Power management method in cabinet system and related network data center

Abstract

The invention discloses a power management method of a cabinet system and a related network data center. The cabinet system comprises at least one server and a power supply with a plurality of power supply units. After the system is started, all power supply units in the power supply are firstly started to supply electric power required by operation of at least one server. The maximum output power value of the power supply and a conversion efficiency table are read, the real-time conversion efficiency of the power supply is calculated, and the conversion efficiency table comprises the corresponding relation between the load rate of the power supply and the conversion efficiency of the power supply; when it is determined that the power supply does not operate at the optimal conversion efficiency according to the conversion efficiency table and the real-time conversion efficiency, the number of started power supply units in the power supply is adjusted according to a predetermined rule.

Description

Technical Field

[0001] The present invention relates to a power management method in a cabinet system and a related network data center, and more particularly to a power management method and a related network data center for ensuring that a power supply of the cabinet system can continuously operate with optimal conversion efficiency. Background Art

[0002] An Internet Data Center (IDC) is a facility with many networked computers that work together to collect, store, process, and transmit data in a centralized manner. A typical IDC usually contains a large number of servers to store data and execute applications, which can be accessed by users through remotely connected computer devices. In order to facilitate unified management, multiple servers are usually set up in the same rack or tank, and server resources are allocated and managed. Each server in each rack system will contain a baseboard management controller (BMC) to manage internal operations and handle network communications with the central management station in the IDC.

[0003] In a cabinet system, the utilization rate of each server depends on its usage pattern, current time, number of users, and other factors. Under the above conditions, the server workload sometimes reaches 100% hardware resource utilization, and sometimes 50% or less, where the above hardware resources generally include processor cores, system memory, storage controllers, Ethernet controllers, and input / output peripherals. The power required for the operation of each server can be centrally supplied by the power distribution unit (PDU) in the cabinet / slot, or separately supplied by the power supply unit (PSU) built into each server. The energy conversion efficiency of the power distribution unit and the power supply unit is related to the workload rate of the server. Generally speaking, when the server is under load, the energy conversion efficiency of the power distribution unit and the power supply unit will be relatively poor, resulting in energy waste.

[0004] Therefore, a power management method for a cabinet system is needed to ensure that the power supply can continue to operate at the best conversion efficiency to achieve the purpose of energy saving and carbon reduction. Summary of the invention

[0005] The present invention provides a method for power management in a cabinet system, wherein the data center includes a power supply and at least one server, and the power supply includes a plurality of power supply units. The method includes turning on all power supply units in the power supply to supply the power required for the operation of the at least one server; reading a maximum output power value and a conversion efficiency table of the power supply, wherein the conversion efficiency table includes a corresponding relationship between the load rate of the power supply and the conversion efficiency of the power supply; calculating a real-time conversion efficiency of the power supply; judging whether the power supply operates at an optimal conversion efficiency based on the conversion efficiency table and the real-time conversion efficiency; and when it is determined that the power supply does not operate at the optimal conversion efficiency, adjusting the number of power supply units in the power supply to be turned on according to a predetermined rule.

[0006] The present invention also provides a cabinet system, which includes multiple servers, a power supply and a cabinet management controller. The power supply includes multiple power supply units. The cabinet management controller is used to turn on all power supply units in the power supply after startup to supply the power required for the operation of the multiple servers; read a maximum output power value and a conversion efficiency table of the power supply, wherein the conversion efficiency table includes the corresponding relationship between the load rate of the power supply and the conversion efficiency of the power supply; calculate a real-time conversion efficiency of the power supply; determine whether the power supply operates at an optimal conversion efficiency based on the conversion efficiency table and the real-time conversion efficiency; and when it is determined that the power supply does not operate at the optimal conversion efficiency, adjust the number of the multiple power supply units in the power supply according to a predetermined rule.

[0007] The present invention also provides a cabinet system, which includes at least one server and a cabinet management controller. The at least one server includes a power supply and a baseboard management controller. The power supply includes a plurality of power supply units. The baseboard management controller is used to control the operation of the plurality of power supply units in the power supply to supply the power required for the operation of the at least one server. The cabinet management controller is used to read a maximum output power value and a conversion efficiency table of the power supply through the baseboard management controller, wherein the conversion efficiency table includes a corresponding relationship between the load rate of the power supply and the conversion efficiency of the power supply; calculate a real-time conversion efficiency of the power supply; determine whether the power supply operates at an optimal conversion efficiency based on the conversion efficiency table and the real-time conversion efficiency; and when it is determined that the power supply does not operate at the optimal conversion efficiency, instruct the baseboard management controller to adjust the number of open power supply units in the power supply according to a predetermined rule. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1Schematic diagram of a cabinet system in an embodiment of the present invention.

[0009] Figure 2 It is a schematic diagram of a cabinet system in another embodiment of the present invention.

[0010] Figure 3 FIG. 4 is a flow chart of the operation of the cabinet system according to an embodiment of the present invention.

[0011]

Explanation of symbols

[0012] 100, 200 cabinet system

[0013] 110, 210: Management system

[0014] 120, 220: Cabinet management controller

[0015] 130, 230: Power supply

[0016] 140: Switch

[0017] 150: Rail power supply system

[0018] 160, 260: Cabinet / slot

[0019] 240: Baseboard Management Controller

[0020] 310-360: Steps

[0021] SV, SV_1-SV_M: Server

[0022] PSU_1-PSU_N: Power supply unit DETAILED DESCRIPTION

[0023] Figure 1 Schematic diagram of a rack system 100 in an embodiment of the present invention. The rack system 100 includes a management system 110, a rack management controller (RMC) 120, a power supply 130, a switch 140, a busbar 150, and servers SV_1-SV_M (M is a positive integer). The rack system 100 adopts a centralized power supply architecture, wherein the power supply 130 includes a plurality of power supply units (PSU) PSU_1-PSU_N (N is an integer greater than 1), and power can be transmitted to the servers SV_1-SV_M through the busbar 150. In an embodiment of the present invention, the rack management controller 120, the power supply 130, the switch 140, the busbar 150, and the servers SV_1-SV_M can be arranged in the same rack / tank 160, but are not limited thereto.

[0024] Figure 2 Schematic diagram of a cabinet system 200 in another embodiment of the present invention. The cabinet system 200 includes a management system 210, a cabinet management controller 220, and at least one server SV. The cabinet system 200 adopts a separate power supply architecture, wherein each server SV includes a power supply 230 and a baseboard management controller (BMC) 240. The power supply 230 built into each server SV includes a plurality of power supply units PSU_1-PSU_N (N is an integer greater than 1), which can supply the power required for its own operation. In an embodiment of the present invention, the cabinet management controller 220 and the server SV can be implemented as a cabinet / slot 260, but are not limited thereto.

[0025] Figure 3 1 is a flowchart of the operation of the cabinet systems 100 and 200 according to an embodiment of the present invention, which includes the following steps:

[0026] Step 310: System startup.

[0027] Step 320: Turn on all power supply units in the power supply.

[0028] Step 330: Read the maximum output power value and conversion efficiency table of the power supply.

[0029] Step 340: Calculate the real-time conversion efficiency of the power supply.

[0030] Step 350 : Determine whether the power supply operates at the optimal conversion efficiency according to the conversion efficiency table; if so, execute step 340 ; if not, execute step 360 .

[0031] Step 360: Adjust the number of enabled power supply units in the power supply according to a predetermined rule.

[0032] As is known to those skilled in the art, a cabinet system generally includes computing infrastructure (e.g., servers, internal storage, and applications), network infrastructure (e.g., cables, switches, routers, and firewalls), storage infrastructure (e.g., block storage devices and file storage devices), and support infrastructure (e.g., power subsystems, uninterruptible power systems, backup generators, ventilation and cooling equipment, fire extinguishing systems, and built-in security systems). To simplify the description, Figure 1 The cabinet system 100 and Figure 2 The cabinet system 200 shown only shows the structure related to the present invention.

[0033] The management systems 110 and 210 are respectively management consoles for centralized and unified resource processing in the cabinet systems 100 and 200, which can monitor the processing requirements, components and / or connection status of the cabinet systems 100 and 200 in real time, allowing users to manage them remotely. The cabinet management controllers 120 and 220 can serve as centralized connection points for environmental monitoring, asset location, physical access, and other monitoring and information security sensors. However, the implementation of the management systems 110 and 210 does not limit the scope of the present invention.

[0034] In one embodiment, the maximum output power value that each power supply unit in the power supply 130 or 230 can provide is the same. In another embodiment, each power supply unit in the power supply 130 or 230 can provide a maximum output power of a different value. However, the type and specification of each power supply unit in the power supply 130 and 230 do not limit the scope of the present invention.

[0035] When the rack system 100 executes the booting procedure in step 310, the rack management controller 220 then turns on all the power supply units in the power supply 130 in step 320, thereby transmitting power to the servers SV_1-SV_M through the rail power supply system 150. When the rack system 200 executes the booting procedure in step 310, the rack management controller 220 then instructs the baseboard management controller 240 to turn on all the power supply units in the built-in power supply 230 of the server SV in step 320, thereby supplying the power required for the operation of the server SV.

[0036] With the rise of environmental awareness, countries have set energy-saving standards for consumer electronics, office equipment, home appliances and external power supplies. For example, for the backup power supply (230V internal redundant) used in servers and cabinet systems, if the power supplies 130 and 230 meet the 80 Plus Titanium specification, it means that the minimum conversion efficiency of each power supply unit at 10% load rate, 20% load rate, 50% load rate and full load (100%) must reach at least 90%, 94%, 96% and 91% respectively.

[0037] exist Figure 1 In the rack system 100 shown in FIG. 1 , the rack management controller 220 reads the maximum output power value P of the power supply 130 in step 330 . MAX And conversion efficiency table. Figure 2 In the rack system 200 shown in FIG. 1 , the rack management controller 220 reads the maximum output power value P of the corresponding power supply 230 through the baseboard management controller 240 of each server SV in step 330 . MAXand a conversion efficiency table. The conversion efficiency table includes the corresponding relationship between the load rate of the power supply 130 or 230 and its conversion efficiency. Assuming that each power supply unit in the power supply 130 or 230 meets the Titanium Full 80+ specification, the following table 1 shows the conversion efficiency table read in step 330, wherein the load rate of the power supply is defined as the ratio of the actual output power to the maximum power value, and the conversion efficiency is defined as the ratio of AC to DC.

[0038] Power supply load factor Power supply conversion efficiency 0％-19％ 90％ 20％-37％ 94％ 38％-82％ 96％ 83％-100％ 91％

[0039] Table 1

[0040] As shown in Table 1, when the load rate of the power supply 130 or 230 is between 38% and 82%, the power supply 130 or 230 can operate at an optimal conversion efficiency of 96%, thereby achieving the purpose of saving electricity and reducing waste heat. It is worth noting that due to differences in the actual operation of the device and environmental factors, the conversion efficiency tables of different power supplies may be slightly different. The conversion efficiency table shown in Table 1 above is only an example, but does not limit the scope of the present invention.

[0041] The maximum output power value P of the power supply 130 and 230 MAX For the purpose of illustration, it is assumed that the power supply 130 and 230 each include power supply units PSU_1-PSU_5 (N=5), wherein the maximum output power values ​​that the power supply units PSU_1-PSU_5 can provide are all 2500W. Since all the power supply units in the power supply 130 and 230 have been turned on in step 320, the maximum output power values ​​P of the power supply 130 and 230 are read in step 330. MAX It is 12500W.

[0042] exist Figure 1 In the rack system 100 shown in FIG. 1 , the rack management controller 120 calculates the real-time conversion efficiency P of the power supply 130 in step 340 . EFF .exist Figure 2 In the rack system 200 shown in FIG. 1 , the rack management controller 220 calculates the real-time conversion efficiency P of the power supply 230 in step 340 . EFF In one embodiment, the rack management controller 120 or the baseboard management controller 240 first measures the average output power of the power supply 130 or 230 within a predetermined period of time (eg, 30 seconds) as its real-time output power P AV , and then according to the maximum output power value P MAX Calculate the corresponding real-time conversion efficiency P EFF , that is, P EFF =PAV / P MAX .

[0043] exist Figure 1 In the rack system 100 shown in FIG. 1 , the real-time output power P of the power supply 130 at a specific time point is AV Related to the load rate of the servers SV_1-SV_M: When the total power consumption value of the servers SV_1-SV_M is larger, the real-time output power P of the power supply 130 is AV The closer it is to the maximum output power value P MAX , and the real-time conversion efficiency P EFF When the total power consumption value of the server SV_1-SV_M is smaller, the real-time output power P of the power supply 130 AV And the maximum output power value P MAX The greater the difference, the higher the real-time conversion efficiency P EFF The lower.

[0044] exist Figure 2 In the rack system 200 shown in FIG. 1 , the real-time output power P of the power supply 230 of each server SV at a specific time point is AV Related to the self-load rate: When the total power consumption value of the server SV is larger, the real-time output power P of the power supply 230 is AV The closer it is to the maximum output power value P MAX , and the real-time conversion efficiency P EFF When the total power consumption value of the server SV is smaller, the real-time output power P of the power supply 230 AV And the maximum output power value P MAX The greater the difference, the higher the real-time conversion efficiency P EFF The lower.

[0045] exist Figure 1 In the rack system 100 shown in FIG. 1 , the rack management controller 120 determines whether the power supply 130 is operating at the optimal conversion efficiency according to the conversion efficiency table of the power supply 130 (such as shown in Table 1) in step 350. In one embodiment, the power supply 130 can be operated by a built-in controller (not shown in FIG. 1 ). Figure 1) to monitor the operation of the plurality of power supply units PSU_1-PSU_N, and the rack management controller 120 directly learns the current workload (related to the total power consumption value of the servers SV_1-SV_M) from the power supply 130, and then determines whether the power supply 130 operates at the best conversion efficiency according to the current workload of the power supply 130. In another embodiment, the rack management controller 120 may receive power information (such as the voltage, current and power consumption of each server, etc.) of the servers SV_1-SV_M through the switch 140, and calculate the total power consumption value of the servers SV_1-SV_M according to the power information to obtain the current workload of the power supply 130, and then determine whether the power supply 130 operates at the best conversion efficiency according to the current workload of the power supply 130. In one embodiment, the switch 140 may be a managed switch, but is not limited thereto.

[0046] exist Figure 2 In the rack system 200 shown, the rack management controller 220 determines whether the power supply 230 operates at the best conversion efficiency according to the conversion efficiency table of the power supply 230 (such as shown in Table 1) in step 350. In one embodiment, the rack management controller 220 can obtain power information (such as the voltage, current and power consumption of each server SV) through the baseboard management controller 240 of each server SV, and then calculate the total power consumption value of the server SV to obtain the current workload of the power supply 230, and then determine whether the power supply 230 operates at the best conversion efficiency according to the current workload of the power supply 230.

[0047] When it is determined that the power supply 130 or 230 is currently operating at the optimal conversion efficiency, the rack system 100 or 200 will execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF When it is determined that the power supply 130 or 230 is not currently operating at the optimal conversion efficiency, the cabinet system 100 or 200 executes step 360 to adjust the number of power supply units in the power supply 130 or 230 to be turned on according to a predetermined rule, and then executes steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0048] Table 2 below shows the status of the cabinet system 100 or 200 at different time points during operation in the embodiment of the present invention. For illustration purposes, it is also assumed that the power supply 130 and 230 each include five power supply units PSU_1-PSU_5, wherein the maximum output power value that the power supply units PSU_1-PSU_5 can provide is 2500W.

[0049]

[0050]

[0051] Table 2

[0052] At time point T1 when step 350 is first executed after the system is powered on, since all power supply units in the power supply 130 or 230 are turned on at this time, the rack management controller 120 or 220 can obtain the maximum output power value P of the power supply 130 or 230. MAX Assume that at time point T1, the servers SV_1-SV_M or the server SV are in full load operation (for example, the total power consumption value is 6400 W). At this time, the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 6400W. Therefore, when all five power supply units PSU_1-PSU_5 are turned on (the maximum output power value is 12500W), the actual load rate is about 51%. By referring to the conversion efficiency table shown in Table 1, the rack management controller 130 or 230 can know that when the power supply 130 or 230 is at a load rate of 51%, it operates at the best conversion efficiency of 96%. In this case, the rack system 100 or 200 will directly execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0053] At time point T2 after time point T1, assuming that the server SV_1-SV_M or server SV is reduced to 30% load operation (for example, the total power consumption value is 1920W), the real-time output power P of the power supply 130 or 230 is AVThe power consumption of the power supply 130 or 230 is reduced to 1920W accordingly. Therefore, when all five power supply units PSU_1-PSU_5 are turned on (the maximum output power value is 12500W), the actual load rate is about 15%. By comparing the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 15%, it operates at a conversion efficiency of 90%, not at the best conversion efficiency of 96%. In this case, the rack management controller 130 or 230 will first execute step 360 to turn off or turn on any power supply unit in the power supply 130 or 230. For example, according to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that the power supply 130 or 230 needs to increase its load rate to operate at the best conversion efficiency, so it will turn off the power supply unit PSU_1 with a maximum output power value of 2500W, and then execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0054] At time point T3 after time point T2, assuming that the server SV_1-SV_M or server SV is still operating at 30% load (for example, the total power consumption value is 1920W), the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 1920W. Therefore, when only four power supply units PSU_2-PSU_5 are turned on (the maximum output power value is 10000W), the actual load rate is about 19%. According to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 19%, it operates at a conversion efficiency of 90%, not at the best conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will execute step 360 again to turn off or turn on any power supply unit in the power supply 130 or 230. For example, according to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that the power supply 130 or 230 needs to increase its load rate to operate at the best conversion efficiency, so it will turn off the power supply unit PSU_2 with a maximum output power value of 2500W, and then execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0055] At time point T4 after time point T3, assuming that the server SV_1-SV_M or server SV is still operating at 30% load (for example, the total power consumption value is 1920W), the real-time output power P of the power supply 130 or 230 is AVThe corresponding value is 1920W. Therefore, when only three power supply units PSU_3-PSU_5 are turned on (the maximum output power value is 7500W), the actual load rate is about 26%. According to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 26%, it operates at a conversion efficiency of 94%, not at the best conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will execute step 360 again to turn off or turn on any power supply unit in the power supply 130 or 230. For example, according to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that the power supply 130 or 230 needs to increase its load rate to operate at the best conversion efficiency, so it will turn off the power supply unit PSU_3 with a maximum output power value of 2500W, and then execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0056] At time point T5 after time point T4, assuming that the server SV_1-SV_M or server SV is still operating at 30% load (for example, the total power consumption value is 1920W), the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 1920W. Therefore, when only two power supply units PSU_4 and PSU_5 are turned on (the maximum output power value is 5000W), the actual load rate is about 38%. By referring to the conversion efficiency table shown in Table 1, the cabinet management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 38%, it is operating at the best conversion efficiency of 96%. In this case, the cabinet system 100 or 200 will directly execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0057] At time point T6 after time point T5, assuming that the server SV_1-SV_M or server SV is increased to 70% load operation (for example, the total power consumption value is 4480W), the real-time output power P of the power supply 130 or 230 is AVThe corresponding value is 4480W. Therefore, when only two power supply units PSU_4 and PSU_5 are turned on (the maximum output power value is 5000W), the actual load rate is about 90%. According to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a 90% load rate, it operates at a conversion efficiency of 91%, not at the optimal conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will execute step 360 again to turn off or turn on any power supply unit in the power supply 130 or 230. For example, according to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that the power supply 130 or 230 needs to reduce its load rate to operate at the best conversion efficiency, so it will turn on the power supply unit PSU_3 with a maximum output power value of 2500W, and then execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0058] At time point T7 after time point T6, assuming that the server SV_1-SV_M or server SV still maintains 70% load operation (for example, the total power consumption value is 4480W), the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 4480W. Therefore, when only three power supply units PSU_3-PSU_5 are turned on (the maximum output power value is 7500W), the actual load rate is about 60%. By referring to the conversion efficiency table shown in Table 1, the cabinet management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 60%, it operates at the best conversion efficiency of 96%. In this case, the cabinet system 100 or 200 will directly execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0059] Table 3 below shows the status of the cabinet system 100 or 200 at different time points during operation in another embodiment of the present invention. For illustration purposes, it is also assumed that the power supply 130 and 230 each include five power supply units PSU_1-PSU_5, wherein the maximum output power value that the power supply units PSU_1--PSU_5 can provide is 2500W.

[0060]

[0061]

[0062] Table 3

[0063] At time point T1 when step 350 is first executed after the system is powered on, since all power supply units in the power supply 130 or 230 are turned on at this time, the rack management controller 120 or 220 can obtain the maximum output power value P of the power supply 130 or 230. MAX Assume that at time point T1, the servers SV_1-SV_M or the server SV are in full load operation (for example, the total power consumption value is 6400 W). At this time, the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 6400W. Therefore, when all five power supply units PSU_1-PSU_5 are turned on (the maximum output power value is 12500W), the actual load rate is about 51%. By referring to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 51%, it operates at the best conversion efficiency of 96%. In this case, the rack system 100 or 200 will directly execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0064] At time point T2 after time point T1, assuming that the server SV_1-SV_M or server SV is reduced to 30% load operation (for example, the total power consumption value is 1920W), the real-time output power P of the power supply 130 or 230 is AV The power consumption of the power supply 130 or 230 is reduced to 1920W accordingly. Therefore, when all five power supply units PSU_1-PSU_5 are turned on (the maximum output power value is 12500W), the actual load rate is about 15%. By comparing the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 15%, it operates at a conversion efficiency of 90%, not at the best conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will first execute step 360 to turn off or turn on a specific number of power supply units in the power supply 130 or 230 according to the historical records. For example, according to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that the power supply 130 or 230 needs to increase its load rate to operate at the best conversion efficiency. In addition, based on the history of previously turning on and off each power supply unit, the rack management controller 120 or 220 can know that turning off the power supply units PSU_2-PSU_5 at the same time can more efficiently and accurately increase the load rate of the power supply 130 or 230, so it will turn off the four power supply units PSU_2-PSU_5 with a maximum total output power value of 10000W, and then execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 and 230.EFF .

[0065] At time point T3 after time point T2, assuming that the server SV_1-SV_M or server SV is still operating at 30% load (for example, the total power consumption value is 1920W), the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 1920W. Therefore, when only one power supply unit PSU_1 is turned on (the maximum output power value is 2500W), the actual load rate is about 77%. By referring to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 77%, it operates at the best conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will directly execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 and 230. EFF .

[0066] Assume that at time point T4 after time point T3, the server SV_1-SV_M or server SV increases to 39% load operation (for example, the total power consumption value is 2500W). At this time, the real-time output power P of the power supply 130 or 230 is AV is 2500W, so the actual load rate is 100%. By comparing the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a load rate of 100%, it operates at a conversion efficiency of 91%, not at the optimal conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will first execute step 360 to turn off or turn on a specific number of power supply units in the power supply 130 or 230 according to the historical records. For example, according to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that at this time, the power supply 130 or 230 needs to reduce its load rate to operate at the optimal conversion efficiency. In addition, based on the history of previously turning on and off each power supply unit, the rack management controller 120 or 220 can know that turning on the power supply units PSU_1 and PSU_2 at the same time can more efficiently and accurately reduce the load rate of the power supply 130, so it will turn on the two power supply units PSU_1 and PSU_2 with a maximum total output power value of 5000W, and then execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supplies 130 and 230. EFF .

[0067] Assume that at time point T5 after time point T4, the server SV_1-SV_M or the server SV is still operating at 39% load (for example, the total power consumption value is 2500W). At this time, the real-time output power P of the power supply 130 or 230 is AV The corresponding value is 2500W. Therefore, when the power supply units PSU_1 and PSU_2 are turned on (the maximum output power value is 5000W), the actual load rate is about 50%. By referring to the conversion efficiency table shown in Table 1, the rack management controller 120 or 220 can know that when the power supply 130 or 230 is at a 50% load rate, it operates at the best conversion efficiency of 96%. In this case, the rack management controller 120 or 220 will directly execute steps 340 and 350 again to continuously monitor and determine the real-time conversion efficiency P of the power supply 130 or 230. EFF .

[0068] In summary, the cabinet system of the present invention monitors the load status of the power supply in real time. When the load rate of the power supply decreases due to the server with a smaller load rate and thus cannot operate at the best conversion efficiency, a specific number of power supply units will be turned off to increase the load rate of the power supply; when the load rate of the power supply increases due to the server with a larger load rate and thus cannot operate at the best conversion efficiency, a specific number of power supply units will be turned on to reduce the load rate of the power supply. Figure 1 ) and server level ( Figure 2 The present invention can ensure that the power supply can continue to operate at the best conversion efficiency without being affected by the load rate changes of the server, so as to achieve the purpose of energy saving and carbon reduction.

[0069] The above descriptions are only preferred embodiments of the present invention. All equivalent changes and modifications made according to the claims of the present invention should fall within the scope of the present invention.

Claims

1. A method for power management in a rack system, wherein the rack system comprises a power supply and at least one server, and the power supply comprises a plurality of power supply units, the method comprising: Turning on all power supply units in the power supply to supply power required for the operation of the at least one server; Reading a maximum output power value and a conversion efficiency table of the power supply, wherein the conversion efficiency table includes a corresponding relationship between a load rate of the power supply and a conversion efficiency of the power supply; Calculating the real-time conversion efficiency of the power supply; determining whether the power supply operates at an optimal conversion efficiency according to the conversion efficiency table and the real-time conversion efficiency; and When it is determined that the power supply does not operate at the optimal conversion efficiency, the number of enabled power supply units in the power supply is adjusted according to a predetermined rule.

2. The method of claim 1, further comprising: When it is determined that the power supply is not operating at the optimal conversion efficiency, turning off or turning on any power supply unit in the power supply; and After any power supply unit in the power supply is turned off or on, the real-time conversion efficiency of the power supply is calculated again to determine whether the power supply can operate at the optimal conversion efficiency.

3. The method of claim 1, further comprising: When it is determined that the power supply is not operating at the optimal conversion efficiency, shutting down or opening a predetermined number of power supply units in the power supply according to the historical records of the maximum output power values ​​of all the power supply units; and After the predetermined number of power supply units in the power supply are turned off or on, the real-time conversion efficiency of the power supply is calculated again to determine whether the power supply can operate at the optimal conversion efficiency.

4. The method of claim 1, further comprising: measuring an average output power of the power supply within a predetermined period of time; and The average output power is divided by the maximum output power value to obtain the real-time conversion efficiency of the power supply.

5. A cabinet system, comprising: Multiple servers; A power supply device, comprising a plurality of power supply units; Rack management controller (RMC), used to: After booting up, turning on all power supply units in the power supply to supply power required for the operation of the plurality of servers; Reading a maximum output power value and a conversion efficiency table of the power supply, wherein the conversion efficiency table includes a corresponding relationship between a load rate of the power supply and a conversion efficiency of the power supply; Calculating the real-time conversion efficiency of the power supply; determining whether the power supply operates at an optimal conversion efficiency according to the conversion efficiency table and the real-time conversion efficiency; and When it is determined that the power supply does not operate at the optimal conversion efficiency, the enabled quantity of the plurality of power supply units in the power supply is adjusted according to a predetermined rule.

6. The cabinet system according to claim 5, further comprising: A busbar power supply system is coupled to the plurality of servers and the power supply, and is used for transmitting power provided by the power supply to the plurality of servers.

7. The cabinet system according to claim 5, further comprising: The switch is coupled between the rack management controller and the plurality of servers.

8. The rack system as claimed in claim 5, wherein the rack management controller is further configured to: When it is determined that the power supply is not operating at the optimal conversion efficiency, turning off or turning on any power supply unit in the power supply; and After any power supply unit in the power supply is turned off or on, the real-time conversion efficiency of the power supply is calculated again to determine whether the power supply can operate at the optimal conversion efficiency.

9. The rack system as claimed in claim 5, wherein the rack management controller is further used for: When it is determined that the power supply is not operating at the optimal conversion efficiency, shutting down or opening a predetermined number of power supply units in the power supply according to a historical record of the maximum output power value of each power supply unit; and After the predetermined number of power supply units in the power supply are turned off or on, the real-time conversion efficiency of the power supply is calculated again to determine whether the power supply can operate at the optimal conversion efficiency.

10. The rack system as claimed in claim 5, wherein the rack management controller is further used for: measuring an average output power of the power supply within a predetermined period of time; and The average output power is divided by the maximum output power value to obtain the real-time conversion efficiency of the power supply.

11. A cabinet system, comprising: At least one server comprising: A power supply device, comprising a plurality of power supply units; a baseboard management controller (BMC), used to control the operation of the plurality of power supply units in the power supply to supply power required for the operation of the at least one server; and Rack management controller (RMC), used to: Reading the maximum output power value and the conversion efficiency table of the power supply through the baseboard management controller, wherein the conversion efficiency table includes the corresponding relationship between the load rate of the power supply and the conversion efficiency of the power supply; Calculating the real-time conversion efficiency of the power supply; determining whether the power supply operates at an optimal conversion efficiency according to the conversion efficiency table and the real-time conversion efficiency; and When it is determined that the power supply does not operate at the optimal conversion efficiency, the baseboard management controller is instructed to adjust the number of enabled power supply units in the power supply according to a predetermined rule.

12. The rack system as claimed in claim 11, wherein the rack management controller is further configured to: When it is determined that the power supply is not operating at the optimal conversion efficiency, instructing the baseboard management controller to turn off or turn on any power supply unit in the power supply; and After instructing the baseboard management controller to turn off or turn on any power supply unit in the power supply, the real-time conversion efficiency of the power supply is calculated again to determine whether the power supply can operate at the optimal conversion efficiency.

13. The rack system as claimed in claim 11, wherein the rack management controller is further configured to: When it is determined that the power supply is not operating at the optimal conversion efficiency, instructing the baseboard management controller to turn off or turn on a predetermined number of power supply units in the power supply according to a historical record of a maximum output power value of each power supply unit, wherein the historical record is related to the maximum output power value of each power supply unit; and After instructing the baseboard management controller to turn off or turn on the predetermined number of power supply units in the power supply, the real-time conversion efficiency of the power supply is calculated again to determine whether the power supply can operate at the optimal conversion efficiency.

14. The rack system as claimed in claim 11, wherein the rack management controller is further configured to: measuring an average output power of the power supply within a predetermined period of time; and The average output power is divided by the maximum output power value to obtain the real-time conversion efficiency of the power supply.

Images