Load balancing method and system for power safety of data center

Abstract

The invention discloses a load balancing method and system for power safety of a data center, and the method comprises the steps: constructing a power space tree based on the hierarchical relation of power equipment in the data center; respectively obtaining a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively carrying out data processing on the first power value and the first load state to obtain a second power value and a second load state; respectively analyzing the second power value and the second load state, determining whether a server in the data center has a security risk, and if the server has the security risk, automatically generating a corresponding scheduling task; and based on the scheduling task, carrying out load balancing on the server with the security risk in the data center. According to the invention, the accuracy and efficiency of power safety load balancing of the data center can be improved, the power safety is improved, and refinement of power regulation and control is promoted.

Description

Technical Field

[0001] The present application relates to the field of power management, and in particular to a load balancing method and system for power safety in a data center. Background Art

[0002] As an important cornerstone of the information age, data centers are responsible for the storage, processing and transmission of massive amounts of data. Their operational stability and security are directly related to all aspects of corporate operations, social services and even national security.

[0003] At present, most data centers rely on advanced power environment monitoring systems (referred to as power environment systems) to monitor the status of power equipment. When the power safety threshold is exceeded, the power alarm function of the system is used to trigger manual intervention to adjust the power. This kind of manual intervention is prone to power adjustment leakage and operational errors. 、 There are many hidden dangers such as low accuracy and low efficiency. Moreover, the power environment monitoring system can only monitor whether the current and voltage values ​​of the equipment are within the safety threshold, and cannot perform automatic scheduling and regulation on the system. Relying on manual regulation cannot respond quickly and is inefficient.

[0004] Application Contents

[0005] The present application provides a load balancing method and system for power safety in a data center, which can improve the accuracy and efficiency of power safety load balancing in a data center, enhance power safety, and promote the refinement of power regulation.

[0006] In a first aspect, the present application provides a load balancing method for power safety in a data center, comprising:

[0007] Based on the hierarchical relationship of power equipment in the data center, a power space tree is constructed;

[0008] Respectively obtaining a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively performing data processing on the first power value and the first load state to obtain a second power value and a second load state, wherein each node of the power space tree has a matching server;

[0009] Analyze the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center, and automatically generate a corresponding scheduling task if there is a security risk;

[0010] Based on the scheduling task, load balancing is performed on servers with security risks in the data center.

[0011] The embodiment of the present application constructs a power space tree based on the hierarchical relationship of power equipment in a data center, so as to facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, and then facilitate the judgment whether there is a security risk; by respectively obtaining the first power value corresponding to each node in the power space tree and the first load status corresponding to each server, it is convenient to quickly and accurately determine the servers with security risks in the data center based on the first power value and the first load status corresponding to each server; by respectively analyzing the second power value and the second load status, it is possible to determine whether there is a security risk for the servers in the data center, so as to facilitate the subsequent load balancing thereof; by scheduling tasks to load balance servers with security risks, the accuracy and efficiency of power safety load balancing in the data center can be improved, power safety can be improved, and the refinement of power regulation can be promoted.

[0012] Furthermore, the power space tree is constructed based on the hierarchical relationship of power equipment in the data center, specifically:

[0013] Decompose the power equipment in the data center to obtain some equipment information;

[0014] Based on the device information, create device nodes corresponding to several levels;

[0015] Establishing association relationships between the device nodes at different levels;

[0016] Based on the device nodes and the association relationships, a power space tree is constructed.

[0017] In this way, by first decomposing the power equipment in the data center, we can understand the hierarchical situation in the data center, facilitate the subsequent construction of a power space tree based on power equipment at different levels, and facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, thereby facilitating the judgment of whether there is a security risk.

[0018] Further, the data processing is performed on the first power value and the first load state respectively to obtain the second power value and the second load state, specifically:

[0019] Performing data cleaning on the first power value and the first load state respectively to obtain a first cleaning result and a second cleaning result;

[0020] Summarizing the first cleaning result and the second cleaning result by time respectively to obtain a first summary result and a second summary result;

[0021] The first summary result and the second summary result are classified according to preset classification standards to obtain a second power value and a second load state.

[0022] In this way, by performing data cleaning, data aggregation and data classification on the first power value and the first load status respectively, the quality and availability of the data can be ensured, which facilitates subsequent data extraction, analysis and decision-making.

[0023] Further, the second power value and the second load state are analyzed respectively to determine whether there is a security risk for the server in the data center, specifically:

[0024] Performing statistical analysis on the second power value to determine a risk time period in which each node has a security risk;

[0025] The second load state corresponding to each node in the risk time period is analyzed to determine that a corresponding server in the data center has a security risk.

[0026] In this way, by analyzing the second power value, the risk time period when there is a security risk can be determined, and the load status during the risk time period can be analyzed, which can effectively reduce the difficulty of data analysis and more quickly determine whether there is a security risk in the server in the data center.

[0027] Furthermore, the determination of whether there is a security risk in the server in the data center is specifically as follows:

[0028] Pre-set the security threshold corresponding to each node in the data center;

[0029] If the second power value or the second load state corresponding to each node exceeds a safety threshold, it is determined that a corresponding server in the data center has a safety risk.

[0030] In this way, by setting the safety threshold in advance, if the second power value or the second load state corresponding to each node exceeds the safety threshold, it can be effectively determined that there are servers with security risks in the data center.

[0031] In a second aspect, the present application provides a load balancing system for power security in a data center, including: a construction module, an acquisition module, an analysis module, and a balancing module;

[0032] The construction module is used to construct a power space tree based on the hierarchical relationship of power equipment in the data center;

[0033] The acquisition module is used to respectively acquire a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively perform data processing on the first power value and the first load state to obtain a second power value and a second load state, wherein each node of the power space tree has a matching server;

[0034] The analysis module is used to analyze the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center, and if there is a security risk, automatically generate a corresponding scheduling task;

[0035] The balancing module is used to perform load balancing on servers with security risks in the data center based on the scheduling tasks.

[0036] The embodiment of the present application constructs a power space tree based on the hierarchical relationship of power equipment in a data center, so as to facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, and then facilitate the judgment whether there is a security risk; by respectively obtaining the first power value corresponding to each node in the power space tree and the first load status corresponding to each server, it is convenient to quickly and accurately determine the servers with security risks in the data center based on the first power value and the first load status corresponding to each server; by respectively analyzing the second power value and the second load status, it is possible to determine the servers with security risks in the data center, so as to facilitate the subsequent load balancing thereof; by scheduling tasks to load balance servers with security risks, the accuracy and efficiency of power safety load balancing in the data center can be improved, power safety can be improved, and the refinement of power regulation can be promoted.

[0037] Further, the construction module includes: a decomposition unit, a first processing unit, a second processing unit and a construction unit;

[0038] The decomposition unit is used to decompose the power equipment in the data center to obtain a number of equipment information;

[0039] The first processing unit is used to create device nodes corresponding to a plurality of levels based on the plurality of device information;

[0040] The second processing unit is used to establish association relationships between the device nodes at different levels;

[0041] The construction unit is used to construct a power space tree based on the device nodes and the association relationships.

[0042] Further, the acquisition module includes: a cleaning unit, a summarizing unit and a classification unit;

[0043] The cleaning unit is used to clean the first power value and the first load state respectively to obtain a first cleaning result and a second cleaning result;

[0044] The summarizing unit is used to summarize the first cleaning result and the second cleaning result by time to obtain a first summary result and a second summary result;

[0045] The classification unit is used to classify the first summary result and the second summary result according to preset classification standards to obtain a second power value and a second load state.

[0046] Further, the analysis module includes: a first analysis unit and a second analysis unit;

[0047] The first analysis unit is used to perform statistical analysis on the second power value to determine a risk time period in which each node has a security risk;

[0048] The second analysis unit is used to analyze the second load status corresponding to each node in the risk time period to determine whether the corresponding server in the data center has a security risk.

[0049] Furthermore, the analysis module further includes: a setting unit and a judgment unit:

[0050] The setting unit is used to pre-set the security threshold corresponding to each node in the data center;

[0051] The judgment unit is used to determine that a corresponding server in the data center has a security risk if the second power value or the second load state corresponding to each node exceeds a safety threshold. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flow chart of an embodiment of a load balancing method for power safety in a data center provided by the present application;

[0053] Figure 2 It is a hierarchical relationship diagram of the hierarchical relationship of power equipment provided by this application;

[0054] Figure 3 It is a structural diagram of an embodiment of a load balancing system for power safety in a data center provided by the present application. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0056] It should be understood that the step numbers used in this article are only for the convenience of description and are not intended to limit the order in which the steps are executed.

[0057] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an" and "the" are intended to include plural forms unless the context clearly indicates otherwise.

[0058] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0059] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0060] Data centers are the cornerstone of the information age, carrying the task of storing, processing and transmitting massive amounts of data. Their stable operation and safety are directly related to corporate operations, social services and national security. Currently, most data centers rely on advanced power environment monitoring systems (power environment systems) to monitor power status and rely on system alarms to trigger manual intervention. However, this model faces hidden dangers such as low accuracy and low efficiency.

[0061] Embodiment 1

[0062] Please refer to Figure 1 , Figure 1 This is a flow chart of an embodiment of a load balancing method for power safety in a data center provided by the present application, including steps S1 to S4:

[0063] Step S1: construct a power space tree based on the hierarchical relationship of power equipment in the data center;

[0064] It is understandable that there are many kinds of power equipment in the data center, including mains power supply, transformer, UPS, cabinet, cabinet and air conditioner. There are hierarchical relationships among the various power equipment, and these hierarchical relationships will be recorded in the power planning drawings of the data center. The hierarchical relationship diagram of the power equipment hierarchical relationship of the data center is as follows: Figure 2 As shown, generally speaking, the AC power supply will be connected to multiple transformers, and each transformer will carry multiple UPS, the UPS will also carry multiple cabinets, the cabinets will carry multiple cabinets, and the cabinets will carry multiple servers.

[0065] It is understandable that the power equipment in the data center can be directly decomposed or the power equipment information and the hierarchical relationship of the power equipment in the data center can be mastered according to the power planning drawings of the data center. If the power equipment in the data center is directly decomposed, the specific steps are: first, directly decompose according to the power equipment situation in the data center, master the power equipment information and the hierarchical relationship of the power equipment in the data center; second, import several power equipment information into the power management system, form power equipment nodes corresponding to different levels in the power management system, and add corresponding equipment information to each equipment node, such as equipment basic information, equipment specifications and equipment power safety thresholds; third, according to the relationship between power equipment, establish an association relationship between the equipment nodes at different levels, i.e., upper and lower levels; finally, construct a power space tree based on the equipment nodes and the association relationship. If the power equipment information and the hierarchical relationship of the power equipment in the data center are mastered according to the power planning drawings of the data center, the specific steps are: obtain the power planning drawings of the data center, import the power planning drawings into the power management system, name each node and write out the specification attributes, and add a description field to each node. For example, for example, if the cabinet is the lowest level setting, you need to enter the corresponding cabinet information under the cabinet category, and at the same time, you need to associate which cabinet is its parent, that is, which cabinet is its parent node. At the same time, the cabinet is also associated with which UPS is its parent, and so on, until it is associated with the final AC power supply.

[0066] It should be noted that after the corresponding device information is entered into each device node in the power space tree, a unique ID will be automatically generated. Based on this unique ID, the corresponding device in the data center can be quickly queried.

[0067] It should be noted that the algorithm steps for constructing the power space tree include: (1) defining the device node class, where each device node contains attributes such as a unique ID, name, parent node, and child node; (2) creating device nodes at each level based on the power hierarchy in the data center; (3) associating each node in the order from the mains to the cabinet according to the power topology; (4) forming a tree structure based on the device nodes and the association relationship to facilitate subsequent power data management and analysis.

[0068] In this way, by first decomposing the power equipment in the data center, we can understand the hierarchical situation in the data center, facilitate the subsequent construction of a power space tree based on power equipment at different levels, and facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, thereby facilitating the judgment of whether there is a security risk.

[0069] Step S2, respectively obtaining a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively performing data processing on the first power value and the first load state to obtain a second power value and a second load state, wherein each node of the power space tree has a matching server;

[0070] It can be understood that after the power space tree is constructed, the power situation of each node in the power space tree can be grasped in real time; specifically, first, the dynamic ring system collects data from each node in the power space tree at a preset time frequency to obtain a first power value, wherein the first power value includes but is not limited to the current, voltage, power, electricity and other related power information of the equipment; secondly, the abnormal equipment power values ​​in the first power value are cleaned to obtain a first cleaning result; thirdly, the first cleaning result is summarized by time to obtain a first summary result, wherein the time in the time summary can be but not limited to hourly, daily and weekly, etc.; finally, the first summary result is classified according to a preset classification standard to obtain a second power value, and the second power value is stored in the corresponding storage table in the power management system, wherein the preset classification standard can be but not limited to equipment type and area, etc. Of course, the load status corresponding to each server in the power space tree can also be mastered in real time; specifically, first, the server software management system collects the load status of all servers in the data center in real time at a preset time frequency to obtain a first load status, wherein the server management system includes but is not limited to: open source systems such as Zabbix and Nagios. When Zabbix is ​​used to collect data, the Zabbix system is connected to the data center management system DCOS through the api interface to transmit the load status to the data center management system. The load status includes but is not limited to network bandwidth utilization, CPU utilization, memory utilization and process status; secondly, the missing part of the data in the first load status is removed or filled, and the abnormal value in the first load status is removed or corrected to obtain a second cleaning result; thirdly, the second cleaning result is summarized by time to obtain a second summary result, wherein the time in the time summary can be but is not limited to: hourly, daily and weekly; finally, the second summary result is classified according to the preset classification standard to obtain a second load state, and the second load state is stored in the corresponding storage table in the power management system according to different dimensions, wherein the preset classification standard can be but is not limited to device type, region, server ID and process status.

[0071] It should be noted that the first power value is proportional to the server load state. When the server's business load increases, its CPU usage rate increases, so the power consumption value also increases. The power consumption value of the parent node corresponds to the load of all servers in its child nodes. For example, the power dynamic value of the head cabinet corresponds to the load of all servers in all its child nodes (cabinets). Therefore, when the business of one or more servers under a node of the space tree increases, the power consumption value of the related nodes will also increase, thereby affecting the power safety of the space node.

[0072] It should be noted that the preset time frequency can be set as needed, which can be 3 minutes, 5 minutes or even other time frequencies, and this application does not impose any restrictions on this.

[0073] It should be noted that there may be abnormal equipment power values ​​in the operation and maintenance of the data center, so data cleaning is required. Specifically, if the power value of some unactivated equipment is 0 during the sampling period, then the power value of the equipment is not used as the basis for calculation and needs to be eliminated. Similarly, if the power value of some equipment is abnormal during a certain test phase, cleaning processing is required to reduce the noise of the data.

[0074] It should be noted that summarizing the first cleaning results by time or category can effectively analyze power usage, identify peak load time periods, and optimize power resource allocation. Classifying and storing data can ensure orderly data management and efficient query.

[0075] In this way, by performing data cleaning, data aggregation and data classification on the first power value and the first load status respectively, the quality and availability of the data can be ensured, which facilitates subsequent data extraction, analysis and decision-making.

[0076] Step S3: Analyze the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center, and if there is a security risk, automatically generate a corresponding scheduling task;

[0077] It can be understood that, first, by calculating the average value, variance, standard deviation, maximum value and minimum value of the second power value in a certain time period, and based on the calculation results, comprehensively determine the risk time period in which there is a security risk in the node, that is, which servers may have some overload conditions in which time periods. Specifically, by calculating the average value of the second power value corresponding to a single node under the power space tree in a period of time, it is determined whether the node is stable in a period of time, thereby reducing the interference of emergencies; calculating the variance and standard deviation of the second power value in a period of time to determine the fluctuation relative to the average value; by calculating the maximum value and minimum value of a single node in a period of time, the peak and valley of the fluctuation are determined. Exemplarily, the ups node in the power space tree is loaded with hundreds of cabinets. During the time periods of 12 noon and 20-21 pm, there are more people playing games, so the server business volume loaded by the ups increases, and the power consumption will also increase. Therefore, it is necessary to detect power safety at all times. Through the above method of calculating power consumption, it can be comprehensively determined which time periods will be peak periods of business and which time periods will be trough periods of business. In the future, when there are overloaded devices during peak hours, the devices can be load balanced. For example, when a UPS is overloaded, the service must be regulated through the server load balancing, and the service must be adjusted to other UPS load servers to keep the UPS load value within the safety threshold.

[0078] It should be noted that this application does not limit the order of calculation methods such as mean, variance, standard deviation, maximum and minimum values, and the calculation method is not the focus of this application, so it will not be expanded here.

[0079] When a node is abnormal, its corresponding server will be abnormal, and the server corresponding to its parent node will also be abnormal, so multiple abnormal information will be received, and due to the conduction effect of the power space tree, a piece of error will appear, so it is necessary to analyze the load status of the server, that is, it is necessary to decompose according to the power value of each node to find the node with real problems. Specifically, after the risk time period is determined, the second load state corresponding to the risk time period can be clustered and analyzed to predict the risk of different nodes in the power space tree. At the same time, the regression algorithm is combined to check the load of each node in different time periods. When the power value or load state corresponding to each node exceeds the pre-set safety threshold, the server with security risk in the data center is determined. Of course, it is also possible to perform machine learning on the operation status and analysis rules of all nodes in the power space tree in advance to train the corresponding prediction model to predict the future risk points of different nodes in the space tree, and visualize the corresponding risk analysis and real-time alarm related content, where the visualization method includes but is not limited to charts and reports.

[0080] It should be noted that cluster analysis and regression algorithm are not the focus of this application, so they will not be expanded here.

[0081] It should be noted that the device itself has a safe power value, which is generally not allowed to be exceeded, otherwise it is easy to cause safety risks. Therefore, the safe power value can be multiplied by a certain coefficient as a safety threshold. When the device power value exceeds the safety threshold, the device location and the location of the space tree node to which it belongs will be sent to the operation and maintenance personnel for an alarm.

[0082] In this way, by analyzing the second power value, the risk time period when there is a security risk can be determined, and the load status during the risk time period can be analyzed, which can effectively reduce the difficulty of data analysis and more quickly determine whether there is a security risk in the server in the data center.

[0083] Finally, after the server with security risks is determined, a scheduling task can be automatically generated and sent to the load balancing system, which will help to subsequently load balance the corresponding servers; the scheduling task will include but is not limited to the server name, cup load rate, memory load rate, the server's corresponding security threshold, the server's location in the space tree, and other related information.

[0084] Step S4: Based on the scheduling task, load balancing is performed on servers with security risks in the data center.

[0085] It is understandable that after receiving the scheduling task, the server can be directional adjusted through the load balancing system. Specifically, the power value can be dispersed to other servers with smaller loads to ensure that the current load status of all servers meets the requirements of the safety threshold, thereby achieving load balancing. For example, when the power value of a node in the power space tree only reaches 30% or less of the same service servers, these servers are the target servers for adjusting the service. By distributing 20% ​​to 30% of the excess server services to these servers, load balancing can be achieved.

[0086] The embodiment of the present application constructs a power space tree based on the hierarchical relationship of power equipment in a data center, so as to facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, and then facilitate the judgment whether there is a security risk; by respectively obtaining the first power value corresponding to each node in the power space tree and the first load status corresponding to each server, it is convenient to quickly and accurately determine the servers with security risks in the data center based on the first power value and the first load status corresponding to each server; by respectively analyzing the second power value and the second load status, it is possible to determine the servers with security risks in the data center, so as to facilitate the subsequent load balancing thereof; by scheduling tasks to load balance servers with security risks, the accuracy and efficiency of power safety load balancing in the data center can be improved, power safety can be improved, and the refinement of power regulation can be promoted.

[0087] Embodiment 2

[0088] Please refer to Figure 3 , Figure 3 It is a structural diagram of an embodiment of a load balancing system for power safety of a data center provided by the present application, including: a construction module 01, an acquisition module 02, an analysis module 03 and a balancing module 04;

[0089] The construction module 01 is used to construct a power space tree based on the hierarchical relationship of power equipment in the data center;

[0090] Specifically, the construction module 01 includes: a decomposition unit, a first processing unit, a second processing unit and a construction unit; the decomposition unit is used to decompose the power equipment in the data center to obtain a number of equipment information; the first processing unit is used to create a number of equipment nodes corresponding to the levels based on the equipment information; the second processing unit is used to establish an association relationship between the equipment nodes at different levels; the construction unit is used to construct a power space tree based on the equipment nodes and the association relationship.

[0091] It is understandable that there are many kinds of power equipment in the data center, including mains power supply, transformer, UPS, cabinet, cabinet and air conditioner. There are hierarchical relationships among the various power equipment, and these hierarchical relationships will be recorded in the power planning drawings of the data center. The hierarchical relationship diagram of the power equipment hierarchical relationship of the data center is as follows: Figure 2 As shown, generally speaking, the AC power supply will be connected to multiple transformers, and each transformer will carry multiple UPS, the UPS will also carry multiple cabinets, the cabinets will carry multiple cabinets, and the cabinets will carry multiple servers.

[0092] It is understandable that the power equipment in the data center can be directly decomposed or the power equipment information and the hierarchical relationship of the power equipment in the data center can be mastered according to the power planning drawings of the data center. If the power equipment in the data center is directly decomposed, the specific steps are: first, directly decompose according to the power equipment situation in the data center, master the power equipment information and the hierarchical relationship of the power equipment in the data center; second, import several power equipment information into the power management system, form power equipment nodes corresponding to different levels in the power management system, and add corresponding equipment information to each equipment node, such as equipment basic information, equipment specifications and equipment power safety thresholds; third, according to the relationship between power equipment, establish an association relationship between the equipment nodes at different levels, i.e., upper and lower levels; finally, construct a power space tree based on the equipment nodes and the association relationship. If the power equipment information and the hierarchical relationship of the power equipment in the data center are mastered according to the power planning drawings of the data center, the specific steps are: obtain the power planning drawings of the data center, import the power planning drawings into the power management system, name each node and write out the specification attributes, and add a description field to each node. For example, for example, if the cabinet is the lowest level setting, you need to enter the corresponding cabinet information under the cabinet category, and at the same time, you need to associate which cabinet is its parent, that is, which cabinet is its parent node. At the same time, the cabinet is also associated with which UPS is its parent, and so on, until it is associated with the final AC power supply.

[0093] It should be noted that after the corresponding device information is entered into each device node in the power space tree, a unique ID will be automatically generated. Based on this unique ID, the corresponding device in the data center can be quickly queried.

[0094] It should be noted that the algorithm steps for constructing the power space tree include: (1) defining the device node class, where each device node contains attributes such as a unique ID, name, parent node, and child node; (2) creating device nodes at each level based on the power hierarchy in the data center; (3) associating each node in the order from the mains to the cabinet according to the power topology; (4) forming a tree structure based on the device nodes and the association relationship to facilitate subsequent power data management and analysis.

[0095] In this way, by first decomposing the power equipment in the data center, we can understand the hierarchical situation in the data center, facilitate the subsequent construction of a power space tree based on power equipment at different levels, and facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, thereby facilitating the judgment of whether there is a security risk.

[0096] The acquisition module 02 is used to respectively acquire a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively perform data processing on the first power value and the first load state to obtain a second power value and a second load state, wherein each node of the power space tree has a matching server;

[0097] The acquisition module 02 includes: a cleaning unit, a summarizing unit and a classification unit; specifically, the cleaning unit is used to perform data cleaning on the first power value and the first load state respectively to obtain a first cleaning result and a second cleaning result; the summarizing unit is used to summarize the first cleaning result and the second cleaning result by time respectively to obtain a first summary result and a second summary result; the classification unit is used to classify the first summary result and the second summary result respectively according to preset classification standards to obtain a second power value and a second load state.

[0098] It can be understood that after the power space tree is constructed, the power situation of each node in the power space tree can be grasped in real time; specifically, first, the dynamic ring system collects data from each node in the power space tree at a preset time frequency to obtain a first power value, wherein the first power value includes but is not limited to the current, voltage, power, electricity and other related power information of the equipment; secondly, the abnormal equipment power values ​​in the first power value are cleaned to obtain a first cleaning result; thirdly, the first cleaning result is summarized by time to obtain a first summary result, wherein the time in the time summary can be but not limited to hourly, daily and weekly, etc.; finally, the first summary result is classified according to a preset classification standard to obtain a second power value, and the second power value is stored in the corresponding storage table in the power management system, wherein the preset classification standard can be but not limited to equipment type and area, etc. Of course, the load status corresponding to each server in the power space tree can also be mastered in real time; specifically, first, the server software management system collects the load status of all servers in the data center in real time at a preset time frequency to obtain a first load status, wherein the server management system includes but is not limited to: open source systems such as Zabbix and Nagios. When Zabbix is ​​used to collect data, the Zabbix system is connected to the data center management system DCOS through the api interface to transmit the load status to the data center management system. The load status includes but is not limited to network bandwidth utilization, CPU utilization, memory utilization and process status; secondly, the missing part of the data in the first load status is removed or filled, and the abnormal value in the first load status is removed or corrected to obtain a second cleaning result; thirdly, the second cleaning result is summarized by time to obtain a second summary result, wherein the time in the time summary can be but is not limited to: hourly, daily and weekly; finally, the second summary result is classified according to the preset classification standard to obtain a second load state, and the second load state is stored in the corresponding storage table in the power management system according to different dimensions, wherein the preset classification standard can be but is not limited to device type, region, server ID and process status.

[0099] It should be noted that the first power value is proportional to the server load state. When the server's business load increases, its CPU usage rate increases, so the power consumption value also increases. The power consumption value of the parent node corresponds to the load of all servers in its child nodes. For example, the power dynamic value of the head cabinet corresponds to the load of all servers in all its child nodes (cabinets). Therefore, when the business of one or more servers under a node of the space tree increases, the power consumption value of the related nodes will also increase, thereby affecting the power safety of the space node.

[0100] It should be noted that the preset time frequency can be set as needed, which can be 3 minutes, 5 minutes or even other time frequencies, and this application does not impose any restrictions on this.

[0101] It should be noted that there may be abnormal equipment power values ​​in the operation and maintenance of the data center, so data cleaning is required. Specifically, if the power value of some unactivated equipment is 0 during the sampling period, then the power value of the equipment is not used as the basis for calculation and needs to be eliminated. Similarly, if the power value of some equipment is abnormal during a certain test phase, cleaning processing is required to reduce the noise of the data.

[0102] It should be noted that summarizing the first cleaning results by time or category can effectively analyze power usage, identify peak load time periods, and optimize power resource allocation. Classifying and storing data can ensure orderly data management and efficient query.

[0103] In this way, by performing data cleaning, data aggregation and data classification on the first power value and the first load status respectively, the quality and availability of the data can be ensured, which facilitates subsequent data extraction, analysis and decision-making.

[0104] The analysis module 03 is used to analyze the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center, and if there is a security risk, automatically generate a corresponding scheduling task;

[0105] The analysis module 03 includes: a first analysis unit and a second analysis unit; specifically, the first analysis unit is used to perform statistical analysis on the second power value to determine the risk time period when each node has a security risk; the second analysis unit is used to analyze the second load state corresponding to each node during the risk time period to determine the servers with security risks in the data center.

[0106] It can be understood that, first, by calculating the average value, variance, standard deviation, maximum value and minimum value of the second power value in a certain time period, and based on the calculation results, comprehensively determine the risk time period in which there is a security risk in the node, that is, which servers may have some overload conditions in which time periods. Specifically, by calculating the average value of the second power value corresponding to a single node under the power space tree in a period of time, it is determined whether the node is stable in a period of time, thereby reducing the interference of emergencies; calculating the variance and standard deviation of the second power value in a period of time to determine the fluctuation relative to the average value; by calculating the maximum value and minimum value of a single node in a period of time, the peak and valley of the fluctuation are determined. Exemplarily, the ups node in the power space tree is loaded with hundreds of cabinets. During the time periods of 12 noon and 20-21 pm, there are more people playing games, so the server business volume loaded by the ups increases, and the power consumption will also increase. Therefore, it is necessary to detect power safety at all times. Through the above method of calculating power consumption, it can be comprehensively determined which time periods will be peak periods of business and which time periods will be trough periods of business. In the future, when there are overloaded devices during peak hours, the devices can be load balanced. For example, when a UPS is overloaded, the service must be regulated through the server load balancing, and the service must be adjusted to other UPS load servers to keep the UPS load value within the safety threshold.

[0107] It should be noted that this application does not limit the order of calculation methods such as mean, variance, standard deviation, maximum and minimum values, and the calculation method is not the focus of this application, so it will not be expanded here.

[0108] When a node is abnormal, its corresponding server will be abnormal, and the server corresponding to its parent node will also be abnormal, so multiple abnormal information will be received, and due to the conduction effect of the power space tree, a piece of error will appear, so it is necessary to analyze the load status of the server, that is, it is necessary to decompose according to the power value of each node to find the node with real problems. Specifically, after the risk time period is determined, the second load state corresponding to the risk time period can be clustered and analyzed to predict the risk of different nodes in the power space tree. At the same time, the regression algorithm is combined to check the load of each node in different time periods. When the power value or load state corresponding to each node exceeds the pre-set safety threshold, the server with security risk in the data center is determined. Of course, it is also possible to perform machine learning on the operation status and analysis rules of all nodes in the power space tree in advance to train the corresponding prediction model to predict the future risk points of different nodes in the space tree, and visualize the corresponding risk analysis and real-time alarm related content, where the visualization method includes but is not limited to charts and reports.

[0109] It should be noted that cluster analysis and regression algorithm are not the focus of this application, so they will not be expanded here.

[0110] It should be noted that the device itself has a safe power value, which is generally not allowed to be exceeded, otherwise it is easy to cause safety risks. Therefore, the safe power value can be multiplied by a certain coefficient as a safety threshold. When the device power value exceeds the safety threshold, the device location and the location of the space tree node to which it belongs will be sent to the operation and maintenance personnel for an alarm.

[0111] In this way, by analyzing the second power value, the risk time period when there is a security risk can be determined, and the load status during the risk time period can be analyzed, which can effectively reduce the difficulty of data analysis and more quickly determine whether there is a security risk in the server in the data center.

[0112] Finally, after the server with security risks is determined, a scheduling task can be automatically generated and sent to the load balancing system, which will help to subsequently load balance the corresponding servers; the scheduling task will include but is not limited to the server name, cup load rate, memory load rate, the server's corresponding security threshold, the server's location in the space tree, and other related information.

[0113] The balancing module 04 is used to perform load balancing on servers with security risks in the data center based on the scheduling tasks.

[0114] It is understandable that after receiving the scheduling task, the server can be directional adjusted through the load balancing system. Specifically, the power value can be dispersed to other servers with smaller loads to ensure that the current load status of all servers meets the requirements of the safety threshold, thereby achieving load balancing. For example, when the power value of a node in the power space tree only reaches 30% or less of the same service servers, these servers are the target servers for adjusting the service. By distributing 20% ​​to 30% of the excess server services to these servers, load balancing can be achieved.

[0115] The embodiment of the present application constructs a power space tree based on the hierarchical relationship of power equipment in a data center, so as to facilitate the subsequent acquisition of the power value of each node and the load status of the corresponding server from the power space tree, and then facilitate the judgment whether there is a security risk; by respectively obtaining the first power value corresponding to each node in the power space tree and the first load status corresponding to each server, it is convenient to quickly and accurately determine the servers with security risks in the data center based on the first power value and the first load status corresponding to each server; by respectively analyzing the second power value and the second load status, it is possible to determine the servers with security risks in the data center, so as to facilitate the subsequent load balancing thereof; by scheduling tasks to load balance servers with security risks, the accuracy and efficiency of power safety load balancing in the data center can be improved, power safety can be improved, and the refinement of power regulation can be promoted.

[0116] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A load balancing method for power safety in a data center, characterized in that: include: Construct a power space tree based on the hierarchical relationship of power equipment in the data center; Respectively obtaining a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively performing data processing on the first power value and the first load state to obtain a second power value and a second load state, wherein each node of the power space tree has a matching server; Analyze the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center, and automatically generate a corresponding scheduling task if there is a security risk; Based on the scheduling task, load balancing is performed on servers with security risks in the data center.

2. The load balancing method for data center power safety according to claim 1, characterized in that: The power space tree is constructed based on the hierarchical relationship of power equipment in the data center, specifically: Decompose the power equipment in the data center to obtain some equipment information; Based on the device information, create device nodes corresponding to several levels; Establishing association relationships between the device nodes at different levels; Based on the device nodes and the association relationships, a power space tree is constructed.

3. The load balancing method for data center power safety according to claim 1, characterized in that: The data processing of the first power value and the first load state is respectively performed to obtain a second power value and a second load state, specifically: Performing data cleaning on the first power value and the first load state respectively to obtain a first cleaning result and a second cleaning result; Summarizing the first cleaning result and the second cleaning result by time respectively to obtain a first summary result and a second summary result; The first summary result and the second summary result are classified according to preset classification standards to obtain a second power value and a second load state.

4. The load balancing method for data center power safety according to claim 1, characterized in that: The analyzing the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center is specifically: Performing statistical analysis on the second power value to determine a risk time period in which each node has a security risk; The second load state corresponding to each node in the risk time period is analyzed to determine that a corresponding server in the data center has a security risk.

5. The load balancing method for data center power safety according to any one of claims 1 to 4, characterized in that: The determining whether there is a security risk in the server in the data center is specifically: Pre-set the security threshold corresponding to each node in the data center; If the second power value or the second load state corresponding to each node exceeds a safety threshold, it is determined that a corresponding server in the data center has a safety risk.

6. A load balancing system for power safety in a data center, characterized in that: include: Building module, acquisition module, analysis module and equalization module; The construction module is used to construct a power space tree based on the hierarchical relationship of power equipment in the data center; The acquisition module is used to respectively acquire a first power value corresponding to each node in the power space tree and a first load state corresponding to each server, and respectively perform data processing on the first power value and the first load state to obtain a second power value and a second load state, wherein each node of the power space tree has a matching server; The analysis module is used to analyze the second power value and the second load state respectively to determine whether there is a security risk for the server in the data center, and if there is a security risk, automatically generate a corresponding scheduling task; The balancing module is used to perform load balancing on servers with security risks in the data center based on the scheduling tasks.

7. The load balancing system for data center power safety according to claim 6, characterized in that: The construction module includes: a decomposition unit, a first processing unit, a second processing unit and a construction unit; The decomposition unit is used to decompose the power equipment in the data center to obtain a number of equipment information; The first processing unit is used to create device nodes corresponding to a plurality of levels based on the plurality of device information; The second processing unit is used to establish association relationships between the device nodes at different levels; The construction unit is used to construct a power space tree based on the device nodes and the association relationships.

8. The load balancing system for data center power safety according to claim 6, characterized in that: The acquisition module includes: a cleaning unit, a summarizing unit and a classification unit; The cleaning unit is used to clean the first power value and the first load state respectively to obtain a first cleaning result and a second cleaning result; The summarizing unit is used to summarize the first cleaning result and the second cleaning result by time to obtain a first summary result and a second summary result; The classification unit is used to classify the first summary result and the second summary result according to preset classification standards to obtain a second power value and a second load state.

9. The load balancing system for data center power safety according to claim 7, characterized in that: The analysis module comprises: a first analysis unit and a second analysis unit; The first analysis unit is used to perform statistical analysis on the second power value to determine a risk time period in which each node has a security risk; The second analysis unit is used to analyze the second load status corresponding to each node in the risk time period to determine whether the corresponding server in the data center has a security risk.

10. The load balancing system for data center power safety according to any one of claims 6 to 9, characterized in that: The analysis module also includes: a setting unit and a judgment unit: The setting unit is used to pre-set the security threshold corresponding to each node in the data center; The judgment unit is used to determine that a corresponding server in the data center has a security risk if the second power value or the second load state corresponding to each node exceeds a safety threshold.

Images