Data center calculation and cooling cooperative scheduling method and system based on optimal working temperature

Abstract

The invention discloses a data center calculation and cooling cooperative scheduling method based on an optimal working temperature, and the method comprises the following steps: obtaining the operation data of an IT system and a cooling system of a data center, and building a server power consumption model and a cooling system power consumption model; establishing a data center energy consumption model based on the server power consumption model and the cooling system power consumption model; on the basis of the data center energy consumption model, with the purpose of minimum overall energy consumption of the data center, calculating different CPU usage rates of the server and the optimal working temperature of the server under outdoor real-time temperature and humidity, and forming a temperature control guide list; and formulating a collaborative scheduling strategy with the target of giving the optimal temperature control of the active server. According to the data center calculation and cooling cooperative scheduling method based on the optimal working temperature, the overall energy efficiency optimization of the data center is converted into the optimal temperature control of the given active server, and a feasible implementation path is provided for the optimal overall energy efficiency of the IT system and the cooling system of the data center.

Description

Technical Field

[0001] The present invention relates to the technical field of data center energy conservation, and in particular to a data center computing and cooling collaborative scheduling method and system based on optimal operating temperature. Background Art

[0002] With the continuous development of big data and artificial intelligence technology, the energy consumption of data centers is also increasing, and IT systems and cooling systems are the two largest components. At present, IT systems often adopt energy-saving measures such as load merging, that is, integrating virtual machines on each server through virtualization technology to reduce the number of active servers, but this can easily lead to the emergence of local hotspots, resulting in an increase in the energy consumption of servers at hotspot locations, resulting in a decrease in the energy consumption of each server.

[0003] Cooling systems often reduce energy consumption by increasing the water supply / air inlet temperature, but this can easily lead to an increase in server energy consumption, resulting in a trade-off between server and cooling system energy consumption.

[0004] In order to reduce the overall energy consumption of the IT system and cooling system in the data center, an effective way is to use a model prediction method, with the goal of optimizing overall energy efficiency, to coordinate and optimize the deployment of virtual machines in the IT system and the control of the cooling system in real time. However, due to the great differences in the scheduling accuracy, speed and frequency of virtual machines and cooling capacity, the above method is difficult to apply to actual projects, and therefore cannot truly achieve the optimal overall energy efficiency of the data center. For this reason, it is urgent to further explore the coordinated scheduling method of computing power and cooling capacity in the data center. Summary of the invention

[0005] In order to overcome the defects in the prior art, the present invention provides a data center computing and cooling collaborative scheduling method and system based on optimal operating temperature. The data center computing and cooling collaborative scheduling method based on optimal operating temperature converts the overall energy efficiency optimization of the data center into the optimal temperature control of a given active server, providing a practical implementation path for the overall energy efficiency optimization of the data center IT system and cooling system.

[0006] To achieve the above object, the technical solution adopted by the present invention is: a data center computing and cooling collaborative scheduling method based on optimal operating temperature, comprising the following steps:

[0007] Obtain the operating data of the IT system and cooling system of the data center, and establish the server power consumption model and cooling system power consumption model;

[0008] Establishing a data center energy consumption model based on the server power consumption model and the cooling system power consumption model;

[0009] Based on the data center energy consumption model, with the goal of minimizing the overall energy consumption of the data center, the optimal operating temperature of the server under different CPU usage rates and outdoor real-time temperature and humidity is calculated, and a temperature control guidance list is formed;

[0010] Develop a collaborative scheduling strategy that targets optimal temperature control for a given set of active servers.

[0011] Through the above solution, the overall energy efficiency optimization of the data center is converted into the optimal temperature control of a given active server, providing a practical implementation path for the overall energy efficiency optimization of the data center IT system and cooling system.

[0012] Furthermore, the operation data of the IT system includes: real-time utilization rate of server CPU resources, real-time power consumption of the server, and real-time temperature of the CPU;

[0013] The cooling system operation data includes: the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, the real-time frequency of the refrigerator compressor, the real-time power consumption of the cooling system and the real-time outdoor temperature and humidity.

[0014] Furthermore, establishing a server power consumption model includes:

[0015] A physical information neural network is used to establish a server power consumption model based on the real-time utilization rate of CPU resources and the real-time temperature of CPU:

[0016] P server (t) = f sever (U server (t), T server,CPU (t));

[0017] Where P server (t) represents the real-time power consumption of the server, U server (t) represents the real-time utilization rate of the server CPU resources, T server,CPU (t) indicates the real-time temperature of the CPU, f sever Indicates the mapping relationship between the real-time usage of the server CPU resources and the temperature and power consumption.

[0018] Furthermore, establishing a cooling system power consumption model includes:

[0019] A physical information neural network is used to establish a cooling system power consumption model based on the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, and the real-time frequency of the chiller compressor;

[0020] P cooling,sys (t) = f cooling,sys，P (TerFan FP (t), ChiPumpFP (t), CdPump FP (t), Chiller FP (t), CT FP (t));

[0021] Where P cooling，sys (t) represents the real-time power consumption of the cooling system, f cooling，sys,P TerFan represents the mapping relationship between the power consumption of the cooling system and the real-time frequency of each terminal fan, chilled water pump, cooling water pump, cooling tower fan, and chiller compressor in the cooling system. FP (t) represents the matrix composed of the terminal fan frequency, ChiPump FP (t) represents the matrix composed of the chilled water pump frequency, CdPump FP (t) represents the matrix composed of the cooling water pump frequency, Chiller FP (t) represents the matrix composed of the compressor frequency of the refrigerator, CT FP (t) shows the matrix composed of the cooling tower fan frequencies.

[0022] Furthermore, establishing a data center energy consumption model includes:

[0023] Establish a data center energy consumption model based on the server power consumption model and cooling system power consumption model:

[0024] EC total =∫(∑P server (t)+P cooling,sys (t))dt

[0025] In the formula, EC total Indicates the total energy consumption of a data center in a certain period of time.

[0026] Furthermore, it also includes establishing a cooling system cooling capacity model:

[0027] A physical information neural network is used to establish a cooling system cooling capacity model based on the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, the real-time frequency of the chiller compressor, and the real-time outdoor temperature and humidity;

[0028] Q cooling,sys (t) = f cooling,sys，Q (TerFan FP (t), ChiPump FP (t), CdPump FP (t), Chiller FP (t), CT FP (t), T wb (t), RH(t));

[0029] In the formula, Q cooling,sys (t) represents the real-time cooling capacity of the cooling system, f cooling,sys，Q It represents the mapping relationship between the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, the real-time frequency of the chiller compressor, the real-time outdoor temperature and humidity and the cooling capacity of the system. wb (t) represents the real-time outdoor temperature, and RH(t) represents the real-time outdoor relative humidity.

[0030] Furthermore, in order to minimize the overall energy consumption of the data center, the optimal operating temperature of the server under different CPU usage rates and outdoor real-time temperature and humidity is calculated, including:

[0031] With the goal of minimizing the overall energy consumption of the data center, a genetic algorithm is used to optimize the real-time utilization rate of CPU resources and the operating conditions under outdoor real-time temperature and humidity based on the data center energy consumption model:

[0032] Optimization goal: min EC total =∫(∑P server (t)+P cooling,sys (t))dt;

[0033] Optimization constraint: T server,CPU (t)≤T server,CPU,max ;

[0034] Optimization variables: {TerFan FP (t), ChiPump FP (t), CdPump FP (t), Chiller FP (t), CT FP (t)}

[0035] Where, min EC total represents the minimum energy consumption of the data center, T server,CPU (t) represents the matrix composed of the real-time CPU temperatures of each server in the data center, T server,CPU,max Indicates the upper limit of the server CPU operating temperature.

[0036] Through optimization calculation, the corresponding min EC under various working conditions is obtained total , the min EC total The corresponding T server,CPU (t) is the optimal operating temperature of the server. The optimal operating temperature of the server corresponding to each working condition optimization result is sorted out to form a temperature control guidance list.

[0037] Furthermore, the collaborative scheduling strategy for optimal temperature control of a given active server is formulated as follows:

[0038] According to the scheduling characteristics of IT systems, formulate a scheduling strategy targeting a given active server:

[0039] Number the data center servers;

[0040] When the IT system receives a new VM deployment request, it obtains the CPU resources required by the VM.

[0041] According to the server number, the remaining CPU resources of the active servers in the data center are sequentially traversed. When the remaining CPU resources of the active server are available for deploying a new virtual machine, the new virtual machine is deployed on the server. If the remaining CPU resources of the active server are not enough to deploy a new virtual machine, a dormant server is made active to deploy the new virtual machine.

[0042] At each preset time interval, the CPU resource requirements of the virtual machines on the active servers are summed up and compared with the rated CPU resources of a single server to determine the minimum number of active servers required;

[0043]

[0044] In the formula, ∑RS VMi (t) represents the sum of CPU resource requirements of the virtual machines hosted in real time on each server, C server Indicates the rated CPU resource capacity of a single server, Num server，active (t) represents the number of active servers required at this moment;

[0045] According to the required number of active servers, the target server to be migrated is determined by server number, and the virtual machines on the active servers in the data center are migrated to the target server with the goal of balancing the CPU resource utilization of the target server.

[0046] During the operation of the data center, virtual machines are automatically released after completing their tasks. If all virtual machines on an active server are released, the server is considered an idle server and goes into hibernation.

[0047] Furthermore, according to the scheduling characteristics of the cooling system, a collaborative scheduling strategy is formulated with the goal of optimal temperature control for a given active server, including:

[0048] Calculate the real-time average CPU usage of active servers, and determine the operating conditions of the data center according to the temperature control guidance list in combination with the outdoor temperature and humidity;

[0049] Determine the optimal operating temperature of the server corresponding to the operating condition from the temperature control guidance list, and use it as the control threshold of the cooling system;

[0050] The real-time maximum temperature of the CPU in the active server is used as the control object, and the feedback control method is used to control it at the optimal operating temperature of the corresponding server according to the set control threshold. In summary of the dynamic operation process, the real-time temperature of each active server may be different, so the one with the highest temperature needs to be used as the control basis.

[0051] A data center computing and cooling collaborative scheduling system, comprising:

[0052] Information collection module, used to obtain the operating data of the IT system and cooling system of the data center;

[0053] A first processing module is used to establish a server power consumption model and a cooling system power consumption model according to the operation data of the IT system and the cooling system, and to establish a data center energy consumption model based on the server power consumption model and the cooling system power consumption model;

[0054] The energy consumption analysis module calculates the optimal operating temperature of the server under different CPU usage rates and outdoor real-time temperature and humidity based on the data center energy consumption model and aims to minimize the overall energy consumption of the data center, and forms a temperature control guidance list;

[0055] The strategy generation module formulates the collaborative scheduling strategy with the goal of optimal temperature control for a given active server;

[0056] The scheduling control module adjusts the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, and the real-time frequency of the refrigeration compressor according to the collaborative scheduling strategy, so as to adjust the real-time maximum temperature of the CPU in the server to the optimal operating temperature of the server.

[0057] By means of the above technical solution, the beneficial effects of the present invention are as follows:

[0058] This application will combine the energy consumption correlation between the IT system and the cooling system in the data center, and convert the overall energy efficiency optimization of the data center into the optimal temperature control of a given active server, providing a practical implementation path for the overall energy efficiency optimization of the IT system and the cooling system in the data center.

[0059] In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 This is an implementation flow chart of a data center computing and cooling collaborative scheduling method based on optimal operating temperature;

[0062] Figure 2 This is the core idea of ​​this patent: to transform the overall energy efficiency optimization of the data center into a mathematical representation of the optimal temperature control for a given active server;

[0063] Figure 3 This is a structural module diagram of a data center to which this patent applies. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and to distinguish similar objects. There is no order of precedence between the two, and they cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0066] Example: Combination Figure 1-3 As shown, embodiment 1 discloses a data center computing and cooling collaborative scheduling method based on optimal operating temperature, comprising the following steps:

[0067] The operation data of the IT system and cooling system of the data center are obtained, wherein the operation data of the IT system includes: the real-time utilization rate of the server CPU resources, the real-time power consumption of the server, and the real-time temperature of the CPU; the operation data of the cooling system includes: the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each cooling tower fan, the real-time frequency of each refrigerator compressor, the real-time power consumption of the cooling system, and the real-time outdoor temperature and humidity.

[0068] A physical information neural network is used to establish a server power consumption model based on the real-time CPU resource utilization and CPU real-time temperature:

[0069] P server (t) = f sever (U server (t),T server,CPU (t));

[0070] Where P server (t) represents the real-time power consumption of the server, U server (t) represents the real-time utilization rate of the server’s CPU resources, T server,CPU (t) indicates the real-time temperature of the CPU, f sever Indicates the mapping relationship between the real-time usage rate of the server's CPU resources and the temperature and power consumption.

[0071] A physical information neural network is used to establish a cooling system power consumption model based on the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each cooling tower fan, and the real-time frequency of each chiller compressor;

[0072] P cooling,sys (t) = f cooling,sys,P (TerFan FP (t), ChiPump FP (t), CdPump F x(t), Chiller FP (t), CT FP (t));

[0073] Where P cooling，sys (t) represents the real-time power consumption of the cooling system, f cooling，sys,P It indicates the mapping relationship between the real-time frequency of each terminal fan, each chilled water pump, each cooling water pump, each chiller compressor, each cooling tower fan and the power consumption of the cooling system. FP (t) represents the matrix composed of the terminal fan frequency, ChiPump FP (t) represents the matrix composed of the chilled water pump frequency, CdPump FP (t) represents the matrix composed of the cooling water pump frequency, Chiller FP (t) represents the matrix composed of the compressor frequency of the refrigerator, CT FP (t) represents the matrix composed of the cooling tower fan frequencies.

[0074] Establish a data center energy consumption model based on the server power consumption model and cooling system power consumption model:

[0075] EC total =∫(∑P server (t)+P cooling,sys (t))dt

[0076] In the formula, EC total Indicates the total energy consumption of a data center in a certain period of time.

[0077] A physical information neural network is used to establish a cooling system cooling capacity model based on the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each refrigerator compressor, the real-time frequency of each cooling tower fan, and the real-time outdoor temperature and humidity;

[0078] Q cooling,sys (t) = f cooling,sys,Q (TerFan FP (t), ChiPump FP (t), CdPump FP (t), Chiller FP (t), CT FP (t), T wb (t), RH(t));

[0079] In the formula, Q cooling,sys (t) represents the real-time cooling capacity of the cooling system, f cooling,sys,Q It represents the mapping relationship between the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each chiller compressor, the real-time frequency of each cooling tower fan, the real-time outdoor temperature and humidity and the cooling capacity of the system. wb (t) represents the real-time outdoor temperature, and RH(t) represents the real-time outdoor relative humidity.

[0080] With the goal of minimizing the overall energy consumption of the data center, a genetic algorithm is used to optimize the real-time utilization rate of CPU resources and the operating conditions under outdoor real-time temperature and humidity based on the data center energy consumption model:

[0081] Optimization goal: min EC total =∫(∑P server (t)+P cooling,sys (t))dt;

[0082] Optimization constraint: T server,CPU (t)≤T server,CPU,max ;

[0083] Optimization variables: {TerFan FP (t), ChiPump FP (t), CdPump FP (t), Chiller FP (t), CT FP (t)}

[0084] Where, min EC total represents the minimum energy consumption of the data center, Tserver,CPU (t) represents the matrix composed of the real-time CPU temperatures of each server in the data center, T server,CPU,max Indicates the upper limit of the server CPU operating temperature.

[0085] By min EC total =∫(∑P server (t)+P cooling,sys (t))dt

[0086] min EC total

[0087] =∫(∑(f sever (U server (t), T server，CPU (t)))+(f cooling,sys,P (TerFan FP (t),ChiPumP FP (t), CdPumP FP (t), Chiller FP (t), CT FP (t))))dt

[0088] Among them, the real-time utilization rate of the server CPU resources U server (t) is the working condition information. Through optimization calculation, the minimum energy consumption min EC of the data center under each working condition is obtained. total At this time, the minimum energy consumption of the data center is min EC total The corresponding T server,CPU (t) is the optimal operating temperature of the server.

[0089] The optimal operating temperature of the server corresponding to each working condition optimization result is statistically organized to form a temperature control guidance list.

[0090] For example:

[0091] Working conditions: real-time utilization rate of server CPU resources U server (t) = 5%, outdoor real-time temperature T wb (t) = 20°C, relative humidity RH(t) = 60%, the optimal operating temperature of the server is 35°C after the optimization calculation, which is expressed in the temperature control guidance list as:

[0092] EC total,min (U server =5%,T wb =20℃,RH(t)=60%)→T server,CPU,optimal (U server (t)=5%,T wb (t)=20℃, RH(t)=60%)=35℃;

[0093] In the formula, (U server =5%,T wb =20℃, RH(t)=60%) is the working condition, T server,CPU,optimal (U server (t)=5%,T wb (t) = 20°C, RH(t) = 60%) indicates the optimal operating temperature of the server under this working condition.

[0094] Optionally, you can obtain typical annual meteorological data for the data center location from the EnergyPlus official website, and filter out the outdoor hourly temperature and humidity to obtain the outdoor real-time temperature T corresponding to each working condition. wb (t) and outdoor real-time relative humidity RH(t).

[0095] According to the above temperature control guidance list, a collaborative scheduling strategy is formulated with the goal of optimal temperature control for a given active server. The steps include:

[0096] Get the real-time average CPU usage of each active server, and combine it with the real-time outdoor temperature and humidity to determine the operating condition of the data center according to the temperature control guidance list. server =XX,T wb =YY, RH(t)=ZZ, where the real-time CPU usage average of the active server replaces the real-time usage U of the server CPU resources. server (t), the real-time outdoor temperature and humidity are obtained through the EnergyPlus official website;

[0097] Select the operating condition U from the temperature control guidance list server =XX,T wb =YY, RH(t) = ZZ corresponding to the optimal server operating temperature T server,CPU,optimal (U server =XX,T wb =YY, RH(t)=ZZ), and taking the optimal operating temperature of the server as the control threshold of the cooling system;

[0098] According to the set control threshold, the server's optimal operating temperature T server,CPU,optimal (U server =XX,T wb =YY, RH(t)=ZZ), and the real-time maximum temperature of the CPU in the active server is adjusted to the threshold value by feedback control.

[0099] Among them, the present application also formulates a scheduling strategy for the IT system targeting a given active server, including the following steps:

[0100] Number the data center servers;

[0101] When the IT system receives a new VM deployment request, it obtains the CPU resources required by the VM.

[0102] According to the server number, the remaining CPU resources of the active servers in the data center are sequentially traversed. When the remaining CPU resources of the active server are available for deploying a new virtual machine, the new virtual machine is deployed on the server. If the remaining CPU resources of the active server are not enough to deploy a new virtual machine, a dormant server is made active to deploy the new virtual machine.

[0103] At each preset time interval, the CPU resource requirements of the virtual machines on the active servers are summed up and compared with the rated CPU resources of a single server to determine the minimum number of active servers required;

[0104]

[0105] In the formula, ∑RS VMi (t) represents the sum of CPU resource requirements of the virtual machines hosted in real time on each server, C server Indicates the rated CPU resource capacity of a single server, Num server,active (t) represents the number of active servers required at this moment;

[0106] According to the required number of active servers, the target server to be migrated is determined by server number, and the virtual machines on the active servers in the data center are migrated to the target server with the goal of balancing the CPU resource utilization of the target server.

[0107] During the operation of the data center, virtual machines are automatically released after completing their tasks. If all virtual machines on an active server are released, the server is considered an idle server and goes into hibernation.

[0108] Embodiment 2 discloses a data center computing and cooling collaborative scheduling system, including:

[0109] The information collection module is used to obtain the operating data of the IT system and cooling system of the data center. The operating data of the IT system includes the real-time utilization rate of the server CPU resources, the real-time power consumption of the server and the real-time temperature of the CPU. The operating data of the cooling system includes the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each cooling tower fan, the real-time frequency of each refrigeration machine compressor, the real-time power consumption of the cooling system and the real-time outdoor temperature and humidity.

[0110] The first processing module is used to establish a server power consumption model according to the real-time utilization rate of the server CPU resources, the real-time power consumption of the server and the real-time temperature of the CPU, establish a cooling system power consumption model according to the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each cooling tower fan, and the real-time frequency of each refrigerator compressor, establish a cooling system cooling capacity model according to the real-time frequency of each terminal fan, the real-time frequency of each chilled water pump, the real-time frequency of each cooling water pump, the real-time frequency of each cooling tower fan, the real-time frequency of each refrigerator compressor and the real-time outdoor temperature and humidity, and establish a data center energy consumption model based on the server power consumption model and the cooling system power consumption model.

[0111] The energy consumption analysis module, based on the data center energy consumption model, aims to minimize the overall energy consumption of the data center, calculates the optimal operating temperature of the server under different CPU usage rates and outdoor real-time temperature and humidity, and forms a temperature control guidance list. Among them, the overall energy consumption of the data center is the sum of the server energy consumption and the cooling system energy consumption. The real-time usage rate of the server CPU resources and the real-time outdoor temperature and humidity are used as working conditions. According to the working conditions, the corresponding CPU real-time temperature in the server power consumption model when the overall energy consumption of the data center is the lowest is traversed and filtered out. The CPU real-time temperature at this time is the optimal operating temperature of the server. The overall minimum energy consumption of all data centers and the corresponding optimal server operating temperatures are statistically organized into a temperature control guidance list.

[0112] The strategy generation module is used to formulate a collaborative scheduling strategy with the goal of optimal temperature control for a given active server. Specifically, the real-time average CPU usage of the active server is obtained, and combined with the real-time outdoor temperature and humidity, the operating condition of the data center is determined according to the temperature control guidance list, and the optimal operating temperature of the server corresponding to the operating condition is determined from the temperature control guidance list, and the optimal operating temperature of the server is used as the control threshold of the cooling system.

[0113] The real-time CPU usage average of the active servers replaces the real-time usage of the server CPU resources, and the real-time outdoor temperature and humidity are obtained through the EnergyPlus official website.

[0114] The scheduling control module uses feedback control to adjust the real-time maximum temperature of the CPU in the active server to the threshold according to the set control threshold, i.e., the optimal operating temperature of the server, and the information in the temperature control guidance list. In summary of the dynamic operation process, the real-time temperature of each active server may be different, so the one with the highest temperature needs to be used as the control basis.

[0115] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A data center computing and cooling collaborative scheduling method based on optimal operating temperature, characterized in that: The following steps are involved: Obtain the operating data of the IT system and cooling system of the data center, and establish the server power consumption model and cooling system power consumption model; Establishing a data center energy consumption model based on the server power consumption model and the cooling system power consumption model; Based on the data center energy consumption model, with the goal of minimizing the overall energy consumption of the data center, the optimal operating temperature of the server under different CPU usage rates and outdoor real-time temperature and humidity is calculated to form a temperature control guidance list; Develop a collaborative scheduling strategy that targets optimal temperature control for a given set of active servers.

2. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 1, characterized in that: The operation data of the IT system includes: real-time utilization rate of server CPU resources, real-time power consumption of the server, and real-time temperature of the CPU; The cooling system operation data includes: the real-time frequency of the terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, the real-time frequency of the refrigerator compressor, the real-time power consumption of the cooling system and the real-time outdoor temperature and humidity.

3. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 2, characterized in that: Building a server power consumption model includes: Establish a server power consumption model based on the real-time CPU resource usage and CPU real-time temperature: P server (t)=f sever (U server (t),T server,CPU (t)); Where P server (t) represents the real-time power consumption of the server, U server (t) represents the real-time utilization rate of the server CPU resources, T server,CPU (t) indicates the real-time temperature of the CPU, f sever Indicates the mapping relationship between the real-time usage of the server CPU resources and the temperature and power consumption.

4. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 3, characterized in that: Modeling the cooling system power consumption includes: Establish a cooling system power consumption model based on the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, and the real-time frequency of the chiller compressor; P cooling,sys (t)=f cooling,sys,P (TerFan FP (t),Chipump FP (t),CdPump FP (t),Chiller FP (t),CT FP (t)); Where P cooling,sys (t) represents the real-time power consumption of the cooling system, f cooling,sys,P TerFan represents the mapping relationship between the power consumption of the cooling system and the real-time frequency of each terminal fan, chilled water pump, cooling water pump, cooling tower fan, and chiller compressor in the cooling system. FP (t) represents the matrix composed of the terminal fan frequency, ChiPump FP (t) represents the matrix composed of the chilled water pump frequency, CdPump FP (t) represents the matrix composed of the cooling water pump frequency, Chiller FP (t) represents the matrix composed of the compressor frequency of the refrigerator, CT FP (t) represents the matrix composed of the cooling tower fan frequencies.

5. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 4, characterized in that: Building a data center energy consumption model includes: Establish a data center energy consumption model based on the server power consumption model and cooling system power consumption model: In the formula, EC total Indicates the total energy consumption of a data center in a certain period of time.

6. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 5, characterized in that: It also includes modeling the cooling system's cooling capacity: A cooling system cooling capacity model is established based on the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, the real-time frequency of the chiller compressor, and the real-time outdoor temperature and humidity; Q cooling,sys (t)=f cooling,sys,Q (TerFan FP (t),ChiPump FP (t),CdPump FP (t),Chiller FP (t),CT FP (t),T wb (t),RH(t)); In the formula, Q cooling,sys (t) represents the real-time cooling capacity of the cooling system, f cooling,sys,Q It represents the mapping relationship between the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, the real-time frequency of the chiller compressor, the real-time outdoor temperature and humidity and the cooling capacity of the system. wb (t) represents the real-time outdoor temperature, and RH(t) represents the real-time outdoor relative humidity.

7. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 6, characterized in that: To minimize the overall energy consumption of the data center, the optimal operating temperature of the server under different CPU usage and outdoor real-time temperature and humidity is calculated as follows: With the goal of minimizing the overall energy consumption of the data center, based on the data center energy consumption model, the real-time utilization rate of CPU resources and the operating conditions under the real-time outdoor temperature and humidity are optimized and calculated: Optimization goal: min EC total =∫(∑P server (t)+P cooling,sys (t))dt; Optimization constraint: T server,CPU (t)≤T server,CPU,max ; Optimization variables: {TerFan FP (t),ChiPump FP (t),CdPump FP (t),Chiller FP (t),CT FP (t)} Where, min EC total represents the minimum energy consumption of the data center, T server,CPU (t) represents the matrix composed of the real-time CPU temperatures of each server in the data center, T server,CPU,max Indicates the upper limit of the maximum operating temperature of the server CPU; The optimal operating temperature of the server corresponding to each working condition optimization result is sorted out to form a temperature control guidance list.

8. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 7, characterized in that: The collaborative scheduling strategy for optimal temperature control of a given set of active servers includes: According to the scheduling characteristics of IT systems, formulate a scheduling strategy targeting a given active server: Number the data center servers; When the IT system receives a new VM deployment request, it obtains the CPU resources required by the VM. According to the server number, the remaining CPU resources of the active servers in the data center are sequentially traversed. When the remaining CPU resources of the active server are available for deploying a new virtual machine, the new virtual machine is deployed on the server. If the remaining CPU resources of the active server are not enough to deploy a new virtual machine, a dormant server is made active to deploy the new virtual machine. At each preset time interval, the CPU resource requirements of the virtual machines on the active servers are summed up and compared with the rated CPU resources of a single server to determine the minimum number of active servers required; In the formula, ∑RS VMi (t) represents the sum of CPU resource requirements of the virtual machines hosted in real time on each server, C server Indicates the rated CPU resource capacity of a single server, Num server,active (t) represents the number of active servers required at this moment; According to the required number of active servers, the target server to be migrated is determined by server number, and the virtual machines on the active servers in the data center are migrated to the target server with the goal of balancing the CPU resource utilization of the target server.

9. The data center computing and cooling collaborative scheduling method based on optimal operating temperature according to claim 8, characterized in that: According to the scheduling characteristics of the cooling system, a collaborative scheduling strategy is formulated with the goal of optimal temperature control for a given active server: Calculate the real-time average CPU usage of active servers, and determine the operating conditions of the data center according to the temperature control guidance list in combination with the outdoor temperature and humidity; Determine the optimal operating temperature of the server corresponding to the operating condition from the temperature control guidance list, and use it as the control threshold of the cooling system; The real-time maximum temperature of the CPU in the active server is taken as the control object, and the various devices in the cooling system are regulated by feedback control according to the set control threshold.

10. A data center computing and cooling collaborative scheduling system, characterized in that: include: Information collection module, used to obtain the operating data of the IT system and cooling system of the data center; A first processing module is used to establish a server power consumption model and a cooling system power consumption model according to the operation data of the IT system and the cooling system, and to establish a data center energy consumption model based on the server power consumption model and the cooling system power consumption model; The energy consumption analysis module calculates the optimal operating temperature of the server under different CPU usage rates and outdoor real-time temperature and humidity based on the data center energy consumption model and aims to minimize the overall energy consumption of the data center, and forms a temperature control guidance list; The strategy generation module formulates the collaborative scheduling strategy with the goal of optimal temperature control for a given active server; The scheduling control module adjusts the real-time frequency of each terminal fan, the real-time frequency of the chilled water pump, the real-time frequency of the cooling water pump, the real-time frequency of the cooling tower fan, and the real-time frequency of the refrigeration compressor according to the collaborative scheduling strategy, so as to adjust the real-time maximum temperature of the CPU in the server to the optimal operating temperature of the server.

Images