A physical machine rack density optimization method, device and equipment

By acquiring historical power consumption data of physical machines, calculating the cumulative distribution function and probability density distribution, and determining the rack density of physical machines, the problem of low rack resource utilization in traditional methods is solved, achieving more efficient resource utilization and power management.

CN111831095BActive Publication Date: 2024-11-19ALIBABA GROUP HOLDING LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201910302220.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2019-04-16
Publication Date
2024-11-19
Estimated Expiration
2039-04-16

AI Technical Summary

Technical Problem

In data centers, traditional methods for configuring physical machine rack density are not accurate enough, resulting in racks being unable to meet the needs of physical machines or wasting resources, which affects business stability and resource utilization.

Method used

By acquiring historical power consumption data from multiple physical machines, the cumulative distribution function and probability density distribution are determined, the target probability that the sum of power consumption does not exceed the rack power consumption threshold is calculated, and the rack density of physical machines is determined based on the target probability.

Benefits of technology

It achieves improved power efficiency and resource utilization, reduced PUE, dynamic adjustment of the number of physical machines, refined rack deployment strategy, and balance of power consumption risks and benefits while ensuring the safe operation of physical machines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111831095B_ABST
    Figure CN111831095B_ABST
Patent Text Reader

Abstract

The present application provides a method, device and equipment for optimizing the density of physical machines on a rack, the method comprising: obtaining historical power consumption data of multiple physical machines; determining a cumulative distribution function based on the historical power consumption data of multiple physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value; determining the target probability that the sum of the power consumption does not exceed the rack power consumption threshold based on the cumulative distribution function; determining the rack density of physical machines on a rack based on the number of physical machines corresponding to the target probability. Through the technical solution of the present application, the power efficiency can be improved while ensuring the safe operation of the physical machines.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a method, device and equipment for optimizing the rack density of physical machines. Background Art

[0002] For data centers, while ensuring business stability, it is necessary to reduce PUE (Power Usage Effectiveness), improve power efficiency, and increase resource utilization. Data centers usually deploy a large number of physical machines in racks. Therefore, in order to improve the resource utilization of data centers, it is necessary to increase the rack density of physical machines while ensuring the safety of physical machine operation.

[0003] In the traditional way, the physical machine rack density can be configured using empirical values. The rack density is the number of physical machines deployed on the rack. Since each rack in the data center has a rated power consumption, if the number of physical machines deployed on the rack is too large, the rack cannot meet the needs of these physical machines. If the number of physical machines deployed on the rack is too small, these physical machines cannot fully utilize the resources of the rack, resulting in resource waste. Summary of the invention

[0004] The present application provides a method for optimizing the density of physical machines on racks, the method comprising:

[0005] Obtain historical power consumption data of multiple physical machines;

[0006] Determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value;

[0007] Determine, according to the cumulative distribution function, a target probability that the sum of power consumption does not exceed a rack power consumption threshold;

[0008] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0009] The present application provides a method for optimizing the density of physical machines on racks, the method comprising:

[0010] Obtain historical power consumption data of multiple physical machines;

[0011] Determine the probability density distribution of the physical machines according to the historical power consumption data of the multiple physical machines;

[0012] Determine the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution;

[0013] Determine a cumulative distribution function according to the power consumption and the distribution probability, wherein the cumulative distribution function includes probability data that the sum of the power consumptions of a specified number of physical machines does not exceed the power consumption value;

[0014] Determine, according to the cumulative distribution function, a target probability that the sum of power consumption does not exceed a rack power consumption threshold;

[0015] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0016] The present application provides a method for optimizing the density of physical machines on racks, the method comprising:

[0017] Obtain historical power consumption data of multiple physical machines;

[0018] Determine cumulative distribution data according to the historical power consumption data of the multiple physical machines;

[0019] The physical machine rack density of the rack is determined according to the cumulative distribution data.

[0020] The present application provides a method for optimizing the density of physical machines on racks, the method comprising:

[0021] Obtain historical power consumption data of multiple physical machines;

[0022] Determine probability data based on the historical power consumption data of the multiple physical machines; wherein the probability data includes probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value;

[0023] Determine, based on the probability data, a target probability that the sum of power consumption does not exceed a rack power consumption threshold;

[0024] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0025] The present application provides a physical machine rack density optimization device, the device comprising:

[0026] An acquisition module is used to obtain historical power consumption data of multiple physical machines;

[0027] A determination module, configured to determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, wherein the cumulative distribution function includes probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value; and determine a target probability that the sum of the power consumption does not exceed a rack power consumption threshold according to the cumulative distribution function;

[0028] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0029] The present application provides a physical machine rack density optimization device, including:

[0030] A processor and a machine-readable storage medium, wherein the machine-readable storage medium stores a plurality of computer instructions, and when the processor executes the computer instructions, the processor performs the following processing:

[0031] Obtain historical power consumption data of multiple physical machines;

[0032] Determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value;

[0033] Determine, according to the cumulative distribution function, a target probability that the sum of power consumption does not exceed a rack power consumption threshold;

[0034] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0035] Based on the above technical solution, in an embodiment of the present application, a cumulative distribution function can be determined based on the historical power consumption data of multiple physical machines, and the cumulative distribution function includes probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value. In this way, the target probability that the sum of the power consumption does not exceed the rack power consumption threshold can be determined based on the cumulative distribution function, and the physical machine rack density of the rack can be determined based on the number of physical machines corresponding to the target probability. Based on the above method, the rack density of the physical machine can be accurately determined, and the power efficiency, resource utilization and PUE can be improved while ensuring the safe operation of the physical machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments of the present application or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings of the embodiments of the present application.

[0037] Figure 1 It is a flow chart of a method for optimizing the density of physical machines on racks in one embodiment of the present application;

[0038] Figure 2 It is a flow chart of a method for optimizing the density of physical machines on racks in one embodiment of the present application;

[0039] Figure 3A and Figure 3B is a schematic diagram of physical machine deployment in one embodiment of the present application;

[0040] Figure 4 It is a structural schematic diagram of a physical machine rack density optimization device in one embodiment of the present application;

[0041] Figure 5 It is a structural diagram of a physical machine rack density optimization device in one implementation of the present application. DETAILED DESCRIPTION

[0042] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, rather than limiting the present application. The singular forms of "a", "said" and "the" used in the present application and claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to any or all possible combinations of one or more associated listed items.

[0043] It should be understood that, although the terms first, second, third, etc. may be used to describe various information in the embodiments of the present application, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, in addition, the word "if" used may be interpreted as "at..." or "when..." or "in response to determination".

[0044] The present application embodiment provides a physical machine rack density optimization method, which can be applied to any device, such as a physical machine rack density optimization device, see Figure 1 As shown, the method may include:

[0045] Step 101: Obtain historical power consumption data of multiple physical machines.

[0046] Specifically, the historical power consumption data of multiple physical machines can be obtained based on the characteristic parameters of the physical machines (such as the model, manufacturer model, business type, etc. of the physical machines); wherein the characteristic parameters of different physical machines may be the same or different, that is, the historical power consumption data of multiple physical machines with the same characteristic parameters can be obtained, or the historical power consumption data of multiple physical machines with different characteristic parameters can be obtained.

[0047] Step 102 : determining a cumulative distribution function according to historical power consumption data of a plurality of physical machines, wherein the cumulative distribution function includes probability data that the sum of power consumptions of a specified number of physical machines does not exceed a power consumption value.

[0048] For example, the cumulative distribution function includes probability data that the sum of power consumption of two physical machines does not exceed power consumption value A, probability data that the sum of power consumption of two physical machines does not exceed power consumption value B; probability data that the sum of power consumption of three physical machines does not exceed power consumption value A, probability data that the sum of power consumption of three physical machines does not exceed power consumption value B, and so on. The specified number may not exceed the number of the multiple physical machines.

[0049] Determining a cumulative distribution function based on historical power consumption data of multiple physical machines may include: determining a probability density distribution of the physical machines based on the historical power consumption data of the multiple physical machines; determining the power consumption and distribution probability of a specified number of physical machines based on the probability density distribution; and determining a cumulative distribution function based on the power consumption and distribution probability.

[0050] In an example, determining the power consumption and distribution probability of a specified number of physical machines based on the probability density distribution may include but is not limited to: if the specified number is 2, the power consumption and distribution probability of the two physical machines may be determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; if the specified number is N, the power consumption and distribution probability of N physical machines may be determined by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine; N is an integer greater than 2.

[0051] Determining the power consumption and distribution probability of the two physical machines by using the probability density distribution of one physical machine and the probability density distribution of another physical machine may include: performing a convolution operation on the probability density distribution of one physical machine and the probability density distribution of another physical machine to obtain the power consumption and distribution probability of the two physical machines.

[0052] Determining the power consumption and distribution probability of N physical machines by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine may include: performing a convolution operation on the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine to obtain the power consumption and distribution probability of 2 physical machines.

[0053] In an example, determining a cumulative distribution function based on power consumption and distribution probability may include but is not limited to: determining probability data that the sum of power consumption of a specified number of physical machines does not exceed the power consumption value based on the power consumption and distribution probability, and then determining the cumulative distribution function based on the probability data.

[0054] Step 103: Determine the target probability that the sum of the power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function. The rack power consumption threshold may be the rated power consumption of the rack, such as 5000 watts, 8000 watts, etc.

[0055] According to the cumulative distribution function, multiple probability data whose sum of power consumption does not exceed the rack power consumption threshold are determined; different probability data correspond to different numbers of physical machines; a target probability is selected from the multiple probability data; the target probability is less than or equal to the probability threshold, and the target probability is the probability data closest to the probability threshold.

[0056] Step 104: Determine the physical machine density of the rack according to the number of physical machines corresponding to the target probability.

[0057] Furthermore, after the physical machine rack density of the rack (ie, the maximum number of physical machines deployed on the rack) is determined, physical machines matching the physical machine rack density may be deployed on the rack.

[0058] In one example, the above execution order is just an example given for the convenience of description. In practical applications, the execution order between the steps can also be changed, and there is no limitation on this execution order. Moreover, in other embodiments, the steps of the corresponding method are not necessarily executed in the order shown and described in this specification, and the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; multiple steps described in this specification may also be combined into a single step for description in other embodiments.

[0059] Based on the above technical solution, in an embodiment of the present application, a cumulative distribution function can be determined based on the historical power consumption data of multiple physical machines, and the cumulative distribution function includes probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value. In this way, the target probability that the sum of the power consumption does not exceed the rack power consumption threshold can be determined based on the cumulative distribution function, and the physical machine rack density of the rack can be determined based on the number of physical machines corresponding to the target probability. Based on the above method, the rack density of the physical machine can be accurately determined, and the power efficiency, resource utilization and PUE can be improved while ensuring the safe operation of the physical machine.

[0060] Based on the same application concept as the above method, another physical machine rack density optimization method is proposed in an embodiment of the present application, which method may include: obtaining historical power consumption data of multiple physical machines, and determining the probability density distribution of the physical machines based on the historical power consumption data of the multiple physical machines. Determine the power consumption and distribution probability of a specified number of physical machines based on the probability density distribution, and determine the cumulative distribution function based on the power consumption and distribution probability, wherein the cumulative distribution function includes probability data that the sum of the power consumption of the specified number of physical machines does not exceed the power consumption value. Then, determine the target probability that the sum of the power consumption does not exceed the rack power consumption threshold based on the cumulative distribution function, and determine the physical machine rack density of the rack based on the number of physical machines corresponding to the target probability.

[0061] In an example, the power consumption and distribution probability of a specified number of physical machines are determined according to the probability density distribution, which may include but is not limited to: if the specified number is 2, the power consumption and distribution probability of the two physical machines can be determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; if the specified number is N, the power consumption and distribution probability of N physical machines can be determined by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine; N is an integer greater than 2.

[0062] Based on the same application concept as the above method, an embodiment of the present application proposes another method for optimizing physical machine rack density, including: obtaining historical power consumption data of multiple physical machines; determining cumulative distribution data based on the historical power consumption data of multiple physical machines; and determining the physical machine rack density of the rack based on the cumulative distribution data.

[0063] In an example, after determining the physical machine rack density of the rack according to the cumulative distribution data, physical machines matching the physical machine rack density may be deployed on the rack.

[0064] In one example, the cumulative distribution data may be a cumulative distribution function. Of course, the cumulative distribution data may also be other types of data, which is not limited. When the cumulative distribution data is a cumulative distribution function, the processing flow of this embodiment can refer to the above embodiment, and will not be repeated here.

[0065] Based on the same application concept as the above method, another physical machine rack density optimization method is also proposed in the embodiment of the present application. The method may include the following steps: obtaining historical power consumption data of multiple physical machines, and determining probability data based on the historical power consumption data of the multiple physical machines; wherein the probability data may include probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value. Furthermore, the target probability that the sum of the power consumption does not exceed the rack power consumption threshold can be determined based on the probability data, and the physical machine rack density of the rack can be determined based on the number of physical machines corresponding to the target probability.

[0066] Among them, determining the probability data based on the historical power consumption data of the multiple physical machines may include: determining the probability density distribution of the physical machines based on the historical power consumption data of the multiple physical machines; determining the power consumption and distribution probability of a specified number of physical machines based on the probability density distribution; and determining the probability data based on the power consumption and distribution probability. For example, determining a cumulative distribution function based on the power consumption and distribution probability, the cumulative distribution function includes probability data that the sum of the power consumption of the specified number of physical machines does not exceed the power consumption value.

[0067] In an example, determining the power consumption and distribution probability of a specified number of physical machines based on the probability density distribution may include but is not limited to: if the specified number is 2, the power consumption and distribution probability of the two physical machines may be determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; if the specified number is N, the power consumption and distribution probability of N physical machines may be determined by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine; N is an integer greater than 2.

[0068] Determining the power consumption and distribution probability of the two physical machines by using the probability density distribution of one physical machine and the probability density distribution of another physical machine may include: performing a convolution operation on the probability density distribution of one physical machine and the probability density distribution of another physical machine to obtain the power consumption and distribution probability of the two physical machines.

[0069] Determining the power consumption and distribution probability of N physical machines by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine may include: performing a convolution operation on the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine to obtain the power consumption and distribution probability of 2 physical machines.

[0070] The following is a further explanation of the physical machine rack density optimization method based on specific application scenarios.

[0071] For a data center (such as a cloud computing data center), one or more racks (ie, RACKs, which may also be called cabinets) may usually be deployed, and each rack may be deployed with multiple physical machines.

[0072] Among them, the rack can be a rack in the computer room for placing physical machines, and the rack has a rated power consumption limit. The power consumption can be the energy consumed per unit time, and the unit of power consumption can be watts (W), for example, a rack with a rated power consumption of 5000 watts, or a rack with a rated power consumption of 8000 watts, etc.

[0073] A physical machine is a device that generates power in a data center. It is a real device and can also be called a physical device. A physical machine may include but is not limited to servers, switches, routers, etc., and there is no restriction on this.

[0074] A cloud computing data center is a data center that uses cloud computing to implement services. Cloud computing is an Internet-based computing method that allows shared hardware and software resources and information to be provided to computers and other devices on demand. In a cloud computing data center, data is transmitted in the cloud, and the cloud computing data center allocates the required computing power and manages the background of the entire infrastructure. It operates and maintains from both software and hardware aspects. The software level continuously debugs the cloud platform according to the actual network usage, and the hardware level ensures the normal operation and allocation of the computer room environment and network resources.

[0075] Since the rack has a rated power consumption limit, the working process of each physical machine needs to consume energy. Therefore, the number of physical machines deployed on the rack (i.e., the rack density) is limited. Based on this, in order to improve the resource utilization of the data center, it is necessary to increase the rack density of the rack while ensuring the safe operation of the physical machine, but it cannot exceed the upper limit. In the traditional way, the rack density can be configured using empirical values, but this rack density configuration is not accurate. If the rack density is too large, the number of physical machines deployed on the rack is too large, and the rated power consumption of the rack cannot meet the needs of these physical machines. The risk of overpowering the rack increases, and even affects the business activities of these physical machines. If the rack density is too small, the number of physical machines deployed on the rack is too small, and these physical machines cannot fully utilize the resources of the rack, resulting in a waste of resources.

[0076] In view of the above findings, the present application embodiment proposes a physical machine rack density optimization method, which can be applied to any device, see Figure 2 , which is a flow chart of the method, the method comprises:

[0077] Step 201 : acquiring multiple data sets according to historical power consumption data (ie, historical power consumption data of each physical machine), each data set may include historical power consumption data of multiple physical machines.

[0078] In one example, since the power consumption performance of physical machines with different characteristic parameters (such as the model, manufacturer model, business type, etc. of the physical machine) is different, they can be processed separately according to different characteristic parameters, that is, multiple data sets can be obtained according to the characteristic parameters of the physical machine. For multiple physical machines in each data set, the characteristic parameters of different physical machines can be the same or different, that is, the same data set can include the historical power consumption data of multiple physical machines with the same characteristic parameters, or the same data set can include the historical power consumption data of multiple physical machines with different characteristic parameters.

[0079] For example, data set 1 includes historical power consumption data of physical machine 1, historical power consumption data of physical machine 2, and so on. Assuming that a rack deploys a maximum of 28 physical machines, data set 1 includes historical power consumption data of 28 physical machines at most, and there is no limit on the number of physical machines in data set 1. Similarly, other data sets (such as data set 2, data set 3, etc.) can include historical power consumption data of multiple physical machines.

[0080] Among them, for data set 1, the characteristic parameters of different physical machines can be the same, such as the model of physical machine 1 and the model of physical machine 2 can be the same, the manufacturer model of physical machine 1 and the manufacturer model of physical machine 2 can be the same, and the business type of physical machine 1 and the business type of physical machine 2 can be the same.

[0081] Among them, for data set 1, the characteristic parameters of different physical machines may be different, such as the model of physical machine 1 may be different from the model of physical machine 2, the manufacturer model of physical machine 1 may be different from the manufacturer model of physical machine 2, and the business type of physical machine 1 may be different from the business type of physical machine 2.

[0082] Of course, the above method is only an example and is not limited to it, as long as multiple data sets can be obtained based on historical power consumption data. For multiple physical machines in the data set, the characteristic parameters of all physical machines can be different, that is, the characteristic parameters of any two physical machines are different; or, the characteristic parameters of all physical machines can be the same, for example, the characteristic parameters of all physical machines are characteristic parameter A; or, the characteristic parameters of some physical machines are the same, and the characteristic parameters of some physical machines are different, for example, the characteristic parameters of some physical machines are characteristic parameter A, and the characteristic parameters of some physical machines are characteristic parameter B.

[0083] Since the processing method for each data set is the same, in the subsequent embodiments, one data set is taken as an example for description, and this data set is hereinafter referred to as data set 1.

[0084] In one example, since the power consumption of a physical machine varies with the time domain, all historical power consumption data in data set 1 may be historical power consumption data of the same time interval, for example, data set 1 includes historical power consumption data of physical machine 1 in time interval 1, historical power consumption data of physical machine 2 in time interval 1, and so on. All historical power consumption data in data set 1 may be historical power consumption data of different time intervals, for example, data set 1 includes historical power consumption data of physical machine 1 in time interval 1, historical power consumption data of physical machine 2 in time interval 2, and so on.

[0085] Of course, the above method is only an example and is not limited to it, as long as multiple data sets can be obtained based on historical power consumption data. Among them, for the historical power consumption data of multiple physical machines in data set 1, the time intervals of the historical power consumption data of all physical machines can be different, that is, the time intervals of the historical power consumption data of any two physical machines are different; or, the time intervals of the historical power consumption data of all physical machines can be the same, for example, the time intervals of the historical power consumption data of all physical machines are time interval 1; or, the time intervals of the historical power consumption data of some physical machines can be the same, and the time intervals of the historical power consumption data of some physical machines can be different, for example, the time intervals of the historical power consumption data of some physical machines are time interval 1, and the time intervals of the historical power consumption data of some physical machines are time interval 2.

[0086] For each physical machine in data set 1, data set 1 may include historical power consumption data of each physical machine, and the historical power consumption data may include multiple power consumption values, and each power consumption value may be the energy value consumed in a unit time. For example, assuming that the unit time is one minute, and the historical power consumption data is 60 minutes of historical power consumption data, the historical power consumption data may include 60 power consumption values, that is, data set 1 may include 60 power consumption values ​​of each physical machine. For example, assuming that the 60 power consumption values ​​of physical machine 1 are 30 300 watts and 30 500 watts, respectively, and the 60 power consumption values ​​of physical machine 2 are 15 400 watts and 45 600 watts, respectively, and so on. For the 60 power consumption values ​​of other physical machines, they will not be repeated here.

[0087] Step 202 : for each data set, according to the historical power consumption data of multiple physical machines in the data set, determine the probability density distribution of each physical machine, that is, the probability density distribution function (PDF).

[0088] For example, based on the 60 power consumption values ​​of physical machine 1, the probability density distribution of physical machine 1 can be determined; based on the 60 power consumption values ​​of physical machine 2, the probability density distribution of physical machine 2 can be determined, and so on.

[0089] Obviously, since the 60 power consumption values ​​of physical machine 1 are 30 of 300 watts and 30 of 500 watts respectively, the probability that the power consumption value of physical machine 1 is 300 watts is 50%, and the probability that the power consumption value of physical machine 1 is 500 watts is 50%, that is, the probability density distribution of physical machine 1 can include: (300:50%), (500:50%).

[0090] Since the 60 power consumption values ​​of physical machine 2 are 15 at 400 watts and 45 at 600 watts respectively, the probability that the power consumption value of physical machine 2 is 400 watts is 25%, and the probability that the power consumption value of physical machine 2 is 600 watts is 75%, that is, the probability density distribution of physical machine 2 can include: (400:25%), (500:75%).

[0091] By analogy, the process of determining the probability density distribution of other physical machines will not be described in detail here.

[0092] Step 203 : determining the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution of each physical machine, where the power consumption and distribution probability are the distribution probability of the sum of the power consumption of the multiple physical machines.

[0093] In one example, if the specified number is 2, the power consumption and distribution probability of the two physical machines can be determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine. If the specified number is N, the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine can be used to determine the power consumption and distribution probability of N physical machines; where N is an integer greater than 2.

[0094] Determining the power consumption and distribution probability of the two physical machines by using the probability density distribution of one physical machine and the probability density distribution of another physical machine may include: performing a convolution operation on the probability density distribution of one physical machine and the probability density distribution of another physical machine to obtain the power consumption and distribution probability of the two physical machines.

[0095] Determining the power consumption and distribution probability of N physical machines by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine may include: performing a convolution operation on the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine to obtain the power consumption and distribution probability of 2 physical machines.

[0096] For example, the power consumption and distribution probability A1 of two physical machines can be determined by using the probability density distribution of physical machine 1 and the probability density distribution of physical machine 2. The power consumption and distribution probability A1 of two physical machines can be determined by using the probability density distribution of physical machine 3. The power consumption and distribution probability A3 of four physical machines can be determined by using the power consumption and distribution probability A2 of three physical machines and the probability density distribution of physical machine 4. The power consumption and distribution probability A4 of five physical machines can be determined by using the power consumption and distribution probability A3 of four physical machines and the probability density distribution of physical machine 5, and so on.

[0097] In summary, according to the convolution formula of independence and probability distribution: when the two variables X and Y are relatively independent, the probability density distribution of Z = X + Y (i.e., the power consumption and distribution probability of Z) is equal to the convolution of the probability density distribution of X and the probability density distribution of Y. For example, the power consumption and distribution probability of 2 physical machines is equal to the convolution of the probability density distribution of 1 physical machine with the probability density distribution of another physical machine, the power consumption and distribution probability of 3 physical machines is equal to the convolution of the power consumption and distribution probability of 2 physical machines with the probability density distribution of another physical machine, the power consumption and distribution probability of 4 physical machines is equal to the power consumption and distribution probability of 3 physical machines with the probability density distribution of another physical machine, and so on, the power consumption and distribution probability of a specified number of physical machines can be determined, and the specified number is 2, 3, 4, 5..., M, where M is the number of physical machines in data set 1.

[0098] For example, referring to the above embodiment, the probability density distribution of physical machine 1 is (300:50%), (500:50%), and the probability density distribution of physical machine 2 is (400:25%), (500:75%), then the power consumption and distribution probability of the two physical machines are (700:12.5%), (800:37.5%), (900:12.5%), (1000:37.5%). Among them, 700 in the power consumption and distribution probability (700:12.5%) is the sum of 300 in the probability density distribution (300:50%) and 400 in the probability density distribution (400:25%), and 12.5% ​​in the power consumption and distribution probability (700:12.5%) is the product of 50% in the probability density distribution (300:50%) and 25% in the probability density distribution (400:25%), and so on.

[0099] Further, assuming that the probability density distribution of physical machine 3 is (100:40%) and (200:60%), since the power consumption and distribution probability of the two physical machines are (700:12.5%), (800:37.5%), (900:12.5%), and (1000:37.5%), the power consumption and distribution probability of the three physical machines are: (800:5%), (900:7.5%), (900:15%), (1000:22.5%), (1000:5%), (1100:7.5%), (1100:15%), and (1200:22.5%). Then, the above power consumption and distribution probabilities are combined to obtain the power consumption and distribution probabilities of the three physical machines: (800:5%), (900:22.5%), (1000:27.5%), (1100:22.5%), and (1200:22.5%).

[0100] By analogy, the power consumption and distribution probability of four physical machines and the power consumption and distribution probability of five physical machines can be obtained. The determination process of other power consumption and distribution probabilities will not be repeated here.

[0101] Step 204 : determining a cumulative distribution function according to the power consumption and distribution probability of the specified number of physical machines, where the cumulative distribution function includes probability data that the sum of the power consumption of the specified number of physical machines does not exceed the power consumption value.

[0102] In an example, probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value may be determined based on the power consumption and the distribution probability, and a cumulative distribution function may be determined based on the probability data.

[0103] For example, the power consumption and distribution probability of two physical machines can be used to determine the cumulative distribution function (i.e., CDF), which can be referred to as cumulative distribution function 1 later, and the cumulative distribution function 1 includes the probability data that the sum of the power consumption of the two physical machines does not exceed the power consumption value. The power consumption and distribution probability of three physical machines can be used to determine the cumulative distribution function 2, and the cumulative distribution function 2 includes the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value. The power consumption and distribution probability of four physical machines can be used to determine the cumulative distribution function 3, and the cumulative distribution function 3 includes the probability data that the sum of the power consumption of the four physical machines does not exceed the power consumption value, and so on.

[0104] For example, cumulative distribution function 1 may include probability data that the sum of power consumption of two physical machines does not exceed power consumption value A, probability data that the sum of power consumption of two physical machines does not exceed power consumption value B, and so on. Cumulative distribution function 2 may include probability data that the sum of power consumption of three physical machines does not exceed power consumption value C, probability data that the sum of power consumption of three physical machines does not exceed power consumption value D, and so on. For power consumption value A, power consumption value B, power consumption value C, or power consumption value D, it may be a power consumption value configured based on experience, or a rack power consumption threshold, or a power consumption value determined based on power consumption and distribution probability, and there is no restriction on this power consumption value.

[0105] For example, referring to the above embodiment, assuming that the power consumption and distribution probability of two physical machines are (700:12.5%), (800:37.5%), (900:12.5%), and (1000:37.5%), the probability data that the sum of the power consumption of the two physical machines does not exceed the power consumption value of 700 is 12.5%, the probability data that the sum of the power consumption of the two physical machines does not exceed the power consumption value of 800 is 50% (i.e., 12.5%+37.5%), the probability data that the sum of the power consumption of the two physical machines does not exceed the power consumption value of 900 is 62.5%, and the probability data that the sum of the power consumption of the two physical machines does not exceed the power consumption value of 1000 is 100%.

[0106] In summary, the cumulative distribution function includes probability data that the sum of the power consumption of two physical machines does not exceed the power consumption value. These probability data include: (700:12.5%), (800:50%), (900:62.5%), (1000:100%), (700:12.5%) means that the probability data that the sum of the power consumption does not exceed the power consumption value 700 is 12.5%, and so on.

[0107] Of course, other power consumption values ​​may also be set. For example, if the power consumption value is 850, the probability data that the sum of the power consumption of two physical machines does not exceed the power consumption value 850 is 50% (ie, 12.5%+37.5%), and there is no restriction on this.

[0108] For another example, referring to the above embodiment, assuming that the power consumption and distribution probabilities of three physical machines are (800:5%), (900:22.5%), (1000:27.5%), (1100:22.5%), and (1200:22.5%), the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value of 800 is 5%, the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value of 900 is 27.5%, the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value of 1000 is 55%, the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value of 1100 is 77.5%, and the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value of 1200 is 100%. In summary, the cumulative distribution function includes probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value, and these probability data include: (800: 5%), (900: 27.5%), (1000: 55%), (1100: 77.5%), (1200: 100%).

[0109] Of course, other power consumption values ​​may also be set. For example, if the power consumption value is 950, the probability data that the sum of the power consumption of the three physical machines does not exceed the power consumption value of 950 is 27.5%, and there is no limit on the size of this power consumption value.

[0110] Obviously, based on the above method, we can also determine the probability data that the sum of the power consumption of 4 physical machines does not exceed the power consumption value, the probability data that the sum of the power consumption of 5 physical machines does not exceed the power consumption value, and so on, and the cumulative distribution function can include these probability data, and the method for determining these probability data will not be repeated.

[0111] Step 205: Determine the target probability that the sum of the power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function. The rack power consumption threshold may be the rated power consumption of the rack, such as 5000 watts, 8000 watts, etc.

[0112] According to the cumulative distribution function, multiple probability data whose sum of power consumption does not exceed the rack power consumption threshold are determined; different probability data correspond to different numbers of physical machines; a target probability is selected from the multiple probability data; the target probability is less than or equal to the probability threshold, and the target probability is the probability data closest to the probability threshold.

[0113] The probability threshold can be configured based on experience, and there is no restriction on the probability threshold. For example, the probability threshold can be a threshold less than 100%, such as 99%, 95%, etc.

[0114] In one example, based on the cumulative distribution function, the probability data that the sum of the power consumption of 2 physical machines does not exceed the rack power consumption threshold (such as 5000 watts) can be determined. Referring to the above embodiment, since the probability data that the sum of the power consumption of 2 physical machines does not exceed 1000 watts is 100%, the probability data that the sum of the power consumption of 2 physical machines does not exceed 5000 watts is 100%. Based on the cumulative distribution function, the probability data that the sum of the power consumption of 3 physical machines does not exceed 5000 watts can be determined. Since the probability data that the sum of the power consumption of 3 physical machines does not exceed 1200 watts is 100%, the probability data that the sum of the power consumption of 3 physical machines does not exceed 5000 watts is 100%. In addition, based on the cumulative distribution function, the probability data that the sum of the power consumption of 4 physical machines does not exceed 5000 watts can be determined, the probability data that the sum of the power consumption of 5 physical machines does not exceed 5000 watts can be determined, and the probability data that the sum of the power consumption of 6 physical machines does not exceed 5000 watts can be determined, and so on.

[0115] In summary, based on the cumulative distribution function, multiple probability data can be obtained, and different probability data correspond to different numbers of physical machines. Assume that the probability data that the sum of the power consumption of 11 physical machines does not exceed 5000 watts is 100%, the probability data that the sum of the power consumption of 12 physical machines does not exceed 5000 watts is 98%, the probability data that the sum of the power consumption of 13 physical machines does not exceed 5000 watts is 95%, and the probability data that the sum of the power consumption of 14 physical machines does not exceed 5000 watts is 90%, and the probability threshold is 99%. Since 98%, 95%, and 90% are all less than the probability threshold, but among 98%, 95%, and 90%, 98% is the probability data closest to the probability threshold, therefore, 98% is the target probability, that is, the probability data that the sum of the power consumption of 12 physical machines does not exceed 5000 watts.

[0116] Step 206: Determine the physical machine density of the rack according to the number of physical machines corresponding to the target probability.

[0117] For example, the probability data that the sum of the power consumption of 12 physical machines does not exceed 5000 watts is 98%, which is the target probability. Therefore, the number of physical machines corresponding to the target probability can be 12, and the physical machine density of the rack is determined to be 12. Of course, the physical machine density of the rack can also be slightly larger than the number of physical machines corresponding to the target probability, such as 13; or, the physical machine density of the rack can also be slightly smaller than the number of physical machines corresponding to the target probability, such as 11. Take the case where the physical machine density of the rack is the number of physical machines corresponding to the target probability as an example.

[0118] Step 207: deploy physical machines matching the rack density of the physical machines on the rack, that is, physical machines matching the rack density of the physical machines can be deployed on the rack, such as deploying 12 physical machines on the rack.

[0119] In one example, when 12 physical machines are deployed on a rack, the characteristic parameters of the 12 physical machines can match the characteristic parameters of the physical machines in data set 1. For example, when the probability data that the sum of the power consumption of the 12 physical machines in data set 1 (these physical machines are referred to as historical physical machines) does not exceed 5000 watts is 98% the target probability, assuming that the characteristic parameters of the 12 historical physical machines are all characteristic parameters A, 12 physical machines with characteristic parameters A can be deployed on the rack. Assuming that the characteristic parameters of 6 physical machines among the 12 historical physical machines are characteristic parameters A and the characteristic parameters of 6 physical machines are characteristic parameters B, 6 physical machines with characteristic parameters A and 6 physical machines with characteristic parameters B can be deployed on the rack.

[0120] Of course, the above method is just an example, and there is no restriction on the deployment method of this physical machine.

[0121] Optionally, in an example, assuming that the physical machines in data set 1 are the same as the physical machines in data set 2, and the time interval of the historical power consumption data of the physical machines in data set 1 is different from the time interval of the historical power consumption data of the physical machines in data set 2, after determining the physical machine rack density A based on data set 1 and determining the physical machine rack density B based on data set 2, the physical machine rack density A and the physical machine rack density B can be compared. If the physical machine rack density A is less than the physical machine rack density B, a physical machine matching the physical machine rack density A is deployed on the rack; if the physical machine rack density B is less than the physical machine rack density A, a physical machine matching the physical machine rack density B is deployed on the rack.

[0122] Optionally, in an example, see Figure 3A As shown, assuming there are multiple racks ( Figure 3A In the example of 2 racks, if the number of physical machines deployed on rack 1 is less than the physical machine rack density of rack 1, and the number of physical machines deployed on rack 2 is less than the physical machine rack density of rack 2, that is, both rack 1 and rack 2 can deploy new physical machines, then for the physical machine 8 to be racked, whether to deploy the physical machine 8 to rack 1 or to rack 2, the following implementation method is also proposed in the embodiment of the present application:

[0123] See also Figure 3BAs shown, the filter performs the following operations: Select a list of racks to be put on the rack. Determine whether the positions of each rack meet the needs. When there are no positions on the rack, the physical machine 8 cannot be carried. When there are positions on the rack, the physical machine 8 can be carried. Determine whether the power consumption of each rack meets the needs. If the sum of the power consumption of the physical machine 8 is greater than the rack power consumption threshold, the physical machine 8 cannot be carried. If the sum of the power consumption of the physical machine 8 is not greater than the rack power consumption threshold, the physical machine 8 can be carried. Mixed deployment strategy, that is, deploy physical machines with different characteristic parameters in different racks as much as possible. New rack strategy, that is, when all racks cannot carry physical machine 8, deploy a new rack and use the new rack to carry physical machine 8.

[0124] See also Figure 3B As shown, the weighter performs the following operations: in the list of racks that meet the conditions, a rack with the highest score is selected for listing based on the weights of the machine type scattering strategy (i.e., deploying physical machines with different characteristic parameters in different racks as much as possible), the port fragmentation rate (i.e., using as many ports as possible), and the power consumption fragmentation rate (i.e., using as much power consumption of each rack as possible to avoid power consumption fragmentation).

[0125] In one example, the above execution order is just an example given for the convenience of description. In practical applications, the execution order between the steps can also be changed, and there is no limitation on this execution order. Moreover, in other embodiments, the steps of the corresponding method are not necessarily executed in the order shown and described in this specification, and the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; multiple steps described in this specification may also be combined into a single step for description in other embodiments.

[0126] Based on the above technical solutions, in the embodiments of the present application, the rack density can be accurately determined, and under the premise of ensuring the safe operation of the physical machine, the power efficiency can be improved, the resource utilization rate can be improved, the PUE can be reduced, the rack density of the physical machine can be increased, and the power consumption utilization rate of the rack can be improved. The number of physical machines on the rack can be dynamically adjusted, and the racking strategy is dynamically adjusted over time. Refined racking can be performed for physical machines with different characteristic parameters. For different racking combinations, the safety risk value under the combination can be clearly given. Based on a scientific packing model, an over-power consumption probability indicator is used to measure the impact of the new racking model on business stability. Through the accumulation of a large amount of power consumption data, a racking density optimization model based on over-power consumption probability calculation is adopted to provide more scientific and refined guidance for the racking of physical machines, which can achieve a balance between over-power risks and benefits.

[0127] Based on the same application concept as the above method, the embodiment of the present application also provides a physical machine rack density optimization device, such as Figure 4 As shown, it is a structural diagram of the device, and the device may include:

[0128] An acquisition module 41 is used to acquire historical power consumption data of multiple physical machines;

[0129] A determination module 42 is configured to determine a cumulative distribution function based on the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value; and determine a target probability that the sum of the power consumption does not exceed a rack power consumption threshold based on the cumulative distribution function;

[0130] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0131] When determining the cumulative distribution function based on the historical power consumption data of the multiple physical machines, the determination module 42 is specifically used to: determine the probability density distribution of the physical machines based on the historical power consumption data of the multiple physical machines; determine the power consumption and distribution probability of a specified number of physical machines based on the probability density distribution; and determine the cumulative distribution function based on the power consumption and distribution probability.

[0132] The determination module 42 is specifically used to determine the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution: if the specified number is 2, then the power consumption and distribution probability of the two physical machines are determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; if the specified number is N, then the power consumption and distribution probability of N physical machines are determined by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine; N is an integer greater than 2.

[0133] Based on the same application concept as the above method, the embodiment of the present application further provides a physical machine rack density optimization device, including: a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores a plurality of computer instructions, and the processor performs the following processing when executing the computer instructions:

[0134] Obtain historical power consumption data of multiple physical machines;

[0135] Determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value;

[0136] Determine, according to the cumulative distribution function, a target probability that the sum of power consumption does not exceed a rack power consumption threshold;

[0137] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0138] The present application also provides a machine-readable storage medium, on which a number of computer instructions are stored; when the computer instructions are executed, the following processing is performed:

[0139] Obtain historical power consumption data of multiple physical machines;

[0140] Determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value;

[0141] Determine, according to the cumulative distribution function, a target probability that the sum of power consumption does not exceed a rack power consumption threshold;

[0142] The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

[0143] See also Figure 5 As shown, it is a structural diagram of a physical machine rack density optimization device, and the physical machine rack density optimization device 50 may include: a processor 51, a network interface 52, a bus 53, and a memory 54. The memory 54 may be any electronic, magnetic, optical or other physical storage device, and may contain or store information, such as executable instructions, data, etc. For example, the memory 54 may be: RAM (Radom Access Memory, random access memory), volatile memory, non-volatile memory, flash memory, storage drive (such as hard disk drive), solid state hard disk, any type of storage disk (such as optical disk, DVD, etc.).

[0144] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which may be in the form of a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email transceiver, a game console, a tablet computer, a wearable device or a combination of any of these devices.

[0145] For the convenience of description, the above device is described in terms of functions and is described separately in various units. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0146] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the embodiments of the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0147] The present application is described with reference to the 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 process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, 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 device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0148] Moreover, these computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0149] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0150] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for optimizing the density of physical machines, characterized in that: The method comprises: Obtain historical power consumption data of multiple physical machines; Determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value; Determine the target probability that the sum of power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function; wherein, determine multiple probability data that the sum of power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function, and different probability data correspond to different numbers of physical machines; select the target probability from the multiple probability data; the target probability is less than or equal to the probability threshold, and the target probability is the probability data closest to the probability threshold; The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

2. The method according to claim 1, characterized in that The determining of the cumulative distribution function according to the historical power consumption data of the multiple physical machines includes: Determine the probability density distribution of the physical machines according to the historical power consumption data of the multiple physical machines; Determine the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution; The cumulative distribution function is determined according to the power consumption and the distribution probability.

3. The method according to claim 2, characterized in that Determining the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution includes: If the specified number is 2, the power consumption and distribution probability of the two physical machines are determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; If the specified number is N, the power consumption and distribution probability of N physical machines are determined by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine; N is an integer greater than 2.

4. The method according to claim 2, characterized in that: Determining the cumulative distribution function according to the power consumption and the distribution probability includes: The probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value is determined according to the power consumption and the distribution probability, and the cumulative distribution function is determined according to the probability data.

5. The method according to claim 1, characterized in that After determining the physical machine density of the rack according to the number of physical machines corresponding to the target probability, the method includes: Physical machines matching the rack density of the physical machines are deployed on the racks.

6. The method according to claim 1, characterized in that The obtaining of historical power consumption data of multiple physical machines includes: Obtain historical power consumption data of multiple physical machines according to characteristic parameters of the physical machines; Among them, the characteristic parameters of different physical machines are the same or different.

7. A method for optimizing the density of physical machines, characterized in that: The data center deploys one or more racks, each rack deploys multiple physical machines, and the method includes: Obtain historical power consumption data of multiple physical machines; Determine the probability density distribution of the physical machines according to the historical power consumption data of the multiple physical machines; Determine the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution; Determine a cumulative distribution function according to the power consumption and the distribution probability, wherein the cumulative distribution function includes probability data that the sum of the power consumptions of a specified number of physical machines does not exceed the power consumption value; Determine the target probability that the sum of power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function; wherein, determine multiple probability data that the sum of power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function, and different probability data correspond to different numbers of physical machines; select the target probability from the multiple probability data; the target probability is less than or equal to the probability threshold, and the target probability is the probability data closest to the probability threshold; The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

8. The method according to claim 7, characterized in that Determining the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution includes: If the specified number is 2, the power consumption and distribution probability of the two physical machines are determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; If the specified number is N, the power consumption and distribution probability of N physical machines are determined by using the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine; N is an integer greater than 2.

9. A physical machine rack density optimization device, characterized in that: The device comprises: An acquisition module is used to obtain historical power consumption data of multiple physical machines; A determination module, used to determine a cumulative distribution function based on the historical power consumption data of the multiple physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value; determine a target probability that the sum of the power consumption does not exceed a rack power consumption threshold according to the cumulative distribution function; wherein the determination module determines the target probability that the sum of the power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function, specifically used to: determine multiple probability data that the sum of the power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function, different probability data corresponding to different numbers of physical machines; select a target probability from the multiple probability data; the target probability is less than or equal to the probability threshold, and the target probability is the probability data closest to the probability threshold; The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.

10. The device according to claim 9, characterized in that The determining module is specifically used to determine the cumulative distribution function according to the historical power consumption data of the multiple physical machines: Determine the probability density distribution of the physical machines according to the historical power consumption data of the multiple physical machines; Determine the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution; The cumulative distribution function is determined according to the power consumption and the distribution probability.

11. The device according to claim 10, characterized in that The determining module is specifically used to determine the power consumption and distribution probability of a specified number of physical machines according to the probability density distribution: If the specified number is 2, the power consumption and distribution probability of the two physical machines are determined by using the probability density distribution of one physical machine and the probability density distribution of another physical machine; if the specified number is N, the power consumption and distribution probability of N-1 physical machines and the probability density distribution of one physical machine are used to determine the power consumption and distribution probability of N physical machines; N is an integer greater than 2.

12. A physical machine rack density optimization device, characterized in that: include: A processor and a machine-readable storage medium, wherein the machine-readable storage medium stores a plurality of computer instructions, and when the processor executes the computer instructions, the processor performs the following processing: Obtain historical power consumption data of multiple physical machines; Determine a cumulative distribution function according to the historical power consumption data of the plurality of physical machines, the cumulative distribution function including probability data that the sum of the power consumption of a specified number of physical machines does not exceed the power consumption value; Determine the target probability that the sum of power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function; wherein, determine multiple probability data that the sum of power consumption does not exceed the rack power consumption threshold according to the cumulative distribution function, and different probability data correspond to different numbers of physical machines; select the target probability from the multiple probability data; the target probability is less than or equal to the probability threshold, and the target probability is the probability data closest to the probability threshold; The physical machine density of the rack is determined according to the number of physical machines corresponding to the target probability.