JP5673167B2 - Power management method for electrical equipment - Google Patents

Description

  The present invention relates to a method for managing the amount of power consumed by a plurality of electrical devices.

  An example of this type of power management method is described in Patent Document 1. In the power management method described in Patent Document 1, a management device or a specific electric device centrally manages calculation of total power consumption of the entire system, selection of a device to be shifted to an energy saving mode, and transmission of a command to the device. Thus, control is performed so that the power consumption of the entire system does not exceed the threshold value.

JP 2010-246389 A

  In the power management method described in Patent Document 1, a special management device for managing the power of each electrical device is required, or when a specific electrical device also serves as a management device, the power management method is applied to the specific electrical device. Therefore, there is a problem that an extra load is applied.

  The object of the present invention is to manage the amount of power consumed by a plurality of electric devices as described above, that is, a special management device is required, or the load of power management is concentrated on a specific electric device. An object of the present invention is to provide a power management method for an electrical device that solves the problem.

An electric equipment power management method according to an aspect of the present invention includes:
A method for managing power consumption of a plurality of electrical devices,
When each electric device receives a data structure having an area for writing the power consumption of each electric device, the current power consumption of the own electric device is written in an area corresponding to the own electric device of the data structure, Passing the data structure to the electrical device and circulating the data structure;
In the above steps,
Calculate the total power consumption of the plurality of electrical devices from the current power consumption of the own electrical device and the power consumption of other electrical devices written in the data structure,
Compare the calculated total power consumption with the threshold,
A configuration is adopted in which the power consumption of the electric device is controlled based on the result of the comparison.

  Since the present invention has the configuration as described above, a special management device is not required when managing the amount of power consumed by a plurality of electric devices, and the load of power management is concentrated on specific electric devices. There is no.

It is a block diagram of a 1st embodiment of the present invention. It is a block diagram of the 2nd Embodiment of this invention. It is a block diagram of the hardware control part of each server in the 2nd Embodiment of this invention. It is a flowchart which shows operation | movement of the hardware control part of each server in the 2nd Embodiment of this invention. It is a figure which shows the specific example of the electric power monitoring method using the table in the 2nd Embodiment of this invention, and the control which suppresses electric power. It is a figure which shows the specific example of the electric power monitoring method using the table in the 2nd Embodiment of this invention, and the control which suppresses electric power. It is a figure which shows the specific example of the electric power monitoring method using the table in the 2nd Embodiment of this invention, and the control which suppresses electric power. It is a figure which shows the process until the power consumption of the whole server falls below a threshold value in the 2nd Embodiment of this invention. It is explanatory drawing of the 3rd Embodiment of this invention. It is a block diagram which shows the general structure of the system which operate | moves a some server with a shared power supply.

Next, embodiments of the present invention will be described in detail with reference to the drawings.
[First embodiment]
Referring to FIG. 1, the first embodiment of the present invention includes a plurality of electric devices U1 to Un. Each electric device U has a function of controlling the power consumption of its own electric device so that the total power consumed by all the electric devices U does not exceed a preset threshold in an autonomous and distributed manner.

  Each electric device U is an IT device, a home appliance, or the like. Each electric device U operates with electric power supplied from a power source (not shown).

  Each electrical device U includes communication means T1 to Tn that communicate with other electrical devices U. Communication by the communication means T may be wired or wireless. The plurality of electrical devices U have control units C1 to Cn that use the communication unit T to rotate the data structure D between the electrical devices U.

  The data structure D has an area in which the power consumption of each electrical device U is written. The circulation of the data structure D has an information sharing function that makes each electric device U recognize the power consumption of the other electric devices U.

  When the control means C of each electric device U receives the data structure D, it writes the current power consumption of the own device in the area corresponding to the own device of the data structure D, and the data structure is transferred to the other electric device U. Pass D and go around.

  When the control unit C receives the data structure D, the control unit C calculates the total of all the electric devices U from the current power consumption of the own device and the power consumption of the other electric devices U written in the data structure D. A function of calculating power consumption, and a function of comparing the calculated total power consumption with a threshold value; Furthermore, the control means C has a function of controlling the power consumption of the own device based on the comparison result. For example, the control means C has a function of switching its own device to an operation state with less power consumption if the total power consumption exceeds a threshold value.

  A method of switching the own device to an operation state with less power consumption is arbitrary. For example, when the electric device U is an IT device having a CPU, the power consumption may be reduced by reducing the frequency of the clock that drives the CPU. Further, when the electric device U has a cooling fan, the power consumption may be reduced by reducing the rotational speed of the fan.

  Next, the operation of this embodiment will be described.

  If the route for circulating the data structure D is the electrical equipment U1, the electrical equipment U2, the electrical equipment U3, ..., the electrical equipment Un, and the electrical equipment U1, the data structure D that has made one round returns to the starting electrical equipment U1. At that time, the power consumption of all the electric devices U is written in the data structure D.

  Upon receiving the data structure D, the control means C1 of the electric device U1 receives all the electric power from the current power consumption of the own device U1 and the power consumption of the other electric devices U2 to Un written in the data structure D. The total power consumption of the device U is calculated. And the control means C1 of the electric equipment U1 compares the total power consumption with a threshold value, and controls the power consumption of the own apparatus U1 based on the comparison result. For example, if the total power consumption exceeds a threshold value, the control unit C1 of the electrical device U1 switches the own device to an operation state with less power consumption in order to reduce the total power consumption. Then, the control means C1 of the electric device U1 writes the current power consumption of the own device in an area corresponding to the own device of the data structure D, and passes the data structure D to the electric device U2.

  The control unit C2 of the electrical device U2 that has received the data structure D performs the same operation as the control unit C1 of the electrical device U1. Thereby, if the total power consumption still exceeds the threshold, the control means C2 of the electric device U2 puts its own device into an operation state with less power consumption in order to reduce the total power consumption of all the electric devices. Switch. Then, the control unit C2 of the electric device U2 writes the current power consumption of the own device in an area corresponding to the own device of the data structure D, and passes the data structure D to the electric device U3.

  By repeating the same operation in the electric devices after the electric device U3, the power consumption of all the electric devices is reduced so as to be lower than the threshold value.

  As described above, according to the present embodiment, a special management device is not required when managing the amount of power consumed by a plurality of electrical devices U, and the load of power management is concentrated on specific electrical devices. Disappear.

  In addition, in the circulation of the data structure D, in addition to the information sharing function that allows each electrical device U to recognize the power consumption of the other electrical devices U, the electrical device that suppresses the power consumption is limited to one electrical device at a time. There is work. For this reason, it can prevent that the total power consumption of all the electric equipments fluctuates rapidly due to the control that suppresses the power consumption.

  Various additions and changes can be made in the present embodiment. For example, the control means C of the electric device U may have a function of switching the own device to an operation state with a higher electric power consumption as long as the total electric power consumption does not exceed the threshold value. The threshold value used at this time may be the same value as the threshold value when switching to an operation state with less power consumption, or may be a smaller value.

[Second Embodiment]
[Overview]
In the present embodiment, when a plurality of servers (information processing apparatuses) have a hardware configuration in which a power source is shared, each server can perform power control so as not to exceed the capacity of the shared power source. The information of the power consumption of each server can be shared by communication.

  Therefore, in this embodiment, each server has a table. In this table, there is an entry for recording the power consumption measurement values of all the servers including the own server. By referring to this, it can be understood which server consumes how much power.

  In addition, in order to share this power consumption information, a hardware control circuit such as a BMC (Base Management Controller) in the server transfers the information in the table to other servers by communication.

When the server receives this table information, the following processing is executed.
-Write to the table how much power the server currently consumes.
-Refer to the table and check the total power consumption of all servers.
・ Comparing the power consumption of all servers with the power threshold, and if the power consumption exceeds the threshold (= when there is no power remaining), power saving control is performed to reduce the power consumption.

  In the present embodiment, by performing the control as described above, power monitoring / control is dynamically performed, and operation with a small capacity power supply is enabled.

[Description of configuration]
FIG. 10 is a diagram illustrating a general configuration when a plurality of servers 1, 2, and 3 have a hardware configuration sharing a power supply 4. Power is supplied from a power facility such as a distribution board to the power supply 4, and power is supplied from the power supply 4 to the servers 1, 2, and 3 via the distribution cable 5. The servers 1, 2, and 3 have hardware control units 11, 12, and 13 therein. However, parts that are not directly related to the present invention, such as a CPU and a memory, are omitted. For example, when three servers (servers 1 to 3) having a maximum power consumption of 300 W share one power source 4, considering the case where all three units have the maximum load, the power source 4 having a power capacity of 900 W or more is used. It is common to select.

  FIG. 2 is a block diagram showing a configuration example of the second embodiment of the present invention. Servers 1, 2 and 3 share the power supply 6. The capacity of the power source 6 is 700 W, which is less than the total 900 W of the maximum power consumption of the three servers 1, 2, 3.

  The hardware control units 11, 12, and 13 in the server are connected by a signal line 7 for mutual communication, and can communicate with the hardware control units of other servers. For communication, a general communication means such as a LAN can be used. In the present embodiment, control is performed so that the total power consumption of all servers does not exceed 700 W while communicating with each server.

  FIG. 3 is a block diagram of the hardware control unit 11. The other hardware control units 12 and 13 have the same configuration. In FIG. 3, portions not directly related to the present invention are omitted.

  The hardware control unit 11 includes a processor 101, a program ROM 102, and a local memory 103. The hardware control unit 11 is assumed to be a controller that manages hardware in the server, such as BMC (Base Management Controller). Further, it is assumed that the server has a function of performing power saving control.

  The table TB used in the present embodiment is a data structure that can hold a server identification number and the power consumption of the server as a set, and is stored in the local memory 103. Furthermore, table information can be transmitted and received with other servers.

[Description of operation]
FIG. 4 is a flowchart showing the operation of this embodiment. 5 to 7 show specific examples of a power monitoring method using a table and control for suppressing power. The operation of this embodiment will be described below with reference to FIGS.

  First, the overall operation of this embodiment will be described with reference to FIG. The program ROM 102 in FIG. 3 stores a program that realizes the flow in FIG. 4, and the processor 101 reads the program and performs processing.

  First, it is confirmed whether or not the processor 101 is receiving power consumption table information (step S1). For example, when the processor 101 in the hardware control unit 11 of the server 1 receives table information from another server through the communication interface, the information is stored in the local memory 103. Only after this process is finished can the process proceed to the next process (step S2). Therefore, the servers 2 and 3 do not proceed to the next process (step S2) until the table information is received from another server.

  Next, the processor 101 writes the current power consumption of its own server in the table of the local memory 103 (step S2). When the processor 101 of the server 1 receives the table, the processor 101 confirms how much W the server itself is consuming from a power sensor (not shown) and writes the power consumption in the table of the local memory 103. . As shown in FIGS. 5 to 7, the table structure is such that the power consumption of the server 1 is xW, the power consumption of the server 2 is yW,... . By this processing, the power consumption information of the own server in the table is updated to the latest state.

  5 and 6 exemplify the state transition of the table. In state (1) (circled numbers in the drawing; the same applies hereinafter), the server 1 inputs the current power consumption into the table and passes the table information to the next server 2. In the state (2), similarly to the state (1), the server 2 inputs the current power consumption into the table and passes the table information to the next server 3. Similarly, in the state (3), the server 3 inputs the current power consumption into the table and passes the table information to the next server 1. In this way, the exchange of table information with (1), (2), (3),... Is repeated. After the second week, the power is overwritten.

  Next, the processor 101 confirms whether or not the power consumption of each server is held in the table of the local memory 103 (step S3). Immediately after starting the server, there is no information in the table, and it is not known how much W each server is currently consuming. By passing the information in the table to servers 1, 2, 3, 1 for one week, the above-described power consumption writing process (step S2) is performed on all servers, and the power consumption of each server is held in the table. It becomes a state. With the current technology, the processing time for writing a value to the table and transferring it to another server can be shortened to the millisecond unit, so that the power consumption value of each server in the table can be regarded as almost the current value.

  When the power consumption of each server is held in the table, the processor 101 refers to the table in the local memory 103 to calculate and calculate how much power is consumed by all the servers, and can be confirmed. (Step S4).

  Next, the processor 101 confirms whether the total power consumption of all the servers does not exceed the threshold (step S5). In this example, 680 W is set as the threshold value so that the total power consumption does not exceed 700 W of the power supply capacity. This threshold value is stored in the local memory 103, and the user can set the threshold value arbitrarily by accessing the hardware control unit 11.

  As a result of the confirmation, if the total power consumption does not exceed the threshold value, the processing of the own server ends, and the processor 101 passes table information to the next server through the communication interface (step S7).

  If the threshold value is exceeded, it means that there is no power supply capacity of the power supply 6, so the processor 101 suppresses power consumption with respect to the circuit having the power saving control function of the server (power saving level). A command is issued to perform control (step S6). Thereafter, the table information is passed to the next server (step S7).

  As described above, the server monitors its own power consumption and the power consumption of all servers, and performs power saving control independently when power is insufficient.

(About power saving level)
FIG. 7 illustrates a process from when the server 2 receives the table information and passes the table information to the next server 3.

  The table information is received from the server 1, and the current power consumption held in the table is updated. Although it was 200 W when it was written last time, 250 W is currently consumed, so the value of server 2 in the table is changed to 250 W. As a result, it is understood that the total power consumption of the servers 1, 2, and 3 is 700 W, which exceeds the threshold value of 680 W, that is, there is no power supply surplus capacity.

  The server 2 has a power saving function, and can prevent a sudden increase in power consumption and perform power ceiling control (upper limit control). In addition, the ceiling value can be set in multiple stages. In this embodiment, the power saving level is defined as an index indicating how much power the server is currently suppressing. As for the power saving control method of the server itself, any one or a combination of known techniques such as CPU clock frequency control and core control can be used as existing techniques.

  In FIG. 7, the server 2 is in a power saving level 3 state with a ceiling value of 260 W. However, since the total power consumption of the servers 1, 2, and 3 is 700 W and exceeds the threshold value, the server 2 performs control to increase the power saving level from 3 to 4 and further suppress the power consumption of the server 2 itself. In this specification, the expressions of increasing the power saving level when strengthening the suppression of power consumption and decreasing the power saving level when reducing the suppression are used.

(Process until the power consumption of the entire server falls below the threshold)
The amount of decrease in power consumption of the servers varies, and it is assumed that the total power consumption cannot be reduced to the threshold only by increasing the power saving level once by one server. Therefore, as shown in FIG. 8, the power consumption is continuously reduced step by step until each server becomes equal to or less than the threshold value.

  In addition, power consumption may exceed the power supply capacity due to a sudden increase in load, but because the table information is communicated and controlled at high speed, the excess time can be suppressed within a few seconds. There is no. The reason is that the power capacity generally has a margin, and even if the power consumption exceeds the power capacity, the power capacity is designed to withstand for a short time.

(Control to reduce power saving level)
In the present embodiment, the power saving level can be lowered, that is, the ceiling value of power consumption can be increased only when the total power consumption of all servers is below the threshold.

(Order of table information delivery)
Immediately after the system construction or start-up for the first time, the local memory 103 table of each server contains no information other than its own server information. Therefore, only in this case, the user needs to access the server hardware control unit and input information on the server that performs power monitoring and control. A user accesses the hardware control unit of one server, and inputs information (such as a physical address) that enables each server to identify and communicate. The server to which this data is input stores the server information in the table of the local memory 103 in the order of input. Delivery of table information starts from this server, and information is delivered in the order entered in the table below.

(Control when communication is impossible due to failure)
When a server becomes unable to communicate due to a failure or the like, the table cannot be transferred, and it is impossible to know how much W is consumed by the entire server. In this case, the operation is performed on the assumption that the server that cannot communicate is operating at the maximum power consumption. As a specific example, if the power supply is shared by three servers (one maximum power consumption 300W), it is assumed that one server that cannot communicate is operating at a maximum power consumption 300W, The power value of the failed server is fixed at 300 W, and the table is transferred with the remaining two units.

  As for the order of table information delivery, the failed server is skipped and the table information is delivered to the next server. For example, when transferring table information, an ACK is returned from the receiving server to the sending server, and whether or not the partner server has failed depending on whether or not the transmitting server has received the ACK. Determine whether. If ACK cannot be received, the partner server is determined to be a failed server, and table information is transmitted to the next server.

[Description of effects]
Servers are becoming smaller and more dense, and the number of servers that can be installed per floor area or rack is increasing. As an example, in a data center or the like, it is possible to install and operate a larger number of servers. However, as the number of servers increases, the power consumption of the entire data center increases, resulting in a problem of power shortage. As countermeasures, there is an increase in power facilities in the data center. However, since the cost is very high, countermeasures on the server side such as suppressing the power consumption of the server and improving the power efficiency are expected. According to the present embodiment, dynamic power consumption ceiling control that monitors power in real time is possible, and a power supply having a capacity smaller than the maximum power consumption of all servers can be used, thereby improving power efficiency. .

  It can be realized with a general electronic circuit without requiring a special control module for grasping the total power consumption and managing the power of each server.

[Other embodiments]
The present invention is not limited to the above embodiment, and various other additions and changes can be made. For example, as shown in FIG. 9A, when the total power consumption of all servers exceeds the threshold value T, the power consumption level is increased by suppressing the power consumption level. The threshold value T can be exceeded immediately. Thus, in order to avoid that the power consumption of all the servers continues to fluctuate around the threshold value T, a gap is provided between the threshold value when the suppression of the power consumption starts and when the power consumption is canceled. For example, as shown in FIG. 9B, an embodiment that prevents frequent switching of the control by setting the threshold Th for starting suppression and the threshold Tl for canceling suppression to different values is also included in the present invention. included.

  Further, for example, when a plurality of servers are mounted on a rack, there is a case where it is desired to set an upper limit on the power consumption of the entire rack according to the capacity of a facility (distribution panel) that supplies power to the rack. The present invention can also be applied to such applications. The power monitoring / control using the table is the same, and the threshold setting based on the power supply capacity is changed to the threshold setting based on the power supply capacity of the distribution board.

1-3 ... Server 5 ... Signal line 6 ... Power supply 11-13 ... Hardware control part

Claims (7)

A method for managing power consumption of a plurality of electrical devices,
When each electric device receives a data structure having an area for writing the power consumption of each electric device and the identification number of the electric device, the electric device has an area corresponding to the identification number of the own electric device in the data structure. Writing the current power consumption and transmitting the data structure to another electrical device having an identification number written in the data structure;
In the above step,
Calculate the total power consumption of the plurality of electrical devices from the current power consumption of the electrical device and the power consumption of other electrical devices written in the data structure,
Compare the calculated total power consumption with a threshold,
Based on the result of the comparison, control the power consumption of the own electrical equipment ,
Further, in the transmission of the data structure, when ACK cannot be received from the destination electrical device, the maximum power consumption of the electrical device is recorded in an area corresponding to the destination electrical device of the data structure, A power management method for an electric device, comprising: transmitting the data structure to another electric device having an identification number written in the data structure . 2. The electric device according to claim 1, wherein in controlling the power consumption of the electric device, the electric device is switched to an operation state with less electric power consumption if the total electric power consumption exceeds a threshold value. Power management method. The control of the power consumption of the self-electric device is characterized in that the self-electric device is switched to an operation state with a higher power consumption if the total power consumption does not exceed a threshold value. Power management method for electrical equipment. The power management of the electrical device according to any one of claims 1 to 3, wherein the plurality of electrical devices share a power source, and the threshold is set to a substantially maximum value of a power supply capability of the power source. Method. 5. The electric equipment power management method according to claim 1, wherein the electric equipment is an IT equipment. When a data structure having an area for writing the power consumption of each electric device and the identification number of the electric device is received, the current power consumption of the own electric device is stored in an area corresponding to the identification number of the own electric device of the data structure. Control means for writing and transmitting the data structure to another electrical device in which an identification number is written in the data structure;
The control means includes
Calculate the total power consumption of the plurality of electrical devices from the current power consumption of the electrical device and the power consumption of other electrical devices written in the data structure,
Compare the calculated total power consumption with a threshold,
Based on the result of the comparison, control the power consumption of the own electrical equipment ,
Further, in the transmission of the data structure, when ACK cannot be received from the destination electrical device, the maximum power consumption of the electrical device is recorded in an area corresponding to the destination electrical device of the data structure, An electric device characterized by transmitting the data structure to another electric device in which an identification number is written in the data structure . Computers that make up electrical equipment
Upon receiving the data structure having a region for writing the identification number of the power consumption and the electrical equipment of the electrical equipment, the current power consumption of the data structure itself electric device itself electrical device in the corresponding region to the identification number of the Control means for writing and transmitting the data structure to another electrical device ,
Calculate the total power consumption of the plurality of electrical devices from the current power consumption of the electrical device and the power consumption of other electrical devices written in the data structure,
Compare the calculated total power consumption with a threshold,
Based on the result of the comparison, control the power consumption of the own electrical equipment ,
Further, in the transmission of the data structure, when ACK cannot be received from the destination electrical device, the maximum power consumption of the electrical device is recorded in an area corresponding to the destination electrical device of the data structure, A program for causing a data structure to function as control means for transmitting the data structure to another electrical device having an identification number written in the data structure .

Images