CN109840180B - Server operation power consumption management method and device and computer readable storage medium - Google Patents
Server operation power consumption management method and device and computer readable storage medium Download PDFInfo
- Publication number
- CN109840180B CN109840180B CN201811550110.XA CN201811550110A CN109840180B CN 109840180 B CN109840180 B CN 109840180B CN 201811550110 A CN201811550110 A CN 201811550110A CN 109840180 B CN109840180 B CN 109840180B
- Authority
- CN
- China
- Prior art keywords
- power consumption
- cabinet
- server
- servers
- actual power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Power Sources (AREA)
Abstract
The scheme relates to pedestal operation and maintenance, and provides a method, a device and a storage medium for managing the running power consumption of a server, wherein the method comprises the following steps: classifying servers of the data center according to the same brand and the same configuration; the method comprises the steps of collecting actual power consumption values of the same type of servers, establishing an actual power consumption load relation table, calculating the actual power consumption of the same type of servers in the following mode, wherein estimated power consumption is the actual power consumption of a cabinet/the number of the cabinet servers, wherein the same type of servers are installed on the cabinet, and according to the estimated power consumption or the average monthly actual power consumption or the average seasonal actual power consumption of the same type of servers, a cabinet configuration server scheme is formed by combining the rated power consumption of the cabinet. According to the invention, the actual power consumption of the server in the cabinet is monitored, and the cabinet is configured according to the actual power consumption of the server, so that the utilization rate of the cabinet can be improved, and the waste of the space of the cabinet is reduced. The power supply utilization rate of the cabinet is improved, and the PUE value is reduced.
Description
Technical Field
The invention relates to pedestal operation and maintenance, in particular to a method and a device for managing running power consumption of a server and a computer readable storage medium.
Background
In the data center, the brand and the model of the server are more, the marked load power consumption of the server leaving the factory is an interval range, however, in order to improve the utilization rate of limited cabinet resources, the data center needs to reasonably set the number of the servers in the cabinet, and along with the increase of the service volume and the service types, the servers need to be added or reduced in the cabinet to adapt to the service requirements. If it is desired to add servers to the rack, as server usage rates scale up, there may be an overload risk from power consumption load increases. Therefore, the existing data center adopts a prudent configuration mode, the cabinet power supply is not fully utilized due to the prudent, the cabinet resources are wasted, the power consumption is evaluated according to the name plate provided by a server manufacturer, the cabinet space is wasted, the power supply use efficiency is low, and the PUE is too high.
Disclosure of Invention
In order to solve the above technical problems, the present invention provides a method for managing power consumption of a server, which is applied to an electronic device, and includes: classifying servers of a data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server; acquiring actual power consumption values of the same type of servers, sampling according to CPU loads at an acquisition time point, and establishing an actual power consumption load relation table, wherein the actual power consumption of the same type of servers is calculated in the following way, and estimated power consumption is the actual power consumption of a cabinet/the number of servers of the cabinet, wherein the same type of servers are installed on the cabinet, monthly average actual power consumption of the servers is also calculated, and the monthly average actual power consumption is the monthly average power consumption of the cabinet/the number of servers of the cabinet, wherein the same type of servers are installed on the cabinet, the seasonal average actual power consumption of the servers is also calculated, and the seasonal average actual power consumption of the servers is the quarterly average power consumption of the cabinet/the number of servers of the cabinet, and the same type of servers are installed on the cabinet; and forming a cabinet configuration server scheme by combining the rated power consumption of the cabinet according to the estimated power consumption or the average monthly actual power consumption or the average seasonal actual power consumption of the same type of server.
Preferably, the actual power consumption of each server is monitored in real time, the difference between the actual power consumption of each server and the average value of the actual power consumption of all servers in the cabinet where the server is located is obtained, if the difference is higher than the difference limit value, the CPU of the server with the highest power consumption is subjected to frequency reduction processing until the difference of the actual power consumption of the server is within the difference limit value range, and if the difference is lower than the difference limit value, the CPU of the server with the highest power consumption is subjected to frequency increase processing, so that the servers are subjected to load average distribution, and the power consumption is balanced.
Preferably, after a cabinet configuration server scheme is formed according to estimated power consumption or monthly average actual power consumption or seasonal average actual power consumption of the same type of server and by combining with cabinet rated power consumption, a machine learning method is further adopted to assist in optimizing the cabinet configuration server scheme, a neural network model is adopted to receive an input cabinet configuration server scheme and process the input cabinet configuration server scheme so as to generate corresponding scores for the cabinet configuration server scheme, and the optimal cabinet configuration server scheme is selected according to the scoring condition.
Preferably, the step of optimizing the rack configuration server scheme by using a machine learning method includes: constructing a cabinet configuration classification model; acquiring a training data set for training a cabinet configuration classification model, wherein the training data set comprises cabinet configuration server schemes described corresponding to different service requirements and scores judged according to a user-defined grading standard; training a cabinet configuration classification model by adopting a training data set, inputting different cabinet configuration server schemes in the training data set into the cabinet configuration classification model, classifying the output of the cabinet configuration classification model by a classifier, and controlling the classification precision of the cabinet configuration classification model by a loss function so as to improve the classification precision of the cabinet configuration classification model; and configuring a server for the cabinet according to the service requirement by utilizing the trained cabinet configuration classification model.
Preferably, constructing the cabinet configuration classification model comprises the following steps: setting cabinet configuration training model parameters, wherein the cabinet configuration training model is a deep neural network model and comprises an input layer, a bidirectional GRU (general purpose unit), a softmax layer and a full connection layer; inputting a plurality of cabinet configuration training schemes into the cabinet configuration training model, training the cabinet configuration training model, and updating bidirectional GRU parameters in the cabinet configuration training model; initializing the bidirectional GRU parameters of the cabinet configuration classification model according to the updated bidirectional GRU parameters of the cabinet configuration training model, and configuring the parameters of the cabinet configuration classification model except the bidirectional GRU parameters; inputting the cabinet configuration training scheme into a cabinet configuration training model, performing supervision and countermeasure training to obtain a cabinet configuration classification model, inputting the newly added cabinet configuration server scheme into the cabinet configuration classification model after the supervision and countermeasure training, performing unsupervised virtual countermeasure training, updating parameters of the cabinet configuration classification model, and obtaining the cabinet configuration classification model.
Preferably, the calculation formula of the GRU is as follows:
zt＝σ(ftUz+s(t-1)Wz)
rt＝σ(ftUr+s(t-1)Wr)
ht＝tanh(ftUh+(s(t-1)*rt)Wh)
st＝(1-zt)*ft+zt*s(t-1)
wherein z istIs an update gate to control how many candidate hidden layers h are addedtThe information of (a);
rtis a reset gate for calculating a candidate hidden layer htControlling how many previous hidden layers s are kept(t-1)The information of (a);
htis a candidate hidden layer;
u, W is a weight matrix;
ftis the input data at time t;
s(t-1)is the activation value of hidden layer neuron at time t-1;
sigma represents a sigmoid activation function;
tanh is the activation function;
stis the activation value of the hidden layer neuron at time t.
Preferably, in the process of forming a cabinet configuration server scheme by combining the estimated power consumption of the same type of server, the monthly average actual power consumption or the seasonal average actual power consumption and the rated power consumption of the cabinet, the servers with the actual power consumption larger than a set threshold and the servers with the actual power consumption smaller than the set threshold are configured for the cabinet in a one-to-one correspondence manner, wherein the set threshold is the average value of the actual power consumption of the servers of different types.
The present invention also provides an electronic device, comprising: a memory and a processor, the memory having stored therein a server-run power consumption management program that when executed by the processor implements the steps of: classifying servers of a data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server; acquiring actual power consumption values of the same type of servers, sampling according to CPU loads at an acquisition time point, and establishing an actual power consumption load relation table, wherein the actual power consumption of the same type of servers is calculated in the following way, and estimated power consumption is the actual power consumption of a cabinet/the number of servers of the cabinet, wherein the same type of servers are installed on the cabinet, monthly average actual power consumption of the servers is also calculated, and the monthly average actual power consumption is the monthly average power consumption of the cabinet/the number of servers of the cabinet, wherein the same type of servers are installed on the cabinet, the seasonal average actual power consumption of the servers is also calculated, and the seasonal average actual power consumption of the servers is the quarterly average power consumption of the cabinet/the number of servers of the cabinet, and the same type of servers are installed on the cabinet; and forming a cabinet configuration server scheme by combining the rated power consumption of the cabinet according to the estimated power consumption or the average monthly actual power consumption or the average seasonal actual power consumption of the same type of server.
Preferably, the actual power consumption of each server is monitored in real time, the difference between the actual power consumption of each server and the average value of the actual power consumption of all servers in the cabinet where the server is located is obtained, if the difference is higher than the difference limit value, the CPU of the server with the highest power consumption is subjected to frequency reduction processing until the difference of the actual power consumption of the server is within the difference limit value range, and if the difference is lower than the difference limit value, the CPU of the server with the highest power consumption is subjected to frequency increase processing, so that the servers are subjected to load average distribution, and the power consumption is balanced.
The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program includes program instructions, and the program instructions, when executed by a processor, implement the server operation power consumption management method described above.
According to the invention, the actual power consumption of the server in the cabinet is monitored, and the cabinet is configured according to the actual power consumption of the server, so that the utilization rate of the cabinet can be improved, and the waste of the space of the cabinet is reduced. The power supply utilization rate of the cabinet is improved, and the PUE value is reduced.
Drawings
The above features and technical advantages of the present invention will become more apparent and readily appreciated from the following description of the embodiments thereof taken in conjunction with the accompanying drawings.
FIG. 1 is a flow chart of a server operation power consumption management method according to an embodiment of the present invention;
FIG. 2 is a diagram of a hardware architecture of an electronic device according to an embodiment of the invention;
fig. 3 is a block configuration diagram of a server running a power consumption manager according to an embodiment of the present invention.
Detailed Description
Embodiments of a server operation power consumption management method, apparatus, and computer-readable storage medium according to the present invention will be described below with reference to the accompanying drawings. Those of ordinary skill in the art will recognize that the described embodiments can be modified in various different ways, or combinations thereof, without departing from the spirit and scope of the present invention. Accordingly, the drawings and description are illustrative in nature and not intended to limit the scope of the claims. Furthermore, in the present description, the drawings are not to scale and like reference numerals refer to like parts.
Fig. 1 is a schematic flow chart of a server operation power consumption management method according to an embodiment of the present invention. The method comprises the following steps:
and step S10, classifying the servers of the data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server.
The primary purpose of a data center is to run applications to process data for businesses and functioning organizations. Each application can lease a server of the data center to perform its intended business function. Taking a certain data center as an example, the data center mainly comprises an HP server, and the main model is DL 380Gen9 power consumption 160W-290W; leonvo is a second multi-brand server, and the main machine type is RD650 power consumption of 170W-245W; the number of Dell servers is slightly less than Lenovo, the main types are R720 power consumption 220W and R730 power consumption 200W-260W, R730XD power consumption is about 275W, and R920 power consumption 450W, R930 power consumption 500W; the main models of the wave server are about 210W power consumption of NF5270M3 and about 130W power consumption of NF 5270M. The stored FAS 8060 double-controller consumes about 3.6 KW-3.7 KW of power.
And step S30, acquiring the actual power consumption value of the same type of server, sampling according to the CPU load at the acquisition time point, and establishing an actual power consumption load relation table.
At present, when each data center configures a server for a cabinet, power consumption is estimated based on a nominal power marked by a server manufacturer, and the estimated power consumption is usually 25% of the nominal power. However, the power consumption range of each server is often related to the service type, and the load formed by different service types is different. Therefore, the power consumption values of the same type of server are collected, sampling is carried out according to the OS (CPU) load at the collection time point, and a power consumption load relation table is established, as shown in the table I.
Watch 1
Wherein, the actual power consumption of the same type of server is calculated in the following way,
the estimated power consumption is the actual power consumption of the cabinet/the number of the servers in the cabinet, wherein the same type of servers are installed on the cabinet,
and the average monthly power consumption of the servers is calculated, wherein the average monthly power consumption is the average monthly power consumption of the cabinet/the number of the cabinet servers, the same type of servers are installed on the cabinet,
in addition, the current season average actual power consumption of the servers is calculated, wherein the current season average actual power consumption is equal to the cabinet season average power consumption/the number of the cabinet servers, and the same type of servers are installed on the cabinets;
and step S50, forming a cabinet configuration server scheme according to the estimated power consumption or the average monthly actual power consumption or the average seasonal actual power consumption of the same type of server and by combining the rated power consumption of the cabinet.
The following describes the present embodiment with specific examples. The data center needs to allocate a reasonable number of configured servers to different application providers, so that the application providers can use the allocated servers to perform their own business, for example, multiple servers of the kyoto rental data center perform business supporting the kyoto shopping mall thereof. In the embodiment, the actual power consumption of the same type of server corresponding to different services is considered, and the efficiency of the server can be exerted to a greater extent. The branding, configuration and number of servers in the rack are set according to the traffic category. For example, the nominal power of the R430 server of Dell is 1100W, the estimated power consumption is about 25% of the nominal power, that is, 275W, and the rated power consumption of the cabinet is 4000W, and if the factors of power supply safety, heat dissipation and the like are not considered, the cabinet can be configured with 14R 430 servers of Dell. If the power supply safety of the cabinet is comprehensively considered, 12 cabinets can be configured. And the estimated power consumption of the R430 server of Dell is 140W, which is lower than the estimated power consumption, and if the influence factors such as cabinet power supply and heat dissipation are not considered, the cabinet is configured only by considering the rated power consumption of the cabinet and the estimated power consumption of the server, and the cabinet can be configured with 28R 430 servers of Dell at most. And the factors of power supply safety, heat dissipation and the like of the cabinet are comprehensively considered, and 20 Dell R430 servers can be configured.
Through the arrangement, although the servers are added in the cabinet, the actual power consumption is within a reasonable range, so that the servers can be arranged to the maximum extent in the cabinet, the power can be supplied to the cabinet by fully utilizing the cabinet, the service efficiency of the power supply of the cabinet is improved, the PUE (power supply use efficiency) value is reduced, and the power supply of the cabinet can be maintained within the range of the power supply threshold value.
In an optional embodiment, the actual power consumption of each server is monitored in real time, the difference between the actual power consumption of each server and the average value of the actual power consumption of all servers in the cabinet where the server is located is obtained, if the difference is higher than the difference limit value, the CPU of the server with the highest power consumption is subjected to frequency reduction processing until the difference of the actual power consumption of the server is within the difference limit value range, and if the difference is lower than the difference limit value, the CPU of the server with the highest power consumption is subjected to frequency increase processing, so that the servers are evenly distributed with loads, and the power consumption is balanced.
In an optional embodiment, in the process of forming a cabinet configuration server scheme by combining the estimated power consumption of the same type of server, the monthly average actual power consumption or the seasonal average actual power consumption and the cabinet rated power consumption, the servers with the actual power consumption greater than a set threshold and the servers with the actual power consumption less than the set threshold are configured for the cabinet in a one-to-one correspondence manner, wherein the set threshold is the average value of the actual power consumption of the different types of servers.
In an optional embodiment, after a cabinet configuration server scheme is formed according to estimated power consumption of the same type of server or monthly average actual power consumption or seasonal average actual power consumption and by combining with cabinet rated power consumption, a machine learning method is further adopted to assist the cabinet configuration server scheme, a neural network model is adopted to receive an input cabinet configuration server scheme and process the input cabinet configuration server scheme so as to generate corresponding scores for the cabinet configuration server scheme, and an optimal cabinet configuration server scheme is selected according to the scoring condition.
Further, the method for assisting the cabinet to configure the server by adopting the machine learning method comprises the following steps:
constructing a cabinet configuration classification model;
acquiring a training data set for training a cabinet configuration classification model, wherein the training data set comprises configuration schemes of cabinets described corresponding to different service requirements and scores judged according to a user-defined grading standard;
the cabinet configuration classification model is trained by adopting a training data set, different cabinet configuration schemes in the training data set are input into the cabinet configuration classification model, the output of the cabinet configuration classification model is classified by a classifier, and then the classification precision of the cabinet configuration classification model is controlled by a loss function, so that the classification precision of the cabinet configuration classification model is improved.
Further, the constructing of the cabinet configuration classification model comprises the following steps:
setting cabinet configuration training model parameters, wherein the cabinet configuration training model is a deep neural network model and comprises an input layer, a bidirectional GRU (general purpose unit), a softmax layer and a full connection layer;
inputting a plurality of cabinet configuration training schemes into the cabinet configuration training model, training the cabinet configuration training model, and updating bidirectional GRU parameters in the cabinet configuration training model;
initializing the bidirectional GRU parameters of the cabinet configuration classification model according to the updated bidirectional GRU parameters of the cabinet configuration training model, and configuring the parameters of the cabinet configuration classification model except the bidirectional GRU parameters;
inputting the cabinet configuration training scheme into the cabinet configuration training model, performing supervision and countermeasure training to obtain a cabinet configuration classification model, inputting the newly added cabinet configuration scheme into the cabinet configuration classification model after the supervision and countermeasure training, performing unsupervised virtual countermeasure training, updating parameters of the cabinet configuration classification model, and obtaining the cabinet configuration classification model.
Further, the calculation formula of GRU is as follows:
zt＝σ(ftUz+s(t-1)Wz)
rt＝σ(ftUr+s(t-1)Wr)
ht＝tanh(ftUh+(s(t-1)*rt)Wh)
st＝(1-zt)*ft+zt*s(t-1)
wherein z istIs an update gate to control how many candidate hidden layers h are addedtThe information of (a);
rtis a reset gate for calculating a candidate hidden layer htControlling how many previous time hidden layers s are reserved(t-1)The information of (a);
htis a candidate hidden layer;
u, W is a weight matrix;
ftis the input data at time t;
s(t-1)is the activation value of hidden layer neuron at time t-1;
sigma represents a sigmoid activation function;
tanh is the activation function;
stis the activation value of the hidden layer neuron at time t.
Fig. 2 is a schematic diagram of a hardware architecture of an electronic device according to an embodiment of the invention. In the present embodiment, the electronic device 2 is a device capable of automatically performing numerical calculation and/or information processing according to a preset or stored instruction. For example, the server may be a smart phone, a tablet computer, a notebook computer, a desktop computer, a rack server, a blade server, a tower server, or a rack server (including an independent server or a server cluster composed of a plurality of servers). As shown in fig. 2, the electronic device 2 includes at least, but is not limited to, a memory 21, a processor 22, and a network interface 23, which are communicatively connected to each other through a system bus. Wherein: the memory 21 includes at least one type of computer-readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a Read Only Memory (ROM), an Electrically Erasable Programmable Read Only Memory (EEPROM), a Programmable Read Only Memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the storage 21 may be an internal storage unit of the electronic device 2, such as a hard disk or a memory of the electronic device 2. In other embodiments, the memory 21 may also be an external storage device of the electronic apparatus 2, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), or the like, provided on the electronic apparatus 2. Of course, the memory 21 may also comprise both an internal memory unit of the electronic apparatus 2 and an external memory device thereof. In this embodiment, the memory 21 is generally used for storing an operating system installed in the electronic device 2 and various types of application software, such as a power consumption management program code executed by the server. Further, the memory 21 may also be used to temporarily store various types of data that have been output or are to be output.
The processor 22 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data Processing chip in some embodiments. The processor 22 is generally configured to control the overall operation of the electronic apparatus 2, such as performing data interaction or communication related control and processing with the electronic apparatus 2. In this embodiment, the processor 22 is configured to execute the program codes or process data stored in the memory 21, for example, execute the server operation power consumption management program.
The network interface 23 may comprise a wireless network interface or a wired network interface, and the network interface 23 is generally used for establishing communication connection between the electronic device 2 and other electronic devices. For example, the network interface 23 is configured to connect the electronic device 2 to a push platform through a network, establish a data transmission channel and a communication connection between the electronic device 2 and the push platform, and the like. The network may be a wireless or wired network such as an Intranet (Intranet), the Internet (Internet), a Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), a 4G network, a 5G network, Bluetooth (Bluetooth), Wi-Fi, and the like.
Optionally, the electronic device 2 may further comprise a display, which may also be referred to as a display screen or a display unit. In some embodiments, the display device can be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an Organic Light-Emitting Diode (OLED) display, and the like. The display is used for displaying information processed in the electronic apparatus 2 and for displaying a visualized user interface.
It is noted that fig. 2 only shows the electronic device 2 with components 21-23, but it is to be understood that not all shown components are required to be implemented, and that more or less components may be implemented instead.
The memory 21 containing the readable storage medium may include an operating system, a server running power consumption management program 50, and the like. The processor 22, when executing the server running power consumption management program 50 in the memory 21, implements the following steps:
and step S10, classifying the servers of the data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server.
And step S30, acquiring the actual power consumption value of the same type of server, sampling according to the CPU load at the acquisition time point, and establishing an actual power consumption load relation table, wherein the actual power consumption can be estimated power consumption or monthly average actual power consumption or current season average actual power consumption.
And step S50, forming a cabinet configuration server scheme according to the estimated power consumption or the average monthly actual power consumption or the average seasonal actual power consumption of the same type of server and by combining the rated power consumption of the cabinet.
In this embodiment, the server running power consumption management program stored in the memory 21 may be divided into one or more program modules, and the one or more program modules are stored in the memory 21 and executed by one or more processors (in this embodiment, the processor 22) to complete the present invention. For example, fig. 3 shows a schematic diagram of program modules of the server running power consumption management program, and in this embodiment, the server running power consumption management program 50 may be divided into a server classification module 501, a power consumption load relation table establishing module 502, and a cabinet configuration module 503. The program module referred to in the present invention refers to a series of computer program instruction segments capable of performing specific functions, and is more suitable than a program for describing the execution process of the server running power consumption management program in the electronic device 2. The following description will specifically describe specific functions of the program modules.
The server classification module 501 is configured to classify servers of the data center according to the same brand and the same configuration, and use the servers of the same brand and the same configuration as the same class server.
The power consumption load relationship table establishing module 502 is configured to establish a power consumption load relationship table, since the power consumption is estimated by using the nominal power marked by the server manufacturer as a reference when each data center configures a server for a cabinet at present, the estimated power consumption is usually 25% of the nominal power. However, the power consumption range of each server is often related to the service type, and the load formed by different service types is different. Therefore, the power consumption values of the same type of server are collected, sampling is carried out according to the OS (CPU) load at the collection time point, and a power consumption load relation table is established, as shown in the table I.
Wherein the actual power consumption of the same type of server is calculated in the following way,
the estimated power consumption is the actual power consumption of the cabinet/the number of the servers in the cabinet, wherein the same type of servers are installed on the cabinet,
and the average monthly power consumption of the servers is calculated, wherein the average monthly power consumption is the average monthly power consumption of the cabinet/the number of the cabinet servers, the same type of servers are installed on the cabinet,
in addition, the current season average actual power consumption of the servers is calculated, wherein the current season average actual power consumption is equal to the cabinet season average power consumption/the number of the cabinet servers, and the same type of servers are installed on the cabinets;
the cabinet configuration module 503 is configured to configure a server for the cabinet according to the estimated power consumption of the same type of server, the monthly average actual power consumption, or the seasonal average actual power consumption, in combination with the rated power consumption of the cabinet.
The following describes the present embodiment with specific examples. The data center needs to allocate a reasonable number of configured servers to different application providers, so that the application providers can use the allocated servers to perform their own business, for example, multiple servers of the kyoto rental data center perform business supporting the kyoto shopping mall thereof. In the embodiment, the actual power consumption of the same type of server corresponding to different services is considered, and the efficiency of the server can be exerted to a greater extent. The branding, configuration and number of servers in the rack are set according to the traffic category. For example, the nominal power of the R430 server of Dell is 1100W, the estimated power consumption is about 25% of the nominal power, namely 275W, the rated power consumption of the cabinet is 4000W, and then the cabinet power supply safety is comprehensively considered, so that 12R 430 servers of Dell can be configured. And the estimated power consumption of the R430 server of Dell is 140W, which is lower than the estimated power consumption, and 20R 430 servers of Dell can be configured by comprehensively considering the cabinet power supply safety.
In an optional embodiment, the system further includes an actual power consumption monitoring module 504, which monitors actual power consumption of each server in real time, and obtains a difference between the actual power consumption of each server and an average value of the actual power consumption of all servers in the cabinet where the server is located, if the difference is higher than a difference limit, the CPU of the server with the highest power consumption is subjected to frequency reduction processing until the difference between the actual power consumption of the server is within the difference limit, and if the difference is lower than the difference limit, the CPU of the server with the highest power consumption is subjected to frequency increase processing, so that the servers distribute loads evenly, and the power consumption is balanced.
In an optional embodiment, in the process of forming the cabinet configuration scheme according to the estimated power consumption of the same type of server, the monthly average actual power consumption, or the seasonal average actual power consumption, and by combining the rated power consumption of the cabinet, the cabinet configuration module 503 configures the servers for the cabinet in a one-to-one correspondence manner between the servers whose actual power consumption is greater than the set threshold and the servers whose actual power consumption is less than the set threshold, where the set threshold is an average value of the actual power consumptions of the servers of different types.
In an optional embodiment, the intelligent configuration module 505 is further included, after a rack configuration server scheme is formed according to estimated power consumption of the same type of server, monthly average actual power consumption or seasonal average actual power consumption, and in combination with rack rated power consumption, the machine learning method of the intelligent configuration module 505 is further used to assist the rack configuration server scheme, an input rack configuration server scheme is received by using a neural network model and is processed to generate a corresponding score for the rack configuration server scheme, and an optimal configuration scheme is selected according to the scoring condition.
Further, the intelligent configuration module 505 adopts a machine learning method to assist the cabinet configuration scheme, and the steps are as follows:
constructing a cabinet configuration classification model;
acquiring a training data set for training a cabinet configuration classification model, wherein the training data set comprises configuration schemes of cabinets described corresponding to different service requirements and scores judged according to a user-defined grading standard;
the cabinet configuration classification model is trained by adopting a training data set, different cabinet configuration schemes in the training data set are input into the cabinet configuration classification model, the output of the cabinet configuration classification model is classified by a classifier, and then the classification precision of the cabinet configuration classification model is controlled by a loss function, so that the classification precision of the cabinet configuration classification model is improved.
Further, the intelligent configuration module 505 building the cabinet configuration classification model includes the following steps:
setting cabinet configuration training model parameters, wherein the cabinet configuration training model is a deep neural network model and comprises an input layer, a bidirectional GRU (general purpose unit), a softmax layer and a full connection layer;
inputting a plurality of cabinet configuration training schemes into the cabinet configuration training model, training the cabinet configuration training model, and updating bidirectional GRU parameters in the cabinet configuration training model;
initializing the bidirectional GRU parameters of the cabinet configuration classification model according to the updated bidirectional GRU parameters of the cabinet configuration training model, and configuring the parameters of the cabinet configuration classification model except the bidirectional GRU parameters;
inputting the cabinet configuration training scheme into a cabinet configuration training model, performing supervision and countermeasure training to obtain a cabinet configuration classification model, inputting the newly added cabinet configuration scheme into the cabinet configuration classification model after the supervision and countermeasure training, performing unsupervised virtual countermeasure training, updating parameters of the cabinet configuration classification model, and obtaining the cabinet configuration classification model.
Further, the calculation formula of GRU is as follows:
zt＝σ(ftUz+s(t-1)Wz)
rt＝σ(ftUr+s(t-1)Wr)
ht＝tanh(ftUh+(s(t-1)*rt)Wh)
st＝(1-zt)*ft+zt*s(t-1)
wherein z istIs an update gate to control how many candidate hidden layers h are addedtThe information of (a);
rtis a reset gate for calculating a candidate hidden layer htControlling how many previous hidden layers s are kept(t-1)The information of (a);
htis a candidate hidden layer;
u, W is a weight matrix;
ftis the input data at time t;
s(t-1)is the activation value of hidden layer neuron at time t-1;
sigma represents a sigmoid activation function;
tanh is the activation function;
stis the activation value of the hidden layer neuron at time t.
Furthermore, the embodiment of the present invention also provides a computer-readable storage medium, which may be any one or any combination of a hard disk, a multimedia card, an SD card, a flash memory card, an SMC, a Read Only Memory (ROM), an Erasable Programmable Read Only Memory (EPROM), a portable compact disc read only memory (CD-ROM), a USB memory, and the like. The computer readable storage medium includes a server running power consumption management program and the like, and the server running power consumption management program 50 when executed by the processor 22 implements the following operations:
and step S10, classifying the servers of the data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server.
And step S30, acquiring actual power consumption values of the same type of servers, sampling according to CPU loads at the acquisition time point, and establishing an actual power consumption load relation table, wherein the actual power consumption values can be estimated power consumption, monthly average actual power consumption or seasonal average actual power consumption.
And step S50, forming a cabinet configuration server scheme according to the estimated power consumption or the average monthly actual power consumption or the average seasonal actual power consumption of the same type of server and by combining the rated power consumption of the cabinet.
The embodiment of the computer-readable storage medium of the present invention is substantially the same as the embodiment of the server operation power consumption management method and the electronic device 2, and will not be described herein again.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (9)
1. A server operation power consumption management method is applied to an electronic device and is characterized by comprising the following steps:
classifying servers of a data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server;
acquiring actual power consumption values of the same type of servers, sampling according to CPU loads at an acquisition time point, and establishing an actual power consumption load relation table, wherein the actual power consumption of the same type of servers is calculated in the following mode, estimated power consumption is the actual power consumption of a cabinet/the number of the cabinet servers, monthly average actual power consumption of the servers is also calculated, the monthly average actual power consumption is the monthly average power consumption of the cabinet/the number of the cabinet servers, the seasonal average actual power consumption of the servers is also calculated, the seasonal average actual power consumption of the cabinets is the seasonal average power consumption of the cabinets/the number of the cabinet servers, and the same type of servers are installed on the cabinets;
according to estimated power consumption or monthly average actual power consumption or seasonal average actual power consumption of the same type of server, a cabinet configuration server scheme is formed by combining with cabinet rated power consumption, a machine learning method is further adopted to assist in optimizing the cabinet configuration server scheme, a neural network model is adopted to receive an input cabinet configuration server scheme and process the input cabinet configuration server scheme so as to generate corresponding scores for the cabinet configuration server scheme, and the optimal cabinet configuration server scheme is selected according to the scoring condition.
2. The server operation power consumption management method according to claim 1, wherein the actual power consumption of each server is monitored in real time, the difference between the actual power consumption of each server and the average value of the actual power consumption of all servers in the cabinet where the server is located is obtained, if the difference is higher than the difference limit value, the CPU of the server with the highest power consumption is subjected to down-frequency processing until the difference of the actual power consumption of the server is within the difference limit value range, and if the difference is lower than the difference limit value, the CPU of the server with the highest power consumption is subjected to up-frequency processing, so that the loads are evenly distributed to the servers, and the power consumption is balanced.
3. The server operation power consumption management method according to claim 1, wherein the step of optimizing the rack configuration server solution by using a machine learning method comprises:
constructing a cabinet configuration classification model;
acquiring a training data set for training a cabinet configuration classification model, wherein the training data set comprises cabinet configuration server schemes described corresponding to different service requirements and scores judged according to a user-defined grading standard;
training a cabinet configuration classification model by adopting a training data set, inputting different cabinet configuration server schemes in the training data set into the cabinet configuration classification model, classifying the output of the cabinet configuration classification model by a classifier, and controlling the classification precision of the cabinet configuration classification model by a loss function so as to improve the classification precision of the cabinet configuration classification model;
and configuring a server for the cabinet according to the service requirement by utilizing the trained cabinet configuration classification model.
4. The server operation power consumption management method according to claim 3, wherein the step of constructing the cabinet configuration classification model comprises the following steps: setting cabinet configuration training model parameters, wherein the cabinet configuration training model is a deep neural network model and comprises an input layer, a bidirectional GRU (general purpose unit), a softmax layer and a full connection layer; inputting a plurality of cabinet configuration training schemes into the cabinet configuration training model, training the cabinet configuration training model, and updating bidirectional GRU parameters in the cabinet configuration training model; initializing the bidirectional GRU parameters of the cabinet configuration classification model according to the updated bidirectional GRU parameters of the cabinet configuration training model, and configuring the parameters of the cabinet configuration classification model except the bidirectional GRU parameters; inputting the cabinet configuration training scheme into a cabinet configuration training model, performing supervision and countermeasure training to obtain a cabinet configuration classification model, inputting the newly added cabinet configuration server scheme into the cabinet configuration classification model after the supervision and countermeasure training, performing unsupervised virtual countermeasure training, updating parameters of the cabinet configuration classification model, and obtaining the cabinet configuration classification model.
5. The server operation power consumption management method according to claim 4, wherein the calculation formula of the GRU is as follows:
zt＝σ(ftUz+s(t-1)Wz)
rt＝σ(ftUr+s(t-1)Wr)
ht＝tanh(ftUh+(s(t-1)*rt)Wh)
st＝(1-zt)*ft+zt*s(t-1)
wherein z istIs an update gate to control how many candidate hidden layers h are addedtThe information of (a);
rtis a reset gate for calculating a candidate hidden layer htControlling how many previous hidden layers s are kept(t-1)The information of (a);
htis a candidate hidden layer;
u, W is a weight matrix;
ftis the input data at time t;
s(t-1)is the activation value of hidden layer neuron at time t-1;
sigma represents a sigmoid activation function;
tanh is the activation function;
stis the activation value of the hidden layer neuron at time t.
6. The method for managing the running power consumption of the server according to claim 1, wherein in the process of forming the cabinet configuration server scheme according to the estimated power consumption of the same type of server, the monthly average actual power consumption or the seasonal average actual power consumption and by combining with the rated power consumption of the cabinet, the servers with the actual power consumption larger than a set threshold value and the servers with the actual power consumption smaller than the set threshold value are configured for the cabinet in a one-to-one correspondence manner, wherein the set threshold value is the average value of the actual power consumption of the different types of servers.
7. An electronic device, comprising: a memory and a processor, the memory having stored therein a server-run power consumption management program that when executed by the processor implements the steps of:
classifying servers of a data center according to the same brand and the same configuration, and taking the servers of the same brand and the same configuration as the same type of server;
acquiring actual power consumption values of the same type of servers, sampling according to CPU loads at an acquisition time point, and establishing an actual power consumption load relation table, wherein the actual power consumption of the same type of servers is calculated in the following mode, estimated power consumption is the actual power consumption of a cabinet/the number of the cabinet servers, monthly average actual power consumption of the servers is also calculated, the monthly average actual power consumption is the monthly average power consumption of the cabinet/the number of the cabinet servers, the seasonal average actual power consumption of the servers is also calculated, the seasonal average actual power consumption of the cabinets is the seasonal average power consumption of the cabinets/the number of the cabinet servers, and the same type of servers are installed on the cabinets;
according to estimated power consumption or monthly average actual power consumption or seasonal average actual power consumption of the same type of server, a cabinet configuration server scheme is formed by combining with cabinet rated power consumption, a machine learning method is further adopted to assist in optimizing the cabinet configuration server scheme, a neural network model is adopted to receive an input cabinet configuration server scheme and process the input cabinet configuration server scheme so as to generate corresponding scores for the cabinet configuration server scheme, and the optimal cabinet configuration server scheme is selected according to the scoring condition.
8. The electronic device according to claim 7, wherein the actual power consumption of each server is monitored in real time, and a difference between the actual power consumption of each server and an average value of the actual power consumption of all servers in the cabinet is obtained, if the difference is higher than a difference limit, the CPU of the server with the highest power consumption is subjected to down-frequency processing until the difference between the actual power consumption of the servers is within the difference limit, and if the difference is lower than the difference limit, the CPU of the server with the highest power consumption is subjected to up-frequency processing, so that the loads are evenly distributed to each server, and the power consumption is balanced.
9. A computer-readable storage medium, characterized in that it stores a computer program comprising program instructions which, when executed by a processor, implement the server operation power consumption management method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811550110.XA CN109840180B (en) | 2018-12-18 | 2018-12-18 | Server operation power consumption management method and device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811550110.XA CN109840180B (en) | 2018-12-18 | 2018-12-18 | Server operation power consumption management method and device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109840180A CN109840180A (en) | 2019-06-04 |
| CN109840180B true CN109840180B (en) | 2022-07-08 |
Family
ID=66883224
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811550110.XA Active CN109840180B (en) | 2018-12-18 | 2018-12-18 | Server operation power consumption management method and device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109840180B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110389637A (en) * | 2019-06-26 | 2019-10-29 | 苏州浪潮智能科技有限公司 | A kind of automatic setting method, BMC and the server of server radiating strategy |
| CN113748398B (en) * | 2019-09-06 | 2023-01-17 | 阿里云计算有限公司 | Data processing and task scheduling method, device, system and storage medium |
| CN110837724B (en) * | 2019-11-07 | 2023-09-29 | 交控科技股份有限公司 | Automatic processing method and device for data communication cabinet |
| CN111198799A (en) * | 2019-12-31 | 2020-05-26 | 苏州浪潮智能科技有限公司 | Machine room power consumption early warning method, system, terminal and storage medium based on LSTM |
| CN111736549A (en) * | 2020-06-11 | 2020-10-02 | 上海申瑞继保电气有限公司 | Production line energy-saving auxiliary equipment control method |
| CN114064282B (en) * | 2021-11-23 | 2023-07-25 | 北京百度网讯科技有限公司 | Resource mining method and device and electronic equipment |
| CN118035051A (en) * | 2022-11-11 | 2024-05-14 | 华为技术有限公司 | Energy efficiency evaluation method, device, system and related equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101277200A (en) * | 2007-03-30 | 2008-10-01 | 联想(北京)有限公司 | Method and device for managing multiserver power supply |
| CN105930253A (en) * | 2016-05-05 | 2016-09-07 | 华中科技大学 | Non-invasive server energy consumption decomposition system and method |
| CN105975047A (en) * | 2016-04-29 | 2016-09-28 | 北京百度网讯科技有限公司 | Method and system for adjusting power consumption of cloud data centre |
| CN107329811A (en) * | 2017-06-09 | 2017-11-07 | 北京云集智造科技有限公司 | A kind of power consumption of data center adjusting method and device |
| CN108874112A (en) * | 2018-06-28 | 2018-11-23 | 郑州云海信息技术有限公司 | A kind of whole machine cabinet power consumption limit method, apparatus, equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2446516A4 (en) * | 2009-06-25 | 2015-09-02 | Server Tech Inc | Power distribution apparatus with input and output power sensing and method of use |
| US9958178B2 (en) * | 2014-03-06 | 2018-05-01 | Dell Products, Lp | System and method for providing a server rack management controller |
-
2018
- 2018-12-18 CN CN201811550110.XA patent/CN109840180B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101277200A (en) * | 2007-03-30 | 2008-10-01 | 联想(北京)有限公司 | Method and device for managing multiserver power supply |
| CN105975047A (en) * | 2016-04-29 | 2016-09-28 | 北京百度网讯科技有限公司 | Method and system for adjusting power consumption of cloud data centre |
| CN105930253A (en) * | 2016-05-05 | 2016-09-07 | 华中科技大学 | Non-invasive server energy consumption decomposition system and method |
| CN107329811A (en) * | 2017-06-09 | 2017-11-07 | 北京云集智造科技有限公司 | A kind of power consumption of data center adjusting method and device |
| CN108874112A (en) * | 2018-06-28 | 2018-11-23 | 郑州云海信息技术有限公司 | A kind of whole machine cabinet power consumption limit method, apparatus, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109840180A (en) | 2019-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109840180B (en) | Server operation power consumption management method and device and computer readable storage medium | |
| CN105940378B (en) | For distributing the technology of configurable computing resource | |
| DE102006051672B4 (en) | System and method for adaptive cooling profiles for an information processing system | |
| DE112011100143B4 (en) | Optimize electrical power consumption in a data center | |
| CN105929923B (en) | A kind of management method and device of power | |
| CN112506444A (en) | Kubernetes cluster-based expansion and contraction capacity control method and device and electronic equipment | |
| CN109800082B (en) | Method, device and storage medium for purchasing server by combining actual power consumption | |
| WO2019108649A1 (en) | Efficiency indexes | |
| US20150271023A1 (en) | Cloud estimator tool | |
| CN103795804A (en) | Storage resource scheduling method and storage calculation system | |
| US20120271678A1 (en) | Chargeback reduction planning for information technology management | |
| CN103645795A (en) | Cloud computing data center energy saving method based on ANN (artificial neural network) | |
| CN103399781A (en) | Cloud server and virtual machine management method thereof | |
| CN110619407B (en) | Object sales prediction method and system, electronic device and storage medium | |
| CN114442794A (en) | Server power consumption control method, system, terminal and storage medium | |
| CN112463372A (en) | Target system capacity expansion method, device, equipment and storage medium | |
| CN112702350B (en) | Out-of-band management method and device, electronic equipment and readable medium | |
| US20050075937A1 (en) | Rack equipment management information coordination system and method | |
| CN116647560A (en) | Method, device, equipment and medium for coordinated optimization control of Internet of things computer clusters | |
| CN110674150A (en) | BMC setting strategy updating method and equipment | |
| CN112734454B (en) | User information determining method and device, electronic equipment and storage medium | |
| CN106326001B (en) | Information processing method and electronic equipment | |
| US11586964B2 (en) | Device component management using deep learning techniques | |
| CN103873388A (en) | Network content control method and network equipment | |
| US8627314B2 (en) | Method for managing power of host computer |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |