Server power supply network design method and system

Abstract

The invention discloses a server power supply network design method and system, and belongs to the technical field of server power supplies, and the method comprises the steps: carrying out the redundancy design of a server power supply through employing dual-path or multi-path power supply input; distributing a server power supply to each server node; hardware resources are dynamically adjusted according to load requirements, and loads are dynamically distributed; the server power supply state is monitored in real time, and abnormity early warning and management and control are carried out according to the monitoring condition; an uninterruptible power supply and a generator are configured, and a power supply module is designed as an independent replaceable unit. According to the invention, the problem that the reliability, efficiency and manageability of the server power supply are low due to the fact that effective network design cannot be carried out on the server power supply in the prior art is solved. According to the invention, redundancy design is carried out on the server power supply, so that when one path of power supply of each server fails, the server can still operate normally, effective network design can be carried out on the server power supply, and the reliability, the efficiency and the manageability of the server power supply are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of server power supplies, and particularly to a server power network design method and system. Background Art

[0002] Server power network design refers to rationally planning and designing a power distribution network in a server system to ensure that the server can operate stably and efficiently.

[0003] Chinese Patent with Publication No. CN202906936U discloses a server power management system, which includes a controlled server. The controlled server includes a power supply; a power management bus interface; a baseboard management controller for collecting voltage values and energy consumption data, and the baseboard management controller is connected to the power supply through the power management bus interface; a CPU whose operating frequency is adjusted by the baseboard management controller, and the CPU is connected to the baseboard management controller; and a network interface connected to the CPU. The server power management system further includes a data center management device for controlling the controlled server, and this device is sequentially connected to the controlled server through a transmission network and a switch. The baseboard management controller collects voltage values and energy consumption data through the power management bus interface and sends the data to the data center management device, and the latter issues instructions to adjust the operating frequency of the CPU, so that the controlled server operates according to the specified energy consumption, achieving the purpose of reducing energy consumption without affecting the performance of the application system. However, this patent has the following defects:

[0004] The existing technology cannot effectively design the network of the server power supply, resulting in low reliability, operating efficiency, and manageability of the server power supply. Summary of the Invention

[0005] The purpose of the present invention is to provide a server power network design method and system, which can effectively design the network of the server power supply, improve the reliability, efficiency, and manageability of the server power supply, and solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A server power network design method includes:

[0008] Performing redundant design on the server power supply by using dual or multiple power inputs;

[0009] Distributing the server power supply to each server node, with each area independently powered;

[0010] Dynamically adjusting hardware resources according to load requirements, optimizing high-energy-consuming components, and dynamically distributing loads;

[0011] Monitor the server power status in real time, and issue abnormal warnings and control according to the monitoring situation;

[0012] Configure an uninterruptible power supply and a generator, and design the power module as an independent replaceable unit for online replacement of the power module;

[0013] Among them, compare the change ratios of the CPU usage rate and the memory usage rate, and determine the server load situation according to the comparison results, and then dynamically adjust the power distribution.

[0014] Preferably, a redundant design is adopted for the server power supply with dual or multiple power inputs, including:

[0015] According to the business requirements of the server power network design, clarify the reliability level that the server power network needs to achieve, and determine the server power network reliability goal;

[0016] Analyze the server power network according to the server power network reliability goal, and identify the single-point failures existing in the server power network, including single power modules and single power supply lines;

[0017] Select the redundancy level according to the identification results, and perform a redundant design on the server power network;

[0018] Among them, dual or multiple power inputs are adopted, and each server is configured with dual or multiple power inputs, so that the server can still run normally when one power supply of each server fails.

[0019] Preferably, distribute the server power supply to each server node, and each area is independently powered, including:

[0020] Calculate the total power demand of the server cluster, including the power consumption of each server and the reserved capacity for future expansion, determine the load demand, select single-phase or three-phase power supply according to the load demand, and determine the power supply method;

[0021] Select an intelligent power distribution unit according to the load demand, so that it can meet the current and future load demands, and determine the capacity of the power distribution unit;

[0022] According to the power interface types of the servers and devices, configure the number and type of output sockets of the power distribution unit, perform a redundant design on the power distribution unit, and configure a dual power distribution unit for critical devices;

[0023] Connect the servers to different power distribution units, so that when one power distribution unit fails, the servers can still be powered by the other power distribution unit, and divide the server cluster into multiple power supply areas, and each area is independently powered.

[0024] Preferably, the power distribution unit dynamically adjusts power distribution according to the load condition of the server, including:

[0025] Collecting the load data of the server in real time, where the load data includes current and voltage;

[0026] Performing data cleaning, denoising and normalization processing on the collected load data of the server to obtain the processed load data;

[0027] Monitoring the change amplitude ratio of the current and voltage in real time;

[0028] Comparing the change amplitude ratios corresponding to the current and voltage with their respective preset change amplitude ratio thresholds;

[0029] When the change amplitude ratios of the current and voltage both exceed their respective change amplitude ratio thresholds, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change;

[0030] When the change amplitude ratios of the current and voltage do not both exceed their respective change amplitude ratio thresholds, it is determined that the load of the server has not changed;

[0031] When only one of the change amplitude ratios of the current and voltage exceeds its respective change amplitude ratio threshold, the CPU usage rate and memory usage rate of the server are retrieved;

[0032] Comparing the change ratios of the CPU usage rate and memory usage rate;

[0033] When the percentage increase in the memory usage rate corresponding to each unit percentage increase in the CPU usage rate is not less than the unit percentage of the first proportionality coefficient, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change; where the value range of the unit percentage is 3%-8%;

[0034] When the percentage increase in the CPU usage rate corresponding to each unit percentage increase in the memory usage rate is not less than the unit percentage of the second proportionality coefficient, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change.

[0035] Preferably, the first proportionality coefficient and the second proportionality coefficient are set according to the change conditions of the current and voltage, including:

[0036] Retrieving the average value of the current change amplitude ratio and the average value of the voltage change amplitude ratio;

[0037] Comparing the average value of the voltage change amplitude ratio and the average value of the current change amplitude ratio to obtain a comparison result;

[0038] Retrieve the type of load data corresponding to the larger value among the average value of the voltage change amplitude ratio and the average value of the current change amplitude ratio as the reference data;

[0039] Extract the load data that exceeds its corresponding change amplitude ratio threshold as the target data;

[0040] Compare the type of the target data with the reference data;

[0041] When the target data and the reference data are of the same type, set the first proportionality coefficient and the second proportionality coefficient using the change amplitude ratio by which the target data exceeds its corresponding change amplitude ratio threshold;

[0042] Among them, the first proportionality coefficient and the second proportionality coefficient are obtained through the following formula:

[0043]

[0044] Among them, P 01 represents the first proportionality coefficient; P 02 represents the second proportionality coefficient; S ym represents the change amplitude ratio threshold corresponding to the target data; S m represents the change amplitude ratio corresponding to the target data; B represents a preset reference proportionality coefficient, and the value range of the reference proportionality coefficient is 0.52 - 0.65;

[0045] When the target data and the reference data are of different types, set the first proportionality coefficient and the second proportionality coefficient using the reference data;

[0046] Among them, the first proportionality coefficient and the second proportionality coefficient are obtained through the following formula:

[0047]

[0048] Among them, P 01 represents the first proportionality coefficient; P 02 represents the second proportionality coefficient; S yc represents the change amplitude ratio threshold corresponding to the reference data; S c represents the change amplitude ratio corresponding to the reference data; B represents a preset reference proportionality coefficient, and the value range of the reference proportionality coefficient is 0.52 - 0.65.

[0049] Preferably, dynamically adjust the hardware resources according to the load demand, optimize the high - energy - consuming components, and dynamically allocate the load, including:

[0050] Use a power meter or an intelligent power distribution unit to measure the total power consumption of the server cluster, calculate the current energy efficiency level, and determine the current energy efficiency;

[0051] Analyze the energy consumption of the server and power supply module to identify high-energy-consuming components;

[0052] Adopt a modular server design, dynamically adjust hardware resources according to load requirements, optimize high-energy-consuming components, dynamically allocate the load situation. When the load is low, switch some servers or power supply modules to the sleep mode, and when the load increases, automatically wake up the sleeping servers;

[0053] Adopt server virtualization technology to integrate multiple applications onto a small number of physical servers, and use dynamic resource scheduling tools to automatically adjust the distribution of virtual machines according to load requirements.

[0054] Preferably, conduct abnormal early warning and control according to the monitoring situation, including:

[0055] Based on intelligent acquisition devices, monitor and collect the voltage, current, power and temperature of the server power supply in real time to obtain real-time operation data of the server power supply;

[0056] Process the real-time operation data of the server power supply to determine the characteristic data of the server power supply operation;

[0057] Based on deep learning technology, use the historical operation data of the server power supply to train and optimize the deep learning model, and construct a server power supply fault prediction model;

[0058] Input the characteristic data of the server power supply operation into the server power supply fault prediction model, analyze the characteristic data of the server power supply operation based on the server power supply fault prediction model, and predict the server power supply fault to determine the server power supply fault prediction result;

[0059] When the server power supply anomaly is predicted, immediately issue an alarm and record the log, and take countermeasures in advance according to the server power supply fault prediction result to remotely manage and control the server power supply.

[0060] Preferably, configure an uninterruptible power supply and a generator, including:

[0061] Configure an uninterruptible power supply and a generator for the server according to the server power supply fault prediction result;

[0062] Through the configured uninterruptible power supply, when the main power supply fails, provide temporary power support through the uninterruptible power supply. At the same time, in the case of a long-term power outage, start the standby generator to provide power support through the standby generator to ensure the continuous operation of the server,

[0063] Among them, the operating state of the main power supply is monitored in real time, and the monitored operating state data of the main power supply is compared and analyzed with the preset standard data of the operating state of the main power supply. When the monitored operating state data of the main power supply is not within the standard data of the operating state of the main power supply, a fault warning is automatically triggered. When the main power supply fails, it automatically switches to the standby power supply to ensure the uninterrupted operation of the server.

[0064] Preferably, the power supply module is designed as an independent replaceable unit, including:

[0065] According to the prediction result of the server power supply failure, a modular hot-swappable power supply is configured for the server, and the modular hot-swappable power supply is designed as an independent replaceable unit;

[0066] When the modular hot-swappable power supply is in use, its usage performance is evaluated. When it is evaluated that the modular hot-swappable power supply cannot meet the usage requirements, the modular hot-swappable power supply is replaced online.

[0067] According to another aspect of the present invention, a server power supply network design system is provided for implementing a server power supply network design method as described above, including:

[0068] A power input module configured to perform redundant design of the server power supply by using dual or multiple power inputs according to the server power supply network reliability target;

[0069] A power distribution unit configured to distribute the server power supply to each server node, perform redundant design on the power distribution unit, and configure a dual power distribution unit for critical equipment;

[0070] A power management module configured to dynamically adjust hardware resources according to load requirements, optimize high-power-consuming components, and dynamically distribute loads;

[0071] A monitoring and alarm module configured to monitor the server power supply status in real time, issue an alarm when abnormal, and perform remote management and control;

[0072] A standby power supply module configured to configure an uninterruptible power supply and a generator for the server so that the server can still operate normally when the main power supply fails;

[0073] A hot-swappable power supply module configured to design the power supply module as an independent replaceable unit for online replacement of the power supply module.

[0074] Compared with the prior art, the beneficial effects of the present invention are:

[0075] 1. The present invention adopts dual - path or multi - path power input for redundant design of the server power supply, enabling the server to still operate normally when one path of the power supply of each server fails. An intelligent power distribution unit is selected according to the load demand, and redundant design is carried out on the power distribution unit. A dual - path power distribution unit is configured for key devices, and the server is connected to different power distribution units, so that when one power distribution unit fails, the server can still be powered by the other power distribution unit, and the hardware resources are dynamically adjusted according to the load demand, and the high - energy - consuming components are optimized, and the load is dynamically distributed.

[0076] 2. The present invention monitors the server power supply status in real - time, obtains the real - time operation data of the server power supply and processes it to determine the operation characteristic data of the server power supply. Based on the server power supply fault prediction model, the operation characteristic data of the server power supply is analyzed, and the server power supply fault is predicted to determine the server power supply fault prediction result. An uninterruptible power supply and a generator are configured, and the power module is designed as an independent replaceable unit for online replacement of the power module, and an effective network design can be carried out on the server power supply, improving the reliability, operation efficiency and manageability of the server power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 is an operation diagram of the server power supply network design method of the present invention;

[0078] Figure 2 is a module diagram of the server power supply network design system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0079] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0080] To solve the problem that the existing technology cannot effectively network - design the server power supply, resulting in low reliability, operation efficiency and manageability of the server power supply, please refer to Figure 1-2 , this embodiment provides the following technical solutions:

[0081] A server power supply network design system includes: a power input module, a power distribution unit, a power management module, a monitoring and alarm module, a backup power supply module and a hot - plug power module.

[0082] Among them, the power input module is configured to adopt dual - path or multi - path power input for redundant design of the server power supply according to the server power supply network reliability target;

[0083] Specifically, N+1 redundancy: An additional standby module is configured in the power supply module to ensure that the system can still operate normally in case of a single module failure.

[0084] Specifically, 2N redundancy: A completely independent dual system is configured for the entire power supply network to provide the highest level of reliability.

[0085] Among them, the power distribution unit is configured to distribute the server power to each server node. A redundant design is carried out for the power distribution unit, and a dual-channel power distribution unit is configured for critical equipment;

[0086] Specifically, an intelligent power distribution unit is used to distribute the power to each server, supporting remote monitoring and management. By dynamically allocating the load, single-channel power overload is avoided, the power life is extended, the server cluster is divided into multiple power supply areas, and each area is independently powered to reduce the impact range of single-point failures.

[0087] Among them, the power management module is configured to dynamically adjust the hardware resources according to the load demand, optimize the high-energy-consuming components, and dynamically allocate the load;

[0088] Specifically, high-efficiency power modules are selected to reduce energy loss. The voltage is dynamically adjusted according to the load condition to reduce energy consumption. When the load is low, some power modules are switched to the sleep mode to reduce unnecessary energy consumption. Through the intelligent power management module, the load is dynamically allocated to avoid single-channel power overload, extend the power life, adopt high-efficiency power technologies and load balancing strategies to reduce energy consumption and operating costs, and ensure the high availability of the server power supply through redundant design and intelligent management.

[0089] Among them, the monitoring and alarm module is configured to monitor the server power status in real time, issue an alarm in case of an abnormality, and perform remote management and control;

[0090] Specifically, the real-time monitoring and fault prediction functions improve the operation and maintenance efficiency and reduce the fault risk.

[0091] Among them, the standby power supply module is configured to configure an uninterruptible power supply and a generator for the server so that the server can still operate normally in case of a main power failure;

[0092] Specifically, an uninterruptible power supply is configured to provide temporary power support in case of a main power failure; in case of a long-term power outage, the standby generator is started to ensure the continuous operation of the server; when the main power fails, it automatically switches to the standby power supply to ensure the uninterrupted operation of the system.

[0093] Among them, the hot-swappable power module is configured to design the power module as an independent replaceable unit for online replacement of the power module.

[0094] Specifically, the power supply module is designed as a hot-swappable module, which is convenient for maintenance and upgrade, supports online replacement of the power supply module, and reduces maintenance downtime.

[0095] To better demonstrate the server power network design process, this embodiment provides a server power network design method, which is implemented based on the above server power network design system, including:

[0096] Redundantly design the server power supply by using dual or multiple power inputs;

[0097] Distribute the server power supply to each server node, and each area is independently powered;

[0098] Dynamically adjust the hardware resources according to the load requirements, optimize the high-power-consuming components, and dynamically allocate the load;

[0099] Real-time monitor the server power supply status, and give early warnings and control for abnormalities according to the monitoring situation;

[0100] Configure an uninterruptible power supply and a generator, and design the power supply module as an independent replaceable unit for online replacement of the power supply module.

[0101] In this embodiment, redundantly designing the server power supply by using dual or multiple power inputs includes: according to the business requirements of the server power network design, clarifying the reliability level that the server power network needs to achieve, and determining the server power network reliability target;

[0102] Analyze the server power network according to the server power network reliability target, and identify the single-point failures existing in the server power network, including single power supply modules and single power supply lines;

[0103] Select the redundancy level according to the identification situation, and redundantly design the server power network;

[0104] Among them, when using dual or multiple power inputs, configure dual or multiple power inputs for each server, so that the server can still operate normally when one power supply of each server fails.

[0105] In this embodiment, distributing the server power supply to each server node and independently powering each area includes:

[0106] Calculate the total power demand of the server cluster, including the power consumption of each server and the reserved capacity for future expansion, determine the load demand, select single-phase or three-phase power supply according to the load demand, and determine the power supply method;

[0107] Select an intelligent power distribution unit according to the load demand, so that it can meet the current and future load demands, and determine the capacity of the power distribution unit;

[0108] Configure the number and type of output sockets of the power distribution unit according to the power interface types of the server and the devices, perform redundant design on the power distribution unit, and configure dual power distribution units for critical devices;

[0109] Connect the servers to different power distribution units so that when one power distribution unit fails, the servers can still be powered by the other power distribution unit, and divide the server cluster into multiple power supply areas, with each area being independently powered.

[0110] Specifically, the power distribution unit dynamically adjusts power distribution according to the load conditions of the servers, including:

[0111] Collect the load data of the servers in real time, where the load data includes current and voltage;

[0112] Perform data cleaning, denoising, and normalization processing on the collected load data of the servers to obtain the processed load data;

[0113] Monitor the change amplitude ratios of the current and voltage in real time;

[0114] Compare the change amplitude ratios corresponding to the current and voltage with their respective preset change amplitude ratio thresholds;

[0115] When the change amplitude ratios of both the current and voltage exceed their respective change amplitude ratio thresholds, it is determined that the load of the servers has changed, and power distribution is dynamically adjusted according to the load change;

[0116] When the change amplitude ratios of both the current and voltage do not exceed their respective change amplitude ratio thresholds, it is determined that the load of the servers has not changed;

[0117] When only one of the change amplitude ratios of the current and voltage exceeds its corresponding change amplitude ratio threshold, retrieve the CPU usage rate and memory usage rate of the servers;

[0118] Compare the change ratios of the CPU usage rate and the memory usage rate;

[0119] When the percentage increase in the memory usage rate corresponding to each one-unit percentage increase in the CPU usage rate is not less than the unit percentage of the first proportionality coefficient, it is determined that the load of the servers has changed, and power distribution is dynamically adjusted according to the load change; where the value range of the one-unit percentage is 3% - 8%;

[0120] When the percentage increase in CPU usage corresponding to each unit percentage increase in memory usage is not less than the unit percentage of the second proportionality coefficient, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change.

[0121] The technical effects of the above technical solution are as follows: The above solution can monitor the load data (including current and voltage) of the server in real time and dynamically adjust the power distribution according to the load change. This ensures that the server can obtain sufficient power support when the load changes, thereby improving the power distribution efficiency. By precisely controlling the power distribution, it is possible to avoid wasting power resources when the load is low or insufficient power supply when the load is high, thus optimizing energy utilization. The above solution can detect load changes in a timely manner and prevent potential failures by monitoring the load data of the server in real time and performing data cleaning, denoising, and normalization processing to obtain accurate load data. By monitoring the change ratio of current and voltage and comparing it with a preset threshold to determine whether the load of the server has changed, potential failures can be prevented and the server can be quickly restored when a failure occurs, thereby enhancing the stability of the server. When only one of the change ratio of current and voltage exceeds the threshold, the above solution will retrieve the CPU usage and memory usage of the server and compare their change ratios to determine the load change. This comprehensive load determination method helps to more accurately evaluate the load situation of the server. Dynamically adjusting the power distribution according to the load change ensures that the server can obtain sufficient power support under different load conditions, which helps to optimize resource utilization and avoid resource waste or overload. By dynamically adjusting the power distribution, it is ensured that the server can obtain sufficient power support when the load changes, thereby avoiding performance degradation. At the same time, it can effectively improve the response speed of the server, thereby enhancing the overall performance. By precisely controlling the power distribution, avoiding resource waste and insufficient power supply can significantly reduce energy consumption. At the same time, it can effectively extend the service life of the server hardware, reduce the hardware failure rate, thereby reducing maintenance costs and downtime.

[0122] Specifically, setting the first proportionality coefficient and the second proportionality coefficient according to the change conditions of current and voltage includes:

[0123] Retrieve the average value of the current change ratio and the average value of the voltage change ratio;

[0124] Compare the average value of the voltage change ratio and the average value of the current change ratio to obtain a comparison result;

[0125] Retrieve the type of load data corresponding to the larger value among the average value of the voltage change ratio and the average value of the current change ratio as the reference data;

[0126] Extract the load data that exceeds its corresponding change amplitude ratio threshold as the target data;

[0127] Compare the type of the target data with the reference data;

[0128] When the target data and the reference data are of the same type, set the first proportionality coefficient and the second proportionality coefficient by using the change amplitude ratio by which the target data exceeds its corresponding change amplitude ratio threshold;

[0129] Among them, the first proportionality coefficient and the second proportionality coefficient are obtained through the following formula:

[0130]

[0131] Among them, P 01 represents the first proportionality coefficient; P 02 represents the second proportionality coefficient; S ym represents the change amplitude ratio threshold corresponding to the target data; S m represents the change amplitude ratio corresponding to the target data; B represents a preset reference proportionality coefficient, and the value range of the reference proportionality coefficient is 0.52 - 0.65;

[0132] When the target data and the reference data are of different types, set the first proportionality coefficient and the second proportionality coefficient by using the reference data;

[0133] Among them, the first proportionality coefficient and the second proportionality coefficient are obtained through the following formula:

[0134]

[0135] Among them, P 01 represents the first proportionality coefficient; P 02 represents the second proportionality coefficient; S yc represents the change amplitude ratio threshold corresponding to the reference data; S c represents the change amplitude ratio corresponding to the reference data; B represents a preset reference proportionality coefficient, and the value range of the reference proportionality coefficient is 0.52 - 0.65.

[0136] The technical effects of the above technical solution are as follows: By retrieving the average value of the current voltage change amplitude ratio and the average value of the current change amplitude ratio, and comparing them, selecting the load data type with the larger value as the reference data can effectively improve the accuracy of the proportional coefficient setting. Since the comprehensive change conditions of the current and voltage are considered, the limitations of a single data source are avoided. The above solution further extracts the load data exceeding its corresponding change amplitude ratio threshold as the target data, and compares the type of the target data with the reference data, ensuring that the setting of the proportional coefficient is more in line with the actual load change conditions and improving the setting accuracy. According to the type comparison result of the target data and the reference data, the above solution flexibly uses the change amplitude ratio of the target data or the reference data to set the proportional coefficient. This flexibility enhances the adaptive ability of the system, enabling it to better cope with different load change conditions and maintain the stability and performance of the system. By real-time monitoring the changes in current and voltage, the above solution can provide real-time feedback and adjust the proportional coefficient. This real-time nature ensures that the system can respond promptly to load changes and further enhances the adaptive ability of the system. By dynamically adjusting the proportional coefficient, the above solution can optimize the power distribution efficiency. Dynamically adjusting the power distribution according to the load change conditions ensures that the server can obtain sufficient power support when the load changes, avoiding power waste and insufficient power supply. Optimizing the power distribution efficiency also helps to improve the energy efficiency ratio, reduce the operating cost and environmental impact. By precisely controlling the power distribution and reducing unnecessary energy consumption, it conforms to the development trend of green data centers. By real-time monitoring and adjusting the proportional coefficient, the above solution can prevent potential failures, detect load changes in a timely manner and adjust the power distribution, avoiding situations such as overload or insufficient power supply, and improving the reliability of the system. Dynamically adjusting the proportional coefficient also helps to reduce the fluctuations and uncertainties of the system. Maintaining a stable power supply and load balance helps to improve the overall performance and stability of the system.

[0137] In this embodiment, dynamically adjusting the hardware resources according to the load requirements, optimizing the high-power-consuming components, and dynamically distributing the load includes:

[0138] Using a power meter or an intelligent power distribution unit to measure the total power consumption of the server cluster, calculating the current energy efficiency level, and determining the current energy efficiency;

[0139] Analyzing the energy consumption of the server and the power supply module to identify high-power-consuming components;

[0140] Adopting a modular server design, dynamically adjusting the hardware resources according to the load requirements, optimizing the high-power-consuming components, and dynamically distributing the load. When the load is low, switch some servers or power supply modules to the sleep mode, and when the load increases, automatically wake up the sleeping servers;

[0141] Adopt server virtualization technology to integrate multiple applications onto a few physical servers, and use dynamic resource scheduling tools to automatically adjust the distribution of virtual machines according to load requirements.

[0142] In this embodiment, perform anomaly warning and control according to the monitoring situation, including:

[0143] Based on intelligent acquisition devices, monitor and collect the voltage, current, power, and temperature of the server power supply in real time, and obtain real-time operation data of the server power supply;

[0144] Process the real-time operation data of the server power supply to determine the operation characteristic data of the server power supply;

[0145] Based on deep learning technology, use the historical operation data of the server power supply to train and optimize the deep learning model, and construct a server power supply fault prediction model;

[0146] Input the operation characteristic data of the server power supply into the server power supply fault prediction model, analyze the operation characteristic data of the server power supply based on the server power supply fault prediction model, and predict the server power supply fault to determine the server power supply fault prediction result;

[0147] When it is predicted that the server power supply is abnormal, immediately issue an alarm and record a log, and take preventive measures in advance according to the server power supply fault prediction result to remotely manage and control the server power supply.

[0148] In this embodiment, configure an uninterruptible power supply and a generator, including:

[0149] According to the server power supply fault prediction result, configure an uninterruptible power supply and a generator for the server;

[0150] Through the configured uninterruptible power supply, when the main power supply fails, provide temporary power support through the uninterruptible power supply. At the same time, in the case of a long-term power outage, start the standby generator to provide power support through the standby generator to ensure the continuous operation of the server;

[0151] Among them, monitor the operation status of the main power supply in real time, and compare and analyze the monitored main power supply operation status data with the preset main power supply operation status standard data. When the monitored main power supply operation status data is not within the main power supply operation status standard data, automatically trigger a fault warning. When the main power supply fails, automatically switch to the standby power supply to keep the server running uninterruptedly.

[0152] In this embodiment, design the power module as an independent replaceable unit, including:

[0153] According to the prediction result of the server power failure, configure a modular hot-swappable power supply for the server, and design the modular hot-swappable power supply as an independent replaceable unit;

[0154] When the modular hot-swappable power supply is in use, evaluate the performance of the modular hot-swappable power supply. When it is evaluated that the modular hot-swappable power supply cannot meet the usage requirements, perform an online replacement of the modular hot-swappable power supply.

[0155] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0156] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A server power supply network design method, characterized in that, Including: Redundantly design the server power supply by using dual or multiple power inputs; Distribute the server power supply to each server node, with each area independently powered; Dynamically adjust hardware resources according to load requirements, optimize high-power-consuming components, and dynamically allocate loads; Real-time monitor the server power supply status, and give early warnings and control for abnormalities according to the monitoring situation; Configure an uninterruptible power supply and a generator, and design the power supply module as an independent replaceable unit for online replacement of the power supply module; Among them, compare the change ratios of the CPU usage rate and the memory usage rate, and determine the load situation of the server according to the comparison result, and then dynamically adjust the power supply distribution.

2. The server power network design method according to claim 1, wherein, Redundantly design the server power supply by using dual or multiple power inputs, including: According to the business requirements of the server power supply network design, clarify the reliability level that the server power supply network needs to achieve, and determine the reliability target of the server power supply network; Analyze the server power supply network according to the server power supply network reliability target, and identify single points of failure in the server power supply network, including single power supply modules and single power supply lines; Select the redundancy level according to the identification result, and redundantly design the server power supply network; Among them, use dual or multiple power inputs, configure dual or multiple power inputs for each server, so that the server can still run normally when one power supply of the server fails.

3. The server power supply network design method according to claim 1, characterized in that Distribute the server power supply to each server node, with each area independently powered, including: Calculate the total power demand of the server cluster, including the power consumption of each server and the reserved capacity for future expansion, determine the load demand, select single-phase or three-phase power supply according to the load demand, and determine the power supply method; Select an intelligent power distribution unit according to the load demand, so that it can meet the current and future load demands, and determine the capacity of the power distribution unit; According to the power interface types of the servers and devices, configure the number and types of output sockets of the power distribution unit, redundantly design the power distribution unit, and configure a dual-channel power distribution unit for key devices; Connect the servers to different power distribution units, so that when one power distribution unit fails, the servers can still be powered by the other power distribution unit, and divide the server cluster into multiple power supply areas, with each area independently powered.

4. The server power supply network design method according to claim 3, characterized in that The power distribution unit dynamically adjusts the power distribution according to the load situation of the server, including: Real-time collect the load data of the server, where the load data includes current and voltage; Perform data cleaning, denoising, and normalization processing on the collected load data of the server to obtain the processed load data; Real-time monitor the change amplitude ratios of the current and voltage; Respectively compare the change amplitude ratios corresponding to the current and voltage with their corresponding preset change amplitude ratio thresholds; When the change amplitude ratios of the current and voltage both exceed their corresponding change amplitude ratio thresholds, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change; When the change amplitude ratios of the current and voltage both do not exceed their corresponding change amplitude ratio thresholds, it is determined that the load of the server has not changed; When only one of the change amplitude ratios of the current and voltage exceeds its corresponding change amplitude ratio threshold, the CPU usage rate and memory usage rate of the server are retrieved; The change ratios of the CPU usage rate and memory usage rate are compared; When the percentage increase in the memory usage rate corresponding to each one-unit percentage increase in the CPU usage rate is not less than the one-unit percentage of the first proportionality coefficient, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change; wherein, the value range of the one-unit percentage is 3%-8%; When the percentage increase in the CPU usage rate corresponding to each one-unit percentage increase in the memory usage rate is not less than the one-unit percentage of the second proportionality coefficient, it is determined that the load of the server has changed, and the power distribution is dynamically adjusted according to the load change.

5. The server power supply network design method according to claim 4, characterized in that Setting the first proportionality coefficient and the second proportionality coefficient according to the change conditions of the current and voltage includes: Retrieving the average value of the voltage change amplitude ratio and the average value of the current change amplitude ratio; Comparing the average value of the voltage change amplitude ratio and the average value of the current change amplitude ratio to obtain a comparison result; Retrieving the type of load data corresponding to the larger value among the average value of the voltage change amplitude ratio and the average value of the current change amplitude ratio as the reference data; Extracting the load data that exceeds its corresponding change amplitude ratio threshold as the target data; Comparing the type of the target data with the reference data; When the target data and the reference data are of the same type, the first proportionality coefficient and the second proportionality coefficient are set using the change amplitude ratio by which the target data exceeds its corresponding change amplitude ratio threshold; When the target data and the reference data are of different types, the first proportionality coefficient and the second proportionality coefficient are set using the reference data.

6. The server power supply network design method according to claim 1, characterized in that, Dynamically adjusting the hardware resources according to the load requirements, optimizing the high-energy-consuming components, and dynamically distributing the load, including: Measuring the total power consumption of the server cluster using a power meter or an intelligent power distribution unit, calculating the current energy efficiency level, and determining the current energy efficiency; Analyzing the energy consumption of the server and the power supply module to identify high-energy-consuming components; Adopting a modular server design, dynamically adjusting the hardware resources according to the load requirements, optimizing the high-energy-consuming components, dynamically distributing the load situation, switching some servers or power supply modules to the sleep mode under low load, and automatically waking up the sleeping servers when the load increases; Adopting server virtualization technology to integrate multiple applications onto a small number of physical servers, and using a dynamic resource scheduling tool to automatically adjust the virtual machine distribution according to the load requirements.

7. The server power network design method according to claim 1, characterized in that Conducting abnormal warning and control according to the monitoring situation, including: Based on intelligent acquisition devices, real-time monitoring and acquisition of the voltage, current, power, and temperature of the server power supply are carried out to obtain real-time operation data of the server power supply; Processing the real-time operation data of the server power supply to determine the characteristic data of the server power supply operation; Based on deep learning technology, using the historical operation data of the server power supply to train and optimize the deep learning model, and constructing a server power supply fault prediction model; Input the server power operation characteristic data into the server power failure prediction model, analyze the server power operation characteristic data based on the server power failure prediction model, predict the server power failure, and determine the server power failure prediction result; When a server power anomaly is predicted, immediately issue an alarm and record a log, and take preventive measures in advance according to the server power failure prediction result to remotely manage and control the server power.

8. The server power network design method according to claim 7, characterized in that, Configure an uninterruptible power supply and a generator, including: According to the server power failure prediction result, configure an uninterruptible power supply and a generator for the server; Through the configured uninterruptible power supply, when the main power fails, provide temporary power support through the uninterruptible power supply. At the same time, in the case of a long-term power outage, start the standby generator to provide power support through the standby generator to ensure the continuous operation of the server. Among them, the operation status of the main power supply is monitored in real time, and the monitored main power supply operation status data is compared and analyzed with the preset main power supply operation status standard data. When the monitored main power supply operation status data is not within the main power supply operation status standard data, a fault warning is automatically triggered. When the main power supply fails, it automatically switches to the standby power supply to keep the server running uninterruptedly.

9. The server power network design method according to claim 8, characterized in that Design the power module as an independent replaceable unit, including: According to the server power failure prediction result, configure a modular hot-swappable power supply for the server, and design the modular hot-swappable power supply as an independent replaceable unit; When the modular hot-swappable power supply is in use, evaluate the performance of the modular hot-swappable power supply. When it is evaluated that the modular hot-swappable power supply cannot meet the usage requirements, perform an online replacement of the modular hot-swappable power supply.

10. A server power supply network design system for implementing a server power supply network design method as described in claim 9, characterized in that, Including: A power input module configured to adopt dual or multiple power inputs for redundant design of the server power supply according to the server power network reliability target; A power distribution unit configured to distribute the server power supply to each server node, perform redundant design on the power distribution unit, and configure a dual power distribution unit for critical equipment; A power management module configured to dynamically adjust hardware resources according to load requirements, optimize high-power-consuming components, and dynamically distribute loads; A monitoring and alarm module configured to monitor the server power supply status in real time, issue an alarm when abnormal, and perform remote management and control; A standby power supply module configured to configure an uninterruptible power supply and a generator for the server, so that the server can still operate normally when the main power supply fails; A hot-swappable power supply module configured to design the power module as an independent replaceable unit for online replacement of the power module.

Images