condition.

[0167] In some embodiments of the present application, the processor 30b, in obtaining the reference machine type of the to-be-tested machine type, is specifically configured to: obtain a number of target services running on the to-be-tested machine type deployed in the cabinet; the target service refers to a service running which generates actual power consumption data having a data quantity greater than or equal to a set data quantity threshold; and if the number of target services is greater than or equal to a set number threshold, determine that the to-be-tested machine type is the reference machine type. Accordingly, if the number of target services is less than the set number threshold and the to-be-tested service is the target service, determine, from the machine types deployed in the cabinet, a machine type running the to-be-tested service most frequently as the reference machine type. Further, if the number of target services is less than the set number threshold and the to-be-tested service is not the target service, determine the reference machine type according to the power consumption dictionary of the machine type deployed in the cabinet and the power consumption dictionary of the to-be-tested machine type.

[0168] Optionally, when determining the reference model, the processor 30b is specifically used to: determine the similarity between the power consumption dictionary of the model deployed in the rack and the power consumption dictionary of the model under test, based on the power consumption dictionary of the model deployed in the rack and the power consumption dictionary of the model under test; and select the model whose similarity meets the set conditions from the models deployed in the rack as the reference model.

[0169] In some embodiments, the processor 30b is further configured to: after determining the power consumption profile of the machine under test running the service under test, deploy the server of the machine under test running the service under test in the rack according to the power consumption profile of the machine under test running the service under test and the power information supported by the rack.

[0170] In some alternative implementations, such as Figure 3b As shown, the computing device may further include components such as a communication component 30c and a power supply component 30d. In some embodiments, the computing device may be a terminal device such as a computer or a mobile phone. Accordingly, the computing device may also include optional components such as a display component 30e and an audio component 30f. Figure 3b The diagram only shows some components and does not mean that the computing device must contain them. Figure 3b The inclusion of all components does not imply that a computing device can only include... Figure 3b The components shown.

[0171] The computing device provided in this application embodiment can obtain a reference model and reference service for the new service model; and obtain the historical power consumption profile of the reference model running the reference service, the power consumption dictionary of the reference model, and the power consumption dictionary of the new model; further, based on the historical power consumption profile, the initial power consumption profile of the model under test running the service under test (new service model) can be determined. Further, the actual power consumption data generated by the model under test running the service under test (new service model) deployed in the rack can be obtained; and the initial power consumption profile can be corrected using the actual power consumption data to achieve adaptive adjustment of the new service model profile, making the obtained power consumption profile of the new service model as close as possible to its actual power consumption profile, thus helping to improve the accuracy of the power consumption profile.

[0172] In this embodiment, the memory is used to store computer programs and can be configured to store various other data to support operation on its host device. The processor can execute the computer programs stored in the memory to implement corresponding control logic. The memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0173] In embodiments of the present application, the processor can be any hardware processing device that can execute the logic of the above-described methods. Optionally, the processor can be a central processing unit (CPU), a graphics processing unit (GPU), or a microcontroller unit (MCU); it can also be a programmable device such as a field-programmable gate array (FPGA), a programmable array logic (PAL), a generic array logic (GAL), a complex programmable logic device (CPLD), etc.; or it can be an advanced RISC machines (ARM) processor or a system on chip (SOC), etc., but is not limited thereto.

[0174] In embodiments of the present application, the communication component is configured to facilitate wired or wireless communication between the device in which it is located and other devices. The device in which the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, 4G, 5G, or a combination thereof. In an example embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component can also be implemented based on near field communication (NFC) technology, radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra wideband (UWB) technology, Bluetooth (BT) technology, or other technology.

[0175] In embodiments of the present application, the display component can include a liquid crystal display (LCD) and a touch panel (TP). If the display component includes a touch panel, the display component can be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense a touch, a slide, and a gesture on the touch panel. The touch sensor can not only sense a boundary of a touch or a slide action, but also detect a duration and a pressure related to the touch or slide action.

[0176] In embodiments of the present application, the power supply component is configured to provide power to various components of the device in which it is located. The power supply component can include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device in which the power supply component is located.

[0177] In the embodiments of the present application, the audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC) that is configured to receive an external audio signal when the device in which the audio component is located is in an operation mode, such as a calling mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory or transmitted via the communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals. For example, for a device with a language interaction function, voice interaction with a user can be realized through the audio component.

[0178] It should be noted that the "first", "second", and the like descriptions herein are used to distinguish different messages, devices, modules, and the like, and do not represent the order of sequence, nor limit the "first" and "second" to be different types.

[0179] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) containing computer-usable program code.

[0180] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as a combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0181] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0182] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 Figure 1 Figure 1

[0183] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0184] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) about which the computer stores the information. The memory is an example of computer readable media.

[0185] The computer's storage media can be implemented using any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disks (DVDs) or other optical storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information for access by a computing device. In accordance with the teachings herein, computer readable media does not include transitory media, such as modulated data signals and carrier waves.

[0186] It is also important to note that the terms "comprises", "comprising", or other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0187] ​​The above merely provides an example of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the scope of claims of the present application.

Claims

1. A power consumption profiling method, characterized by, The method comprises the following steps: obtaining a reference model of a to-be-tested model; obtaining a historical power consumption profile of a reference service of the reference model running the to-be-tested service; determining an initial power consumption profile of the to-be-tested model running the to-be-tested service according to the historical power consumption profile; obtaining actual power consumption data of the to-be-tested model running the to-be-tested service deployed in a cabinet; correcting the initial power consumption profile by using the actual power consumption data to obtain a power consumption profile of the to-be-tested model running the to-be-tested service; wherein the actual power consumption data comprises actual power consumption data corresponding to a plurality of profile periods; and the correcting the initial power consumption profile by using the actual power consumption data comprises: determining an actual power consumption profile of each profile period corresponding to the to-be-tested model running the to-be-tested service deployed in the cabinet according to the actual power consumption data corresponding to each profile period; determining weights of the actual power consumption profiles of the plurality of profile periods according to a time sequence of the plurality of profile periods; correcting the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the actual power consumption profile of each profile period to obtain the power consumption profile of the to-be-tested model running the to-be-tested service.

2. The method of claim 1, wherein, The correcting the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the actual power consumption profile of each profile period comprises: determining weights of the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods; performing weighted processing on the actual power consumption profiles of the plurality of profile periods and the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the weights of the initial power consumption profile to obtain the power consumption profile of the to-be-tested model running the to-be-tested service.

3. The method of claim 2, wherein, The power consumption profile is characterized by a power consumption mean and a power consumption standard deviation; and the performing weighted processing on the actual power consumption profiles of the plurality of profile periods and the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the weights of the initial power consumption profile comprises: performing weighted processing on power consumption means corresponding to the actual power consumption profiles of the plurality of profile periods and power consumption means corresponding to the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the weights of the initial power consumption profile to obtain power consumption means corresponding to the power consumption profile of the to-be-tested model running the to-be-tested service; performing weighted processing on power consumption standard deviations corresponding to the actual power consumption profiles of the plurality of profile periods and power consumption standard deviations corresponding to the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the weights of the initial power consumption profile to obtain power consumption standard deviations corresponding to the power consumption profile of the to-be-tested model running the to-be-tested service.

4. The method of claim 1, wherein, The determining the weights of the actual power consumption profiles of the plurality of profile periods according to the time sequence of the plurality of profile periods comprises: determining actual power consumption profiles corresponding to each weight update period according to the time of the plurality of profile periods; one weight update period comprises a plurality of profile periods; setting the same weight coefficient for the actual power consumption profiles in each weight update period according to a time sequence of the plurality of weight update periods; and According to the number and weight coefficient of the actual power consumption image of each image cycle, the weight of the actual power consumption image of the image cycle is determined; The closer the weight update cycle is to the current time, the greater the weight coefficient of the actual power consumption image of the weight update cycle.

5. The method of claim 1, wherein, The initial power consumption image of the to-be-tested service running on the to-be-tested machine type is determined according to the historical power consumption image. The initial power consumption image of the to-be-tested service running on the to-be-tested machine type is determined according to the historical power consumption image, the power consumption dictionary of the reference machine type and the power consumption dictionary of the to-be-tested machine type. The power consumption dictionary stores the relationship between the CPU load rate and the power consumption of the corresponding machine type.

6. The method of claim 5, wherein, The power consumption image is represented by the power consumption mean and the power consumption standard deviation. The initial power consumption image of the to-be-tested service running on the to-be-tested machine type is determined according to the historical power consumption image, the power consumption dictionary of the reference machine type, and the power consumption dictionary of the to-be-tested machine type. The average value of the power consumption mean corresponding to the historical power consumption image is calculated. The power consumption mean corresponding to the initial power consumption image is determined according to the power consumption dictionary of the reference machine type, the power consumption dictionary of the to-be-tested machine type and the average value of the power consumption mean corresponding to the historical power consumption image. The average value of the power consumption standard deviation corresponding to the historical power consumption image is calculated as the power consumption standard deviation corresponding to the initial power consumption image.

7. The method of claim 6, wherein, The power consumption mean corresponding to the initial power consumption image is determined according to the power consumption dictionary of the reference machine type, the power consumption dictionary of the to-be-tested machine type and the average value of the power consumption mean corresponding to the historical power consumption image. The target CPU load rate corresponding to the average value of the power consumption mean corresponding to the historical power consumption image in the power consumption dictionary of the reference machine type is determined according to the power consumption dictionary of the reference machine type. The power consumption corresponding to the target CPU load rate in the power consumption dictionary corresponding to the to-be-tested machine type is determined as the power consumption mean corresponding to the initial power consumption image according to the power consumption dictionary corresponding to the to-be-tested machine type.

8. The method of claim 1, wherein, The reference machine type is the to-be-tested machine type deployed in the cabinet; the actual power consumption data generated by the reference machine type running the reference service has a data volume greater than or equal to a set threshold; Or, The to-be-tested service is a target service; the reference machine type is a machine type deployed in the cabinet and running the to-be-tested service that meets the set requirements; the target service refers to a service whose actual power consumption data generated by the machine type deployed in the cabinet has a data volume greater than or equal to a set data volume threshold; Or, The reference machine type is a machine type deployed in the cabinet and having a similarity degree between the power consumption dictionary and the power consumption dictionary of the to-be-tested machine type that meets a set condition.

9. The method according to any one of claims 1 to 8, characterized in that, Further comprising: According to the power consumption image of the to-be-tested machine type running the to-be-tested service and the power information supported by the cabinet, a server of the to-be-tested machine type running the to-be-tested service is deployed in the cabinet.

10. A data center, comprising: Comprising: Multiple cabinets, servers deployed in the cabinets and a management device; The machine type of the server deployed in the cabinet comprises: a to-be-tested machine type; The management device is configured to: obtain a reference model of the to-be-tested model from models of servers deployed in the cabinet; obtain a historical power consumption profile of a reference service run by the reference model; determine an initial power consumption profile of the to-be-tested model running the to-be-tested service according to the historical power consumption profile; obtain actual power consumption data of the to-be-tested model running the to-be-tested service deployed in the data center; and correct the initial power consumption profile by using the actual power consumption data to obtain a power consumption profile of the to-be-tested model running the to-be-tested service; wherein the actual power consumption data includes actual power consumption data corresponding to a plurality of profile periods; When correcting the initial power consumption profile by using the actual power consumption data, the management device is specifically configured to: determine an actual power consumption profile of each profile period corresponding to the to-be-tested model running the to-be-tested service deployed in the cabinet according to actual power consumption data corresponding to each profile period; Determine weights of the actual power consumption profiles of the plurality of profile periods according to a time sequence of the plurality of profile periods; Correct the initial power consumption profile according to the weights of the actual power consumption profiles of the plurality of profile periods and the actual power consumption profile of each profile period to obtain the power consumption profile of the to-be-tested model running the to-be-tested service.

11. A computing device, comprising: Comprise: A memory and a processor; wherein the memory is configured to store a computer program; The processor is coupled to the memory and is configured to execute the computer program to perform the steps in the method of any one of claims 1-9.

12. A computer readable storage medium having stored thereon computer instructions, wherein, When the computer instructions are executed by one or more processors, the one or more processors are caused to perform the steps in the method of any one of claims 1-9.

Images