Control system and control method

Abstract

A facility in which a plurality of server groups, each comprising of a plurality of servers arranged in a first horizontal direction, are installed at intervals in a second horizontal direction orthogonal to the first horizontal direction, includes a cooling system for cooling the servers, and a control device for controlling the servers and the cooling system, wherein: the control device includes a first allocating unit which, on the basis of schedule information of jobs for performing arithmetic processing on the servers and server information including the arrangement of the servers, allocates the jobs in a concentrated manner in order from the server group having the shortest flow path distance; and the opening degrees of open / closed valves provided in secondary side piping of the cooling system are changed in accordance with the processing loads of the servers.

Description

Control system and control method

[0001] This application claims priority to Japanese Patent Application No. 2024-010175, filed on January 26, 2024, the contents of which are incorporated herein by reference.

[0002] In data centers and the like where multiple servers are installed, there is a need to reduce the power consumption of the servers and the power consumption of the server cooling equipment. For example, Patent Document 1 describes a technology for allocating workloads in order, giving priority to information processing devices with high power efficiency, so that the workload processing volume falls within a threshold range for each information processing device.

[0003] Japanese Patent No. 4768082

[0004] When performing computational processing using multiple servers in a data center, etc., it is common to have some predefined servers perform the computational processing. However, simply allocating computational processing to predefined servers or servers with high power efficiency can make it difficult to minimize the power consumption of the entire data center, including the power required to cool the servers.

[0005] An object of the present disclosure is to provide a control system and a control method that can minimize the power consumption of the entire data center, including the cooling systems of the servers.

[0006] According to one aspect of the present disclosure, a control system is provided in a facility in which server groups each consisting of a plurality of servers arranged in a first horizontal direction are installed at intervals in a second horizontal direction perpendicular to the first horizontal direction, the control system including a cooling system for cooling the servers and a control device for controlling the servers and the cooling system, the cooling system including a cooling device for cooling a first-side refrigerant, a first-side piping provided with a first-side pump for delivering the first-side refrigerant cooled by the cooling device, a heat exchanger for exchanging heat between the first-side refrigerant cooled by the cooling device and a second-side refrigerant used to cool the servers, and a control device for delivering the second-side refrigerant cooled by the heat exchanger. The control device has a secondary side piping equipped with a pump, the secondary side piping having a flow path for each server group consisting of a main pipe, a plurality of first branch pipes branching from the main pipe at branch positions corresponding to each of the server groups, and a plurality of second branch pipes branching from the first branch pipe to each of the servers, and an on-off valve provided in each of the plurality of second branch pipes of the secondary side piping, and the control device has a first allocation unit that allocates the jobs in a concentrated manner in order from the server group with the shortest distance in the flow path based on schedule information of jobs to be processed by the servers and server information including the placement of the servers, and the opening degree of the on-off valve is changed according to the processing load of the servers.

[0007] According to one aspect of the present disclosure, a control method is a control method for controlling a cooling system that cools servers in a facility in which server groups each consisting of a plurality of servers arranged in a first horizontal direction are installed at intervals in a second horizontal direction perpendicular to the first horizontal direction, and the servers, the cooling system including a cooling device that cools a first-side refrigerant, a first-side pipe provided with a first-side pump that delivers the first-side refrigerant cooled by the cooling device, a heat exchanger that exchanges heat between the first-side refrigerant cooled by the cooling device and a second-side refrigerant used to cool the servers, and a second-side pump that delivers the second-side refrigerant cooled by the heat exchanger. the secondary-side piping having a flow path for each server group, the flow path being composed of a main pipe, a plurality of first branch pipes branching from the main pipe at branch positions corresponding to each of the server groups, and a plurality of second branch pipes branching from the first branch pipe to each of the servers; and an on-off valve provided in each of the plurality of second branch pipes of the secondary-side piping, and the control method comprises the steps of: concentrating and allocating the jobs in order from the server group having the shortest distance of the flow path based on schedule information of jobs that perform calculation processing on the servers and server information including the placement of the servers; and changing the opening degree of the on-off valve in accordance with the processing load of the servers.

[0008] According to the above aspect, it is possible to minimize the power consumption of the entire data center including the cooling system of the servers.

[0009] 1 is a schematic diagram showing the overall configuration of a control system according to a first embodiment. FIG. 2 is a diagram showing the configuration of a secondary-side piping according to the first embodiment. FIG. 3 is a block diagram showing the functional configuration of a control device according to the first embodiment. FIG. 4 is a flowchart showing an example of processing by the control device according to the first embodiment. FIG. 5 is a block diagram showing the functional configuration of a control device according to a second embodiment. FIG. 6 is a flowchart showing an example of processing by the control device according to the second embodiment. FIG. 7 is a graph showing an example of a relationship between a secondary-side refrigerant flow rate and secondary-side pump power according to the second embodiment. FIG. 8 is a graph showing an example of a relationship between a secondary-side refrigerant flow rate and power of a cooling device according to the second embodiment. FIG. 9 is a graph showing an example of a relationship between a secondary-side refrigerant flow rate and total power according to the second embodiment. FIG. 10 is a diagram showing an example of a distribution of a processing load according to the second embodiment. FIG. 11 is a diagram showing an example of a distribution of a processing load according to the third embodiment.

[0010] First Embodiment Hereinafter, a first embodiment will be described in detail with reference to FIGS.

[0011] (Overall Configuration) Fig. 1 is a schematic diagram showing the overall configuration of a control system according to Embodiment 1. As shown in Fig. 1, the control system 1 includes a cooling system 10 and a control device 20.

[0012] The cooling system 10 includes a cooling device 11, a primary pump 13, a heat exchanger 14, a primary piping 15, a secondary pump 16, and a secondary piping 17. The cooling system 10 is a system that cools a server 18 installed in, for example, a data center. The server 18 has a processor (heat generating element 19) such as a CPU or GPU. In this embodiment, an example will be described in which the heat generating element 19 is a CPU.

[0013] The cooling device 11 is, for example, a free-cooling cooling tower, an air-cooled heat pump chiller, etc. The cooling device 11 cools a primary refrigerant (for example, water) flowing through a primary pipe 15. The cooling device 11 includes a fan 12.

[0014] The primary pump 13 is provided in the primary pipe 15, and sucks in the primary refrigerant cooled by the cooling device 11 and sends it to the heat exchanger 14. The primary pipe 15 is a pipe through which the primary refrigerant flows, and is provided so that the primary refrigerant can circulate between the cooling device 11 and the heat exchanger 14.

[0015] The secondary piping 17 is a piping through which the secondary refrigerant flows, and is provided so that the secondary refrigerant can circulate between the heat exchanger 14 and the server 18 .

[0016] The heat exchanger 14 exchanges heat between the primary refrigerant and the secondary refrigerant. In the heat exchanger 14, the secondary refrigerant is cooled by the primary refrigerant.

[0017] The secondary refrigerant cools the CPU 19 by removing heat from the CPU 19 , and is cooled by the primary refrigerant in the heat exchanger 14 .

[0018] FIG. 2 is a diagram illustrating the configuration of secondary piping according to the first embodiment. As shown in FIG. 2, the secondary piping 17 circulates a secondary refrigerant through each of multiple servers 18 installed in a facility R, such as a data center, to cool each server 18. In the facility R, multiple server groups G, each consisting of multiple servers 18 arranged in a first horizontal direction (the left-right direction on the paper), are installed at intervals in a second horizontal direction (the up-down direction on the paper) perpendicular to the first horizontal direction. While FIG. 2 illustrates an example in which four server groups G, groups A to D, are installed, the number of server groups may be more or less than this. Furthermore, while FIG. 2 illustrates an example in which each server group G has 15 servers 18, the number of servers may be more or less than this.

[0019] The secondary piping 17 includes a main pipe 171 , a first branch pipe 172 , and a second branch pipe 173 .

[0020] The header pipe 171 is made up of a low-temperature side header pipe 171a and a high-temperature side header pipe 171b extending in the first horizontal direction. The low-temperature side header pipe 171a is a flow path through which the secondary refrigerant (chilled water) cooled by the heat exchanger 14 is supplied, and the high-temperature side header pipe 171b is a flow path through which the secondary refrigerant (hot water) heated by heat taken from the server 18 (CPU 19) is discharged.

[0021] The first branch pipes 172 branch off in the second horizontal direction from the header 171 at branch positions P1 to P4 on the low-temperature side header 171a, which correspond to the respective server groups G (groups A to D). The first branch pipes 172 also merge with the header 171 at merge positions P5 to P8 on the high-temperature side header 171b, which correspond to the respective server groups G (groups A to D). The first branch pipes 172 consist of a low-temperature side first branch pipe 172a and a high-temperature side first branch pipe 172b. The low-temperature side first branch pipe 172a is a flow path through which the secondary refrigerant (chilled water) cooled by the heat exchanger 14 is supplied, and the high-temperature side first branch pipe 172b is a flow path through which the secondary refrigerant (hot water) heated by heat removed from the server 18 (CPU 19) is discharged.

[0022] The second branch pipe 173 branches off from the first low-temperature branch pipe 172a, passes through the server 18, and then merges with the first high-temperature branch pipe 172b. The second branch pipe 173 consists of a second low-temperature branch pipe 173a and a second high-temperature branch pipe 173b. The second low-temperature branch pipe 173a is a flow path to which secondary refrigerant (chilled water) cooled in the heat exchanger 14 is supplied, and the second high-temperature branch pipe 173b is a flow path to which secondary refrigerant (hot water) warmed by heat absorbed from the server 18 (CPU 19) is discharged. The second low-temperature branch pipe 173a and the second high-temperature branch pipe 173b are provided with on-off valves 174 and 175, respectively.

[0023] The on-off valves 174 and 175 adjust the flow rate of the secondary-side refrigerant to the server 18 by changing their opening degrees in accordance with the processing load of the server 18. In this embodiment, the on-off valves 174 and 175 are automatic valves that automatically change their opening degrees in accordance with the temperature difference between the secondary-side refrigerant on the inlet side (low-temperature side second branch pipe 173 a) of the server 18 and the secondary-side refrigerant on the outlet side (high-temperature side second branch pipe 173 b) of the server 18. When the temperature difference is within a predetermined range (close to zero), i.e., when the processing load of the server 18 (CPU 19) is low and no heat is generated, the on-off valves 174 and 175 are closed and stop the supply of the secondary-side refrigerant to the server 18. On the other hand, the on-off valves 174 and 175 increase their opening degrees as the temperature difference increases, thereby increasing the amount of secondary-side refrigerant supplied to the server 18. This prevents the server 18 from being supplied with too much or too little secondary-side refrigerant.

[0024] The control device 20 adjusts the temperature of the primary refrigerant at the inlet side of the heat exchanger 14 by controlling the operation of the cooling device 11, adjusts the flow rate of the primary refrigerant flowing through the primary piping 15 by controlling the rotation speed of the primary pump 13, and adjusts the flow rate of the secondary refrigerant flowing through the secondary piping 17 by controlling the rotation speed of the secondary pump 16, thereby controlling the temperature of the CPU 19 to a desired range (below a predetermined allowable temperature). The control device 20 also allocates jobs to each server 18.

[0025] (Functional configuration of control device) Fig. 3 is a block diagram showing the functional configuration of the control device according to embodiment 1. As shown in Fig. 3, the control device 20 includes a processor 21, a memory 22, a storage 23, a communication interface 24, a display unit 25, and an operation unit 26.

[0026] The processor 21 operates in accordance with a predetermined program to function as an acquisition unit 211, a first allocation unit 212, a heat generation amount estimation unit 213, a power calculation unit 214, and a control unit 215.

[0027] The acquisition unit 211 acquires schedule information for jobs that will perform computational processing on the servers 18 and server information for the servers 18. The job schedule information includes information on jobs to be executed in a predetermined period, etc. The server information includes information on the location of each server 18, etc.

[0028] The first allocation unit 212 allocates jobs in a concentrated manner, starting with the server group G having the shortest path distance, based on the job schedule information and server information.

[0029] The heat generation amount estimation unit 213 estimates the heat generation amount of each server 18 (CPU 19) based on the processing load of each server 18 and information (table, function, etc.) that defines the relationship between the processing load of the CPU 19 and the heat generation amount.

[0030] The power calculation unit 214 determines the operating parameters of the cooling device 11, the primary pump 13, and the secondary pump 16, such as the rotation speed of the fan 12 (if the cooling device 11 is free cooling), the rotation speed of the compressor (not shown) (if the cooling device 11 is a heat pump chiller), and the rotation speeds of the primary pump 13 and the secondary pump 16, based on the estimated heat generation amount of the CPU 19 and the allowable temperature of the CPU 19.

[0031] The control unit 215 controls the cooling device 11 , the primary pump 13 , and the secondary pump 16 based on the operating parameters determined by the power calculation unit 214 .

[0032] The predetermined program executed by the processor 21 is stored on a computer-readable recording medium. Computer-readable recording media include magnetic disks, magneto-optical disks, CD-ROMs, DVD-ROMs, and semiconductor memories. The computer program may be distributed to a computer via a communication line, and the computer that receives the distribution may execute the program. The program may also be a program for implementing some of the above-described functions. Furthermore, the program may be a so-called differential file (differential program) that can realize the above-described functions in combination with a program already stored in the computer system.

[0033] Furthermore, in the present embodiment, an example has been described in which the on-off valves 174, 175 are automatic valves, but the present invention is not limited to this. In other embodiments, the on-off valves 174, 175 may be controlled by the control unit 215. For example, the control unit 215 adjusts the opening degree of each on-off valve 174, 175 in response to an operation by a data center administrator (user). Furthermore, the control unit 215 may automatically adjust the opening degree of each on-off valve 174, 175 in accordance with the heat generation amount of the server (CPU 19) estimated by the heat generation amount estimation unit 213.

[0034] The memory 22 has a memory area necessary for the operation of the processor 21 .

[0035] The storage 23 is a so-called auxiliary storage device, such as a hard disk drive (HDD), a solid state drive (SSD), etc. The storage 23 stores data that each part of the processor 21 acquires, generates, and references during processing.

[0036] The communication interface 24 is an interface for transmitting and receiving data, control signals, measurement values, etc. between each device and various sensors of the cooling system 10.

[0037] The display unit 25 has a display screen such as a liquid crystal display, and displays text, images, etc. represented by given data on the display screen.

[0038] The operation unit 26 includes input devices such as a mouse and a keyboard, receives operations from the administrator (user) of the data center, and outputs an operation signal corresponding to the received operation.

[0039] (Processing Flow) Fig. 4 is a flowchart showing an example of processing by the control device according to the first embodiment. The processing flow of the control device according to this embodiment will be described in detail below with reference to Fig. 4 .

[0040] First, the acquisition unit 211 acquires job schedule information and server information (step S101). For example, information about jobs submitted by clients (job names, start times, end times, etc.) is recorded in the storage 23. The acquisition unit 211 acquires schedule information including information about jobs to be executed within a predetermined period from the job information recorded in the storage 23. The predetermined period is, for example, a future period of 30 minutes starting 5 minutes after the current time, and may be set arbitrarily by the administrator (user) of the data center.

[0041] FIG. 5 is a diagram illustrating an example of server information according to the first embodiment. The storage 23 pre-stores server information such as that shown in FIG. 5 . The server information includes identification information (e.g., group ID) of the server group G, the flow path distance of the server group, identification information (e.g., server ID) of the servers 18 belonging to each server group G, the allowable temperature of each server 18, and the upper load limit. The server ID includes, for example, the group ID of the server group G to which the server belongs and a number indicating the server's position within the group. In the example of FIG. 2 , the server ID of the first server 18 in group A (the server 18 connected to the low-temperature-side second branch pipe 173a, which branches off from the most upstream side of the low-temperature-side first branch pipe 172a) is "A1." In other words, the server ID identifies the location of each server 18. The flow path distance of the server group is the flow path distance of the server group G, which is composed of the main pipe 171 and the first branch pipe 172, and is based on the design data of the cooling system 10, actual measurements, and the like. For example, the flow path distance of group A is the sum of the distance from the starting point of the low-temperature side main pipe 171a (e.g., the inlet of facility R, the outlet of heat exchanger 14, etc.) to the branch position P1 to group A, the distance of the first branch pipe 172 (the low-temperature side first branch pipe 172a and the high-temperature side first branch pipe 172b), and the distance from the junction position P5 with the high-temperature side main pipe 171b to the end point of the high-temperature side main pipe 171b (e.g., the outlet of facility R, the inlet of heat exchanger 14, etc.). The load upper limit value is the upper limit of the CPU occupancy rate or the maximum number of jobs that can be submitted.

[0042] Next, the first allocation unit 212 allocates jobs included in the job schedule information so that the processing load is concentrated on the servers in server group G, which has the shortest path distance (step S102). In the examples of FIGS. 2 and 5, the server group G, which has the shortest path distance, is group A. Therefore, the first allocation unit 212 allocates jobs to each server 18 (A1 to A15) in group D up to the load upper limit. Furthermore, if each server 18 (A1 to A15) in group A has already been allocated up to the load upper limit and there are still unallocated jobs, the first allocation unit 212 allocates jobs to a server in server group G, which has the next shortest path distance. In the examples of FIGS. 2 and 5, group B has the next shortest path distance after group A. Therefore, if the servers 18 (A1 to A15) in group A cannot process all of the jobs, the first allocation unit 212 allocates jobs to the servers 18 (B1 to B15) in group B.

[0043] Fig. 6 is a diagram showing an example of the distribution of processing loads according to the first embodiment. Assume that in step S102, the first allocation unit 212 allocates all jobs intensively to the servers 18 (A1 to A15) of group A, as shown in Fig. 6, and does not allocate jobs to the servers 18 of other groups B to D. The first allocation unit 212 may display a diagram showing the distribution of the processing loads (CPU occupancy rates) of the servers 18 on the display unit 25 using, for example, shading, color, text information, etc., so that the user can check it.

[0044] Next, the operating parameters of each component of the cooling system 10 (the cooling device 11, the primary pump 13, and the secondary pump 16) are determined, and the cooling system 10 is controlled (step S103). For example, the heat generation amount estimation unit 213 calculates the heat generation amount of the server 18 (CPU 19) based on the processing load of the server 18. The heat generation amount estimation unit 213 also acquires the allowable temperature of the server 18 from the server information recorded in the storage 23. The power calculation unit 214 determines operating parameters of the cooling device 11, the primary pump 13, and the secondary pump 16 based on the estimated heat generation amount of the CPU 19 and the allowable temperature of the CPU 19. Known techniques may be used to determine the operating parameters. The control unit 215 then controls the cooling device 11, the primary pump 13, and the secondary pump 16 based on the operating parameters determined by the power calculation unit 214.

[0045] The on-off valves 174, 175 of each server group G are automatically switched between open and closed states (step S103). For example, when the opening degree is automatically changed based on the temperature difference between the inlet and outlet of the server 18, the servers 18 (A1 to A15) of group A have a high processing load and a large temperature difference, so the on-off valves 174, 175 are open. On the other hand, the servers of groups B to D have a low processing load and a temperature difference close to zero, so the on-off valves 174, 175 are closed. In another embodiment, a user may issue an instruction to open or close the on-off valves 174, 175 through the operation unit 26 of the control device 20 while checking the processing load distribution displayed on the display unit 25. In yet another embodiment, the control unit 215 may control the on-off valves 174, 175 to be open or closed based on the temperature difference between the inlet and outlet of the server 18.

[0046] The control device 20 repeatedly executes the series of processes shown in FIG. 4 at predetermined intervals.

[0047] As described above, in the control system 1 according to this embodiment, the control device 20 has the first allocation unit 212 that allocates jobs in a concentrated manner, starting with the server group G having the shortest flow path distance, based on the job schedule information and server information. In addition, the opening degrees of the on-off valves 174 and 175 provided in the second branch pipe 173 of the secondary piping 17 are changed in accordance with the processing load of the server 18.

[0048] In conventional technology, jobs were assigned to servers designated by clients, which distributed the processing load across groups A to D and required the secondary refrigerant to circulate through all groups A to D. In contrast, in this embodiment, the control device 20 concentrates the processing load on group A, which has a shorter flow path distance, allowing the secondary refrigerant to circulate only through group A. This shortens the distance the secondary refrigerant must flow, thereby shortening the secondary refrigerant circulation path compared to circulating the secondary refrigerant to group D. This reduces pressure loss in the main pipe 171, assuming the same flow rate of the secondary refrigerant, and reduces the power (power consumption) of the secondary pump 16. This minimizes the power consumption of the entire data center, including the cooling system 10 for the servers 18.

[0049] The opening degrees of the on-off valves 174 and 175 are automatically changed based on the temperature difference of the secondary-side refrigerant at the inlet and outlet of the server 18 of the second branch pipe 173 .

[0050] In this way, for servers 18 that have no job assignments (processing loads) and have small temperature differences, the on-off valves 174 and 175 can be automatically closed to stop the flow of secondary refrigerant without the user having to monitor or operate them.

[0051] In addition, the control device 20 further includes a control unit 215 that controls the change in opening degree based on a user's operation instruction or based on the temperature difference of the secondary side refrigerant at the inlet and outlet of the server 18 of the second branch pipe 173.

[0052] By doing this, even if on-off valves 174, 175 that are not automatic valves are installed, for servers 18 that have no job assignments (processing load) and have a small temperature difference, the on-off valves 174, 175 can be closed by user operation or by automatic control of the control unit 215 to stop the flow of secondary side refrigerant.

[0053] Second Embodiment A second embodiment will be described below with reference to Figures 7 to 13. Note that, among the configurations of the second embodiment, the same configurations as those of the first embodiment will be described using the same reference numerals as those of the first embodiment.

[0054] (Functional configuration of control device) Fig. 7 is a block diagram showing the functional configuration of a control device according to the second embodiment. As shown in Fig. 7, the control device 20 according to this embodiment further includes a second assigning unit 216 and a determining unit 217. In addition, the function of the power calculating unit 214 differs from that of the first embodiment.

[0055] The second allocation unit 216 allocates jobs to all servers 18 in all server groups G based on the job schedule information and server information so as to distribute the processing loads equally among them. In this embodiment, "uniform processing loads" means that the difference in processing load (CPU occupancy rate) is within a predetermined value (within ±X%).

[0056] The power calculation unit 214 calculates a first predicted value of the power of the cooling system 10 according to the job allocation of the first allocation unit 212 and a second predicted value of the power of the cooling system 10 according to the job allocation of the second allocation unit 216 based on information indicating the relationship between the allowable temperature of the server 18, the heat generation amount, the flow rate of the primary side refrigerant and the secondary side refrigerant, and the total power of the cooling system 10 for supplying the primary side refrigerant and the secondary side refrigerant to the server 18.

[0057] The determination unit 217 determines whether to adopt the job allocation of the first allocation unit 212 or the second allocation unit 216 based on the first predicted value and the second predicted value.

[0058] (Processing Flow) Fig. 8 is a flowchart showing an example of processing by the control device according to the second embodiment. Hereinafter, the processing flow of the control device in this embodiment will be described in detail with reference to Fig. 8 .

[0059] First, the acquisition unit 211 acquires the job schedule information and server information (step S201). This process is the same as step S101 in the first embodiment (FIG. 4).

[0060] Next, a first process of steps S202 to S206 and a second process of steps S207 to S211 are executed in parallel. Note that in other embodiments, the first process and the second process may be executed in order.

[0061] First, the first process (steps S202 to S206) will be described. The first allocation unit 212 allocates jobs included in the job schedule information so that the processing load is concentrated on the servers in the server group G with the shortest path distance (step S202). This process is the same as step S102 in the first embodiment (FIG. 4).

[0062] Next, the power calculation unit 214 sets the flow rate of the primary refrigerant (step S203). The flow rate of the primary refrigerant may be set by the user via the operation unit 26, or a pre-recorded setting value may be read from the storage 23.

[0063] The heat generation amount estimation unit 213 also checks the constraint conditions of the heat generating element (step S204). For example, the heat generation amount estimation unit 213 calculates the heat generation amount of the server 18 based on the processing load of the server 18 (CPU 19). The heat generation amount estimation unit 213 also obtains the allowable temperature of the server 18 from the server information recorded in the storage 23. In the processing from step S205 onwards, the primary-side refrigerant flow rate set in step S203 and the heat generation amount and allowable temperature of the server 18 checked in step S204 are used as constraint conditions, and an optimal operating point under these constraint conditions is determined.

[0064] Next, the power calculation unit 214 determines the secondary-side refrigerant flow rate that satisfies the constraint conditions (step S205).

[0065] 9 is a graph showing an example of the relationship between the secondary-side refrigerant flow rate and the secondary-side pump power according to the second embodiment. As shown in FIG. 9, increasing the secondary-side refrigerant flow rate also increases the power of the secondary-side pump 16. According to the graph in FIG. 9, the power of the secondary-side pump 16 can be calculated by determining the secondary-side refrigerant flow rate.

[0066] 10 is a graph showing an example of the relationship between the secondary-side refrigerant flow rate and the power of the cooling device according to the second embodiment. The graph in FIG. 10 shows the relationship between the secondary-side refrigerant flow rate and the power of the cooling device 11, assuming that the primary-side refrigerant flow rate and the heat generation amount of the server 18 (CPU 19) are constant at certain values. As shown in the figure, increasing the secondary-side refrigerant flow rate can reduce the power of the primary-side cooling device 11. When combined with the graph in FIG. 9 , the power of the secondary-side pump 16 and the power of the cooling device 11 can be determined according to the secondary-side refrigerant flow rate.

[0067] FIG. 11 is a graph showing an example of the relationship between the secondary-side refrigerant flow rate and the total power according to the second embodiment. The graph in FIG. 11 shows the relationship between the secondary-side refrigerant flow rate and the total power, assuming that the primary-side refrigerant flow rate and the heat generation amount of the server 18 (CPU 19) are constant at certain values. The total power is the sum of the power of the cooling device 11, the power of the primary-side pump 13, and the power of the secondary-side pump 16. In combination with the graphs in FIGS. 9 and 10, the power of the cooling device 11, the power of the primary-side pump 13, and the power of the secondary-side pump 16 can be determined by determining the secondary-side refrigerant flow rate. Here, if an optimal PUE (Power Usage Effectiveness) value that minimizes the total power can be selected and the cooling device 11, the primary-side pump 13, and the secondary-side pump 16 can be operated at this operating point, the cooling system 10 can be operated with minimal power (energy consumption), which can contribute to improving the energy efficiency of the data center.

[0068] For example, the storage 23 stores a graph (e.g., FIG. 11 ) showing the relationship between the secondary-side refrigerant flow rate and the total power for each combination of the primary-side refrigerant flow rate set in step S105, the heat generation amount of the server 18 calculated in step S204, and the allowable temperature of the server 18. Based on this graph, the power calculation unit 214 calculates the secondary-side refrigerant flow rate that minimizes the total power, and determines the calculated value as the secondary-side refrigerant flow rate of the cooling system 10 corresponding to the job allocation by the first allocation unit 212 (step S205).

[0069] Once the secondary-side refrigerant flow rate is determined, the power (rotation speed) of the secondary-side pump 16 is determined. For example, the power calculation unit 214 calculates the power of the secondary-side pump 16 based on the graph of FIG. 9 and the secondary-side refrigerant flow rate, and converts the power of the secondary-side pump 16 into a rotation speed. The power calculation unit 214 also determines the temperature of the secondary-side refrigerant required to cool the server 18 to a temperature equal to or lower than the allowable temperature based on the heat generation amount, allowable temperature, and secondary-side refrigerant flow rate of the server 18. For example, the storage 23 stores a table that defines the correspondence between the heat generation amount of the server 18, the allowable temperature of the server 18, the secondary-side refrigerant flow rate, and the secondary-side refrigerant temperature (the temperature of the secondary-side refrigerant at the inlet side of the server 18). The power calculation unit 214 determines the secondary-side refrigerant temperature corresponding to the job allocation of the first allocation unit 212 based on the heat generation amount and allowable temperature estimated in step S204, the secondary-side refrigerant flow rate determined in step S205, and this table. When the secondary-side refrigerant temperature is determined, the temperature of the primary-side refrigerant on the inlet side of the heat exchanger 14 is also determined because the primary-side refrigerant flow rate has been set previously. For example, a table that defines the correspondence between the primary-side refrigerant flow rate, the primary-side refrigerant temperature, the secondary-side refrigerant flow rate, and the secondary-side refrigerant temperature is registered in the storage 23. The power calculation unit 214 calculates the primary-side refrigerant temperature corresponding to the job allocation of the first allocation unit 212 based on this table and the previously determined primary-side refrigerant flow rate, secondary-side refrigerant flow rate, and refrigerant temperature.

[0070] The inlet and outlet temperatures of the primary side of the heat exchanger 14 are determined under the above conditions, and the power (first predicted value) of the cooling device 11 corresponding to the job allocation by the first allocation unit 212 is determined (step S206). The power calculation unit 214 calculates the outlet temperature of the primary refrigerant of the heat exchanger 14 using a predetermined formula based on the primary refrigerant temperature, secondary refrigerant flow rate, and temperature at the inlet side of the heat exchanger 14, and calculates the power (power consumption) of the cooling device 11 from the primary refrigerant temperatures at the inlet and outlet of the heat exchanger 14. For example, if the cooling device 11 is a free-cooling cooling tower, the power calculation unit 214 calculates the power (power consumption) of the fan 12. For example, if the cooling device 11 is an air-cooled heat pump chiller, the power calculation unit 214 calculates the power (power consumption) of the fan 12, a compressor (not shown), and other components. Furthermore, in the case where the cooling device 11 has both a free-cooling cooling tower and an air-cooled heat pump chiller, if the constraint conditions are satisfied with only the free-cooling cooling tower, the power calculation unit 214 calculates only the power for the cooling tower fan 12. On the other hand, if the constraint conditions cannot be satisfied without operating both the free-cooling cooling tower and the air-cooled heat pump chiller, the power calculation unit 214 calculates the total value of the power for the cooling tower fan 12 and the power for the heat pump chiller fan 12 and a compressor (not shown), etc.

[0071] Next, the second process (steps S207 to S211) will be described. The second allocation unit 216 allocates jobs included in the job schedule information so that the processing loads of all servers in all server groups G are approximately equal (step S207).

[0072] 12 is a diagram showing an example of a distribution of processing loads according to the second embodiment. In step S207, the second allocation unit 216 allocates jobs to the servers 18 in all groups A to D so as to distribute the jobs approximately evenly, as shown in FIG.

[0073] The processes in steps S208 to S211 are the same as those in steps S203 to S206. In step S211, the power calculation unit 214 determines the power (second predicted value) of the cooling device 11 corresponding to the job allocation by the second allocation unit 216.

[0074] Next, the determining unit 217 compares the first predicted value with the second predicted value (step S212).

[0075] 13 is a diagram illustrating an example of the first predicted value and the second predicted value according to the second embodiment. For example, as in the example of FIG. 13 , if the second predicted value is smaller than the first predicted value (step S212; NO), the determination unit 217 adopts the job allocation (e.g., FIG. 12 ) of the second allocation unit 216 (step S214). The control unit 215 also controls each component of the cooling system 10 based on the second predicted value. For example, the control unit 215 controls the cooling device 11, the primary pump 13, and the secondary pump 16 based on values ​​obtained by converting the power of the cooling device 11, the primary pump 13, and the secondary pump 16 determined in step S211 into the rotation speed of the fan 12 (or / and the compressor), the rotation speed of the primary pump 13, and the rotation speed of the secondary pump 16.

[0076] Furthermore, if the first predicted value is smaller than the second predicted value (step S212; YES), the determination unit 217 adopts the job allocation (e.g., FIG. 6 ) of the first allocation unit 212 (step S213). Furthermore, the control unit 215 controls each unit of the cooling system 10 based on the first predicted value. For example, the control unit 215 controls the cooling device 11, the primary pump 13, and the secondary pump 16 based on values ​​obtained by converting the power of the cooling device 11, the primary pump 13, and the secondary pump 16 determined in step S206 into the rotation speed of the fan 12 (or / and the rotation speed of the compressor), the rotation speed of the primary pump 13, and the rotation speed of the secondary pump 16.

[0077] The control device 20 repeatedly executes the series of processes shown in FIG. 4 at predetermined intervals.

[0078] (Actions and Effects) As described above, in the control system 1 according to the present embodiment, the control device 20 further includes: a second allocation unit 216 that allocates jobs to all of the servers 18 in all of the server groups G so as to distribute the processing loads evenly based on the schedule information and the server information; a power calculation unit 214 that calculates a first predicted value of the power of the cooling system 10 according to the job allocation of the first allocation unit 212 and a second predicted value of the power of the cooling system 10 according to the job allocation of the second allocation unit 216 based on information indicating the relationship between the allowable temperature, heat generation amount, flow rate of the primary-side refrigerant and the secondary-side refrigerant of the servers 18, and the total power of the cooling system 10 for supplying the primary-side refrigerant and the secondary-side refrigerant to the servers 18; and a determination unit 217 that determines whether the job allocation of the first allocation unit 212 or the second allocation unit 216 is to be adopted based on the first predicted value and the second predicted value.

[0079] In this way, the control device 20 can select and control conditions (job allocation) that allow for further power reduction, thereby making it possible to further effectively reduce the power consumption of the cooling system 10.

[0080] Furthermore, for example, in a configuration in which the cooling device 11 has both a free-cooling cooling tower and an air-cooled heat pump chiller, it may be possible to satisfy the constraints by operating only the free-cooling cooling tower without using the air-cooled heat pump, depending on the conditions. For example, in the job allocation of the second allocation unit 216 (allocating jobs approximately evenly to all servers 18), the thermal load per server is low, so it is possible to sufficiently cool each server 18 even if the refrigerant temperature is relatively high. Furthermore, during periods of low outside temperature, such as winter, a sufficient cooling effect may be obtained using only the free-cooling cooling tower. In such cases, stopping the air-cooled heat pump chiller and operating only the free-cooling cooling tower, which requires less power, allows the servers 18 to be cooled, further reducing power consumption.

[0081] Third Embodiment A second embodiment will be described below with reference to Fig. 13. Note that, among the configurations of the second embodiment, the same configurations as those of the first embodiment will be described using the same reference numerals as those of the first embodiment.

[0082] The facility R according to this embodiment has a plurality of floors, and a plurality of server groups G are set on each floor.

[0083] 4 or step S202 in FIG. 8, the first allocation unit 212 according to this embodiment allocates jobs in a concentrated manner to server groups G on some floors so as to minimize the number of operating floors. The first allocation unit 212 allocates jobs sequentially, starting from floor 1, for example, until the processing load of each server 18 in each server group G reaches its upper load limit. Furthermore, if floor 1 alone is insufficient, the first allocation unit 212 allocates jobs until the processing load of each server 18 on floor 2 reaches its upper load limit.

[0084] FIG. 14 is a first diagram illustrating an example of a processing load distribution according to the third embodiment. FIG. 15 is a second diagram illustrating an example of a processing load distribution according to the third embodiment. FIG. 14 illustrates an example in which the second allocation unit 216 allocates jobs approximately evenly to all servers 18 on all floors. Meanwhile, FIG. 15 illustrates an example in which the first allocation unit 212 according to this embodiment allocates jobs in a concentrated manner to server groups G on some floors (floors 1 and 2) so as to minimize the number of floors in operation. As shown in FIG. 14 , the allocation method of the second allocation unit 216 requires three floors to be operated, which increases the power consumption of the cooling system 10. On the other hand, as shown in FIG. 15 , the allocation method of the first allocation unit 212 requires only a minimum number of floors (two) to be operated, so the flow of secondary refrigerant can be stopped for floor 3, which is not in operation (no jobs are allocated). Therefore, since cooling of floor 3 is not required, the power consumption of the entire facility R may be reduced. Furthermore, since it becomes possible to submit jobs from other clients to the server 18 on floor 3, the data center side can handle jobs from more clients.

[0085] <Other Embodiments> Although one embodiment has been described in detail above with reference to the drawings, the specific configuration is not limited to the above, and various design modifications are possible. That is, in other embodiments, the order of the above-described processes may be changed as appropriate. Furthermore, some processes may be executed in parallel.

[0086] <Additional Notes> The control system and control method described in the above-described embodiment can be understood, for example, as follows.

[0087] (1) According to a first aspect, a control system 1 is provided in a facility R in which a plurality of server groups G, each consisting of a plurality of servers 18 arranged in a first horizontal direction, are installed at intervals in a second horizontal direction perpendicular to the first horizontal direction. The control system 1 includes a cooling system 10 that cools the servers 18 and a control device 20 that controls the servers 18 and the cooling system 10. The cooling system 10 includes a cooling device 11 that cools a first-side refrigerant, a first-side pipe 15 provided with a first-side pump 13 that delivers the first-side refrigerant cooled by the cooling device 11, a heat exchanger 14 that exchanges heat between the first-side refrigerant cooled by the cooling device 11 and a second-side refrigerant used to cool the servers 18, and a second-side pump 16 that delivers the second-side refrigerant cooled by the heat exchanger 14. The secondary side piping 17 has a flow path for each server group G, which consists of a main pipe 171, a plurality of first branch pipes 172 branching from the main pipe 171 at branch positions corresponding to each server group G, and a plurality of second branch pipes 173 branching from the first branch pipe 172 to each server 18, and on-off valves 174, 175 provided on each of the plurality of second branch pipes 173 of the secondary side piping 17, and the control device 20 has a first allocation unit 212 that allocates jobs in a concentrated manner starting from the server group G with the shortest flow path distance based on schedule information for jobs that will perform calculation processing on the servers 18 and server information including the placement of the servers 18, and the opening degrees of the on-off valves 174, 175 are changed according to the processing load of the servers 18.

[0088] In conventional technology, jobs were assigned to servers designated by clients, which distributed the processing load across groups A to D and required the secondary refrigerant to circulate through all groups A to D. In contrast, in this embodiment, the control device 20 concentrates the processing load on group A, which has a shorter flow path distance, allowing the secondary refrigerant to circulate only through group A. This shortens the distance the secondary refrigerant must flow, thereby shortening the secondary refrigerant circulation path compared to circulating the secondary refrigerant to group D. This reduces pressure loss in the main pipe 171, assuming the same flow rate of the secondary refrigerant, and reduces the power (power consumption) of the secondary pump 16. This minimizes the power consumption of the entire data center, including the cooling system 10 for the servers 18.

[0089] (2) According to the second aspect, in the control system 1 relating to the first aspect, the opening degree of the on-off valves 174, 175 is automatically changed based on the temperature difference of the secondary side refrigerant at the inlet and outlet of the server 18 of the second branch pipe 173.

[0090] In this way, for servers 18 that have no job assignments (processing loads) and have small temperature differences, the on-off valves 174 and 175 can be automatically closed to stop the flow of secondary refrigerant without the user having to monitor or operate them.

[0091] (3) According to the third aspect, in the control system 1 relating to the first aspect, the control device 20 further includes a control unit 215 that performs control to change the opening degree based on a user's operation instruction or based on the temperature difference of the secondary side refrigerant at the inlet and outlet of the server 18 of the second branch pipe 173.

[0092] By doing this, even if on-off valves 174, 175 that are not automatic valves are installed, for servers 18 that have no job assignments (processing load) and have a small temperature difference, the on-off valves 174, 175 can be closed by user operation or by automatic control of the control unit 215 to stop the flow of secondary side refrigerant.

[0093] (4) According to a fourth aspect, in the control system 1 relating to any one of the first to third aspects, the control device 20 further includes a second allocation unit 216 that allocates jobs to all of the servers 18 in all of the server groups G so as to distribute the processing load evenly based on schedule information and server information; a power calculation unit 214 that calculates a first predicted value of the power of the cooling system 10 according to the job allocation of the first allocation unit 212 and a second predicted value of the power of the cooling system 10 according to the job allocation of the second allocation unit 216 based on information indicating the relationship between the allowable temperature, heat generation amount, flow rate of the primary-side refrigerant and the secondary-side refrigerant of the servers 18, and the total power of the cooling system 10 for supplying the primary-side refrigerant and the secondary-side refrigerant to the servers 18; and a determination unit 217 that determines which job allocation of the first allocation unit 212 or the second allocation unit 216 to adopt based on the first predicted value and the second predicted value.

[0094] In this way, the control device 20 can select and control conditions (job allocation) that allow for further power reduction, thereby making it possible to further effectively reduce the power consumption of the cooling system 10.

[0095] (5) According to the fifth aspect, in the control system 1 relating to any one of the first to fourth aspects, the facility R has multiple floors on which server groups G are installed, and the first allocation unit 212 concentrates and allocates jobs to the server groups G on some of the floors so as to minimize the number of operating floors.

[0096] By doing this, it is possible to minimize the number of operating floors, thereby further reducing the power consumption of the entire facility R. In addition, it is possible to submit jobs from other clients to servers 18 on floors that are not in operation, allowing the data center to handle jobs from more clients.

[0097] (6) According to a sixth aspect, a control method is a control method for controlling a cooling system 10 that cools the servers 18 and the servers 18 in a facility R in which a plurality of server groups G, each of which is composed of a plurality of servers 18 arranged in a first horizontal direction, are installed at intervals in a second horizontal direction perpendicular to the first horizontal direction, the control method including: controlling a cooling system 10 that cools the servers 18 and the servers 18; controlling the servers 18; and controlling the cooling system 10, the cooling system 10 including a cooling device 11 that cools a first-side refrigerant; a first-side pipe 15 provided with a first-side pump 13 that delivers the first-side refrigerant cooled by the cooling device 11; a heat exchanger 14 that exchanges heat between the first-side refrigerant cooled by the cooling device 11 and a second-side refrigerant used to cool the servers 18; and controlling the second-side pipe 17 provided with a second-side pump 16 that delivers the second-side refrigerant cooled by the heat exchanger 14. The system has a main pipe 171, a plurality of first branch pipes 172 branching from the main pipe 171 at branch positions corresponding to each server group G, and a plurality of second branch pipes 173 branching from the first branch pipe 172 to each server 18, and secondary side piping 17 having a flow path for each server group G, and on-off valves 174, 175 provided on each of the plurality of second branch pipes 173 of the secondary side piping 17, and the control method has a step of concentrating and allocating jobs in order from the server group G with the shortest flow path distance based on schedule information for jobs that perform calculation processing on the servers 18 and server information including the placement of the servers 18, and a step of changing the opening degree of the on-off valves 174, 175 in accordance with the processing load of the servers 18.

[0098] According to the above aspect, it is possible to minimize the power consumption of the entire data center including the cooling system of the servers.

[0099] DESCRIPTION OF SYMBOLS 1 Control system 10 Cooling system 11 Cooling device 12 Fan 13 Primary pump 14 Heat exchanger 15 Primary piping 16 Secondary pump 17 Secondary piping 171 Main pipe 172 First branch pipe 173 Second branch pipe 174, 175 On-off valve 18 Server 19 CPU (heat generating element) 20 Control device 21 Processor 211 Acquisition unit 212 First allocation unit 213 Heat generation amount estimation unit 214 Power calculation unit 215 Control unit 216 Second allocation unit 217 Determination unit 22 Memory 23 Storage 24 Communication interface 25 Display unit 26 Operation unit G Server group

Claims

1. In a facility where a plurality of server groups each composed of a plurality of servers arranged in a first horizontal direction are installed at intervals in a second horizontal direction orthogonal to the first horizontal direction, the facility includes a cooling system for cooling the servers and a control device for controlling the servers and the cooling system. The cooling system includes a cooling device for cooling a primary refrigerant, a primary pipe provided with a primary pump for sending out the primary refrigerant cooled by the cooling device, a heat exchanger for performing heat exchange between the primary refrigerant cooled by the cooling device and a secondary refrigerant used for cooling the servers, and a secondary pipe provided with a secondary pump for sending out the secondary refrigerant cooled by the heat exchanger. The secondary pipe includes a main pipe, a plurality of first branch pipes branched at branch positions corresponding to each server group from the main pipe, and a plurality of second branch pipes branched from the first branch pipes to each server, and has a flow path for each server group. The cooling system further includes an on-off valve provided in each of the plurality of second branch pipes of the secondary pipe. The control device has a first allocating unit for concentrating and allocating the jobs in order from the server group with the shortest flow path distance based on schedule information of jobs for performing arithmetic processing in the servers and server information including the arrangement of the servers. The on-off valve has an opening degree changed according to the processing load of the servers. Control system.

2. The control system according to claim 1, wherein the on-off valve automatically changes the opening degree based on a temperature difference of the secondary refrigerant at an inlet and an outlet of the server in the second branch pipe.

3. The control system according to claim 1, further comprising a control unit that performs control to change the opening degree based on an operation instruction of a user or based on a temperature difference of the secondary refrigerant at an inlet and an outlet of the server in the second branch pipe.

4. The control device further includes: a second allocation unit that allocates all the jobs to all the servers in all the server groups so that the processing load is equal based on the schedule information and the server information; a power calculation unit that calculates a first predicted value of the power of the cooling system corresponding to the job allocation of the first allocation unit and a second predicted value of the power of the cooling system corresponding to the job allocation of the second allocation unit based on information indicating the relationship between the allowable temperature of the server, the heat generation amount, the flow rates of the primary refrigerant and the secondary refrigerant, and the total power of the cooling system for supplying the primary refrigerant and the secondary refrigerant to the server; and a determination unit that determines which job allocation of the first allocation unit and the second allocation unit to adopt based on the first predicted value and the second predicted value. The control system according to any one of claims 1 to 3.

5. The facility has a plurality of floors on which the server groups are installed, and the first allocation unit concentrates and allocates the jobs to the server groups on some of the floors so that the number of operating floors is minimized. The control system according to any one of claims 1 to 3.

6. In a facility where a plurality of server groups each composed of a plurality of servers arranged in a first horizontal direction are installed at intervals in a second horizontal direction orthogonal to the first horizontal direction, a cooling system for cooling the servers and a control method for controlling the servers, wherein the cooling system includes: a cooling device for cooling a primary-side refrigerant; a primary-side pipe provided with a primary-side pump for sending out the primary-side refrigerant cooled by the cooling device; a heat exchanger for performing heat exchange between the primary-side refrigerant cooled by the cooling device and a secondary-side refrigerant used for cooling the servers; a secondary-side pipe provided with a secondary-side pump for sending out the secondary-side refrigerant cooled by the heat exchanger, the secondary-side pipe including a main pipe, a plurality of first branch pipes branching from the main pipe at branch positions corresponding to each of the server groups, and a plurality of second branch pipes branching from each of the first branch pipes to each of the servers, and having a flow path for each server group; and on-off valves provided in each of the plurality of second branch pipes of the secondary-side pipe. The control method includes: a step of concentrating and allocating the jobs in order from the server group with the shortest flow path distance based on server information including scheduled information of jobs for performing arithmetic processing in the servers and the arrangement of the servers; and a step of changing the opening degree of the on-off valves according to the processing load of the servers.

Images