Method for acquiring racking position and computing equipment

Abstract

The embodiment of the invention provides a racking position obtaining method and computing equipment, and the method specifically comprises the steps: obtaining a target feature, determining a reference target feature from the target feature, screening out candidate racking positions from idle positions, and determining a target racking position meeting a load feature from the candidate racking positions through employing the target feature. Therefore, the to-be-shelved equipment is shelved on the basis of the shelving position, so that the phenomena that some machine rooms or cabinets and the like in a data center are often wasted in idle resources and even are completely idle, and other machine rooms or cabinets face performance bottlenecks and potential safety hazards due to excessive load bearing can be improved; the overall performance of the data center is improved.

Description

Technical Field

[0001] The present application relates to the field of server technology, and in particular to a method for obtaining a shelf location and a computing device. Background Art

[0002] As an indispensable cornerstone of modern information technology, data centers carry a variety of information technology equipment such as servers, network equipment and storage devices, and provide strong support for data processing, storage, high-speed transmission, seamless exchange and intelligent management. With the widespread use of cloud computing and big data technologies, data centers are developing rapidly, and the scale of data centers is also getting larger and larger. Super-large-scale data centers with tens of thousands to hundreds of thousands of information technology equipment are already common.

[0003] The expansion of the scale of data centers has led to increasingly severe management of data centers, especially the shelving and de-shelving of a large number of information technology equipment, which has brought great adjustments to the space resource management of data centers. At present, before information technology equipment is put on the shelves, a shelf plan including the shelf location will be obtained first, and then the user will put the information technology equipment on the shelves based on the shelf plan. However, at present, some computer rooms or cabinets in data centers often have idle resources wasted or even completely idle, while other computer rooms or cabinets face performance bottlenecks and safety risks due to excessive loads, affecting the overall performance of the data center. Summary of the invention

[0004] The embodiments of the present application provide a method for obtaining shelf locations and a computing device, which are used to improve the phenomenon in a data center where some computer rooms or cabinets often have idle resources wasted or even completely idle, while other computer rooms or cabinets face performance bottlenecks and safety hazards due to carrying too much load, thereby improving the overall performance of the data center.

[0005] In order to achieve the above objectives, the embodiments of the present application provide the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a method for obtaining a shelf location, the method comprising:

[0007] Acquire target characteristics, wherein the target characteristics include idle positions and load characteristics; determine reference target characteristics from the target characteristics; based on the reference target characteristics, screen out at least one candidate shelving position from the idle positions, wherein the at least one candidate shelving position meets the requirements of the reference target characteristics; based on the target characteristics, determine a target shelving position from a plurality of the candidate shelving positions, wherein the target shelving position is the idle position that meets the requirements of the load characteristics.

[0008] The embodiment of the present application determines the reference target feature from the target feature, selects the candidate racking position from the free position, and uses the target feature to determine the target racking position that meets the load feature from the candidate racking position. Therefore, the embodiment of the present application racks the equipment to be racked based on the racking position, which can improve the phenomenon that some computer rooms or cabinets in the data center often waste idle resources or are completely idle, while other computer rooms or cabinets face performance bottlenecks and safety hazards due to carrying too much load, thereby improving the overall performance of the data center.

[0009] In a specific implementation, the recommendation priority of multiple candidate shelf locations is determined, and the higher the recommendation priority of the candidate shelf location, the more requirements of the target features that the candidate shelf location meets; based on the recommendation priority, the target shelf location is determined. Based on the recommendation priority method, the most suitable candidate shelf location is given priority, which helps to optimize the shelf location resource allocation and improve the overall efficiency of the data center.

[0010] In a specific implementation, the score of each target feature of the candidate shelf position is determined based on the preset score of the feature data of the target feature, and the target feature includes multiple feature data. The higher the preset score of each feature data, the higher the recommendation priority of the target feature; based on the score of each target feature of the candidate shelf position, the recommendation priority of the candidate shelf position is determined. By setting the preset score of the feature data of the target feature, the sum of each candidate shelf position is obtained, and the sum is used as the recommendation priority of the candidate shelf position. This method of quantifying the recommendation priority by score can reduce subjective judgment, improve the objectivity and consistency of determination, and all candidate shelf positions are evaluated based on the same features and preset scores, so that the comparison between different positions is more open, the determination results are more accurate, and more in line with customer expectations.

[0011] In a specific implementation, the target characteristics include free space, power capacity, cooling capacity, network bandwidth, maintenance and access, cost-effectiveness, compliance, scalability, redundancy, security, network latency, or disaster recovery, and the reference target characteristics include: available space, power capacity, cooling capacity, or network bandwidth.

[0012] In a specific implementation, scores of multiple candidate shelf positions are obtained based on target features and a scoring card model. The scoring card model is used to determine the scores of candidate shelf positions based on preset scores of feature data of the target features; based on multiple scores, the target shelf position is determined from multiple candidate shelf positions. The scoring card model is a model created based on existing historical shelf data, so the target features and target feature weights used in the scoring card model can reflect the statistical management between each feature and the best shelf position. This approach reduces the impact of subjective judgment and improves the accuracy of decision-making. In addition, the scoring card model can capture hidden patterns and regularities in the data, thereby accurately predicting new candidate shelf positions. This predictive capability helps to discover and avoid potential risks and problems in advance.

[0013] In a specific implementation, the scorecard model includes a prediction algorithm for predicting a candidate listing position, and the prediction algorithm is:

[0014] Score=∑ i (P i ×Q i )

[0015] Among them, Score is the score of the candidate listing position, i represents the scoring items included in the scoring card model, i is the same as the number of target features, and each scoring item represents a target feature. i represents the preset score of the feature data of scoring item i, Q i It represents the weight value of rating item i. The weight value of rating item i is positively correlated with the importance of rating item i in determining the listing position.

[0016] In a specific implementation, at least one candidate shelf position is screened out from an idle position based on a decision tree model and reference target features; wherein the decision tree model is a decision tree model trained based on historical shelf data, and is used to determine reference target features from target features and screen out at least one candidate shelf position from an idle position, and the historical shelf data includes target features and reference shelf positions of historical shelf devices. The decision tree model can automatically adjust its structure and parameters according to the different historical shelf data, so as to adapt to different data sets and business scenarios. This adaptability and flexibility enables the decision tree model to maintain high prediction accuracy and robustness in a complex and changing environment. And as the amount of data increases and the data is continuously updated, the decision tree model can continue to learn and optimize its decision rules. This means that the model can gradually discover the laws and patterns hidden in the data, so as to more accurately predict future shelf positions or results.

[0017] Among them, the decision tree model includes a connected root node, multiple child nodes and leaf nodes: the root node of the decision tree model is a node for whether the power capacity is sufficient, the multiple child nodes include a sub-node for whether the cooling capacity is sufficient, a sub-node for whether the available space is sufficient and a sub-node for whether the network bandwidth is sufficient, and the leaf nodes include leaf nodes that do not meet the conditions and leaf nodes that meet all the conditions.

[0018] In another implementation, if the power capacity is sufficient, jump to the sub-node of whether the cooling capacity is sufficient; if the power capacity is insufficient, jump to the sub-node of whether the available space is sufficient; if the cooling capacity is sufficient, jump to the sub-node of whether the network bandwidth is sufficient; if the cooling capacity is insufficient, jump to the leaf node that does not meet the conditions; if the available space is sufficient, jump to the sub-node of whether the network bandwidth is sufficient; if the available space is insufficient, jump to the leaf node that does not meet the conditions; if the network bandwidth is sufficient, jump to the leaf node that meets all conditions; if the network bandwidth is insufficient, jump to the leaf node that does not meet the conditions; the shelf position of the leaf node that meets all conditions is the candidate shelf position.

[0019] In a second aspect, an embodiment of the present application provides a computing device, comprising: a memory and a processor; the memory is used to store programs; the processor is used to execute the programs stored in the memory, and when the program stored in the memory is executed, the processor is used to execute any method as in the first aspect.

[0020] In a third aspect, an embodiment of the present application provides a device for obtaining a shelf position, including:

[0021] An acquisition unit, used for acquiring target characteristics, the target characteristics including idle position and load characteristics;

[0022] A feature determination unit, used for determining a reference target feature from the target feature;

[0023] A screening unit, configured to screen out at least one candidate shelf position from the idle positions based on the reference target feature, wherein at least one candidate shelf position meets the requirement of the reference target feature;

[0024] The position determination unit is used to determine a target racking position from a plurality of candidate racking positions based on the target characteristics, wherein the target racking position is an idle position that meets the requirements of the load characteristics.

[0025] In a fourth aspect, the present application provides a computer storage medium for storing a computer program, which, when executed, is used to implement any one of the methods of the first aspect of the embodiments of the present application.

[0026] In a fifth aspect, the present application provides a computer program product comprising instructions, which, when executed on at least one computing device, enables the at least one computing device to implement any one of the methods of the embodiments of the present application.

[0027] The computing device, computer-readable storage medium, or computer program product of any of the above-mentioned shelf location acquisition methods are used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A schematic diagram of a data center provided in an embodiment of the present application;

[0029] Figure 2 An example diagram of the architecture of a method for obtaining a shelf location provided in an embodiment of the present application;

[0030] Figure 3 An interactive diagram of a method for obtaining a shelf location provided in an embodiment of the present application;

[0031] Figure 4 A flow chart of a method for obtaining a decision tree model provided in an embodiment of the present application;

[0032] Figure 5 A schematic diagram of a decision tree model provided in an embodiment of the present application;

[0033] Figure 6 A flowchart of the construction method of the scorecard model provided for the implementation of this application;

[0034] Figure 7 A flow chart of another method for obtaining a shelf location provided in an embodiment of the present application;

[0035] Figure 8 A schematic diagram of the structure of a shelf position acquisition device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] It should be noted that the embodiments described in this application are only part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0037] First, the application scenarios of the embodiments of the present application are introduced.

[0038] The method for obtaining a shelf location provided in an embodiment of the present application is applied in a data center scenario. A data center refers to a facility used to house computer systems and related components, such as telecommunications and storage systems. Generally, it includes redundant and backup power supplies, redundant data communication connections, environmental controls (such as air conditioning, fire extinguishers), and various safety devices.

[0039] Attached Figure 1A data center 100 is a schematic diagram of a data center provided in an embodiment of the present application. The data center 100 generally includes several computer rooms, such as Figure 1 There are m computer rooms on display, namely computer room 1 to computer room m. Each computer room includes multiple cabinets, such as the attached Figure 1 It is shown that each computer room includes n cabinets, namely cabinet 1 to cabinet n. Each cabinet includes multiple information technology (IT) devices. IT devices usually refer to servers, network devices, storage devices and other devices. For example, Figure 1 N servers are shown, namely server 1 to server N. m≥1, and m is an integer; n≥1, and n is an integer, and N≥1, and N is an integer. In some possible scenarios, the number of cabinets in the data center is more than 1,000, and the number of IT devices is more than 5,000.

[0040] For data centers, IT equipment often needs to be put on shelves, that is, to add IT equipment to the space resources of the computer room or cabinet of the data center. For the convenience of description, the IT equipment is put on shelves by taking the example of putting servers on shelves. The IT equipment is put on shelves in the data center by taking the example of putting servers on shelves in cabinets.

[0041] The embodiment of the present application provides a method for obtaining a shelf position, which is used to obtain the shelf position of a device to be shelf mounted. The device to be shelf mounted is shelf mounted based on the shelf position, which can effectively improve the problem of unbalanced load of space resources in the data center and improve the overall performance of the data center.

[0042] The following is a detailed description of the method for obtaining the shelf location provided by the embodiment of the present application in conjunction with the accompanying drawings. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0043] Figure 2 The present invention provides a method for obtaining a shelf location, which can be applied to the following examples: Figure 2 As shown in the system architecture diagram. Figure 2 As shown, the racking system includes a terminal 210 , a control platform 220 and a data center 230 .

[0044] Among them, the terminal 210 is composed of a terminal device and application software deployed on the terminal device. Among them, the terminal device can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. Exemplary embodiments of the terminal devices involved in this solution include, but are not limited to, electronic devices equipped with iOS, Android, Windows, Harmony OS, or other operating systems. The embodiment of this application does not specifically limit the type of electronic device.

[0045] The control platform 220 can run on a server, for example, a general-purpose server, a GPU server, an AI server, a heterogeneous server, etc. Alternatively, the control platform 220 can be composed of a device cluster and application software deployed on the device cluster. The device cluster can be implemented with an independent computing device or a device cluster composed of multiple computing devices. In some possible implementations, the computing device in the device cluster 220 can be a terminal or a computer, or a server. In one example, the server involved in this solution can be used to provide cloud services, which can be a server or a super terminal that can establish a communication connection with other devices and can provide computing functions and / or storage functions for other devices. Among them, the server involved in this solution can be a hardware server, or it can be implanted in a virtualized environment. For example, the server involved in this solution can be a virtual machine executed on a hardware server including one or more other virtual machines.

[0046] The terminal 210 communicates with the control platform 220 through the network, and the control platform 220 interacts with the data center 230 through the network. The network can be a wired network or a wireless network. For example, the wired network can be a cable network, an optical fiber network, a digital data network (Digital Data Network, DDN), etc., and the wireless network can be a telecommunication network, an internal network, the Internet, a local area network (Local Area Network, LAN), a wide area network (Wide Area Network, WAN), etc. or any combination thereof. It is understandable that the network can use any known network communication protocol to achieve communication between different client layers and gateways, and the above-mentioned network communication protocol can be various wired or wireless communication protocols.

[0047] In the embodiment of the present application, the terminal 210 obtains the target feature and sends the target feature to the control platform 220. The control platform 220 obtains the shelf position of the device to be shelf-mounted from the idle positions of the data center 230 based on the monitored idle positions of the data center 230. Thus, the data center 230 shelves the device to be shelf-mounted based on the shelf position, which can improve the phenomenon that some computer rooms or cabinets in the data center often waste idle resources or are completely idle, while other computer rooms or cabinets face performance bottlenecks and safety hazards due to carrying too many electronic devices, thereby improving the overall performance of the data center.

[0048] It should be noted that the system architecture and application scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Ordinary technicians in this field can know that with the evolution of the system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0049] The following is an exemplary introduction to the method for obtaining the shelf location provided in the embodiment of the present application in conjunction with the accompanying drawings.

[0050] Attached Figure 3 An interactive diagram of a method for obtaining a shelf location provided in an embodiment of the present application. The method is applied in the attached Figure 2 On the rack system shown, the method includes the following:

[0051] S310, the terminal inputs target features.

[0052] In the embodiment of the present application, the target characteristics include the idle position and load characteristics of the current data center. The load characteristics refer to the characteristics related to the load (such as business load or workload, etc.) at the specific location of the current data center, such as power capacity, cooling capacity, network bandwidth, network delay and redundancy.

[0053] The free space is also called available space, which refers to the remaining space or available space in the rack or computer room in the data center. The available space of the data center is used to measure whether the rack position meets the shelf requirements for the equipment to be put on the rack. If the available space of the data center is greater than or equal to the space required for the equipment to be put on the rack, the available space of the data center meets the shelf requirements for the equipment to be put on the rack. For example, if the height of the server to be put on the rack is 2U, and the rack space of the data center is 1U, 2U>1U, the rack space of the data center cannot meet the height requirement of the server to be put on the rack, and the server to be put on the rack cannot be put on the rack at this location. If the rack space of the data center is 3U, 3U≥2U, the rack space of the data center should meet the height requirement of the server to be put on the rack.

[0054] Power capacity refers to the power provided by the power supply at the current moment in the specific location of the data center. When the equipment to be put on the rack is running, the power supply needs to provide corresponding power to support it. If the power provided by the power supply does not meet the power required by the equipment to be put on the rack, the equipment to be put on the rack may not be able to run, or some functions may be shut down to meet the power requirements provided by the power supply, affecting the function of the equipment to be put on the rack.

[0055] Cooling capacity refers to the ability of a specific location in a data center to handle heat loads. When the equipment to be put on the rack is running, heat loads will be generated. Excessive heat loads will affect the performance of the equipment to be put on the rack, such as affecting the running speed of the equipment to be put on the rack. In addition, excessive heat loads will affect the life of the equipment to be put on the rack. Therefore, in order to increase the life of the equipment to be put on the rack and increase the performance of the equipment to be put on the rack, non-business characteristics may also include cooling capacity, that is, the ability of the data center to handle the heat load of the equipment to be put on the rack at that location.

[0056] Network bandwidth refers to the network bandwidth allocated by the data center to a specific location at the current moment. When the device to be put on the shelf performs a certain function, it sometimes needs the help of the network to operate normally. Therefore, non-business characteristics can also include network bandwidth, that is, the network bandwidth allocated by the data center to a specific location. By ensuring that the network bandwidth allocated by the data center to a specific location meets the communication needs of the device to be put on the shelf, the performance of the device to be put on the shelf is guaranteed.

[0057] Network latency refers to the time required for data to be transmitted in the network at the specific location of the data center at the current moment, that is, the time interval from the start of data transmission at the sender to the receipt of the data at the receiver. Network latency is also an important indicator for evaluating network performance, which directly affects the user experience when using network services.

[0058] Redundancy refers to the artificially added duplication in the data center, which is used to ensure the high availability, data security and business continuity of the rack equipment. By introducing redundancy at different levels, even if a component or system fails, the entire data center can continue to operate normally, thereby avoiding or reducing the impact on the business. In the embodiment of the present application, redundancy includes power redundancy, storage redundancy, network redundancy and cooling system redundancy. Therefore, when determining the rack location of the equipment to be racked, the control platform 220 intelligently plans the rack location according to the redundancy requirements of the data center, such as single redundancy and dual redundancy, to ensure the optimal configuration of resources and the effective implementation of the redundancy strategy.

[0059] In a specific implementation, the target characteristics also include business characteristics associated with the business, such as maintenance and access, cost-effectiveness, compliance, scalability, security, and disaster recovery.

[0060] Maintenance and access refer to whether the location of the data center is convenient for routine maintenance and troubleshooting. Cost-effectiveness refers to the rental and operation and maintenance costs of the shelf location. Compliance refers to whether the location of the data center meets industry standards and regulatory requirements. Scalability refers to whether the location of the data center has the possibility of expanding the number of servers, etc. Disaster recovery refers to the availability of backups and disaster recovery plans for the location of the data center. Cost-effectiveness refers to the rent and costs of the specific location of the current data center. Security refers to the state in which the data at the specific location of the data center is protected from threats such as unauthorized access, leakage, tampering or destruction, including high security level, medium security level and low security level. In the embodiment of the present application, the high security level can be data protected from threats such as unauthorized access, leakage, tampering and destruction, the medium security level can be data exempt from some unauthorized access, leakage, tampering and destruction, and the low security level can be data protected from threats such as unauthorized access, leakage, tampering and destruction. In addition, the embodiment of the present application can also be divided into high security level, low security level and medium security level in other ways, which are not specifically limited in the embodiment of the present application.

[0061] In addition, the target feature may also include other features, which are not specifically limited in the embodiments of the present application.

[0062] S320. The terminal sends the target features to the control platform.

[0063] That is, after the terminal 210 obtains the target feature, it sends the target feature to the control platform 220. The control platform 220 obtains the target feature and determines the shelf position of the device to be shelf-mounted based on the target feature.

[0064] S330: The control platform inputs the target features and the spatial resources of the data center into a decision tree model to obtain at least one candidate shelf location.

[0065] The decision tree model provided in the embodiment of the present application is a supervised learning process, which is a tree structure for classifying samples. The classification process of the decision tree model is: recursively select reference target features such as optimal partitioning features from a given data set based on feature selection criteria, and segment the data set based on the reference target features so that each data subset includes an optimal classification result.

[0066] In an embodiment of the present application, the decision tree model can segment the data set based on idle and load characteristics so that each data subset includes an optimal classification result. Thus, the decision tree model can filter out at least one candidate shelf position from the spatial location of the data center based on the input target characteristics, the spatial resources of the current data center, and the decision tree model. Among them, the candidate shelf position is a shelf position in the idle position that meets the reference target characteristics in the target characteristics.

[0067] The decision tree model provided by the present application is a decision tree model trained with historical shelf data, and the historical shelf data includes the shelf locations of historical shelf devices and corresponding target features. Figure 4 A flowchart of a method for acquiring a decision tree model provided in an embodiment of the present application. The execution subject of the method for acquiring a decision tree model may be the control platform 220 or other devices, which is not specifically limited in the embodiment of the present application. The following is an example of the method for acquiring a decision tree model executed by the control platform 220. The method includes:

[0068] S330.1: Obtain historical listing data.

[0069] The control platform 220 obtains a plurality of historical shelf data. The historical data refers to the shelf locations of historical shelf devices and the corresponding historical target characteristics. For example, the target characteristics of historical shelf devices include but are not limited to free locations, power capacity, cooling capacity, network bandwidth, maintenance and access, cost-effectiveness, compliance, scalability, redundancy, security, network latency or disaster recovery, etc.

[0070] Furthermore, considering that the acquired historical data may have problems such as missing data, data anomalies or errors that may affect the accuracy of the model, the control platform 220 may continue to execute S330.2 to pre-process the historical shelf data.

[0071] S330.2: Preprocess historical shelf data.

[0072] The control platform 220 pre-processes the historical shelf data to improve the data quality of the historical shelf data.

[0073] In one example, the control platform 220 can clean the historical data on the shelves, remove missing values, outliers or erroneous data in the historical data on the shelves, thereby avoiding the impact of missing values, outliers or erroneous data on the decision tree model and improving training efficiency and training accuracy.

[0074] In another example, the control platform 22 may also normalize the historical shelf data after clearing, convert data of different dimensions into the same dimension, enhance the generalization ability of the model, and improve the training effect.

[0075] Furthermore, in order to improve the training effect of the decision tree model, the control platform 220 can also divide the preprocessed historical shelf data into data sets, using part of the historical shelf data as a training set and another part of the data as a test set. For example, the preprocessed historical shelf data consists of N groups, of which N1 group of historical shelf data is used as a training set, and the remaining N-N1 groups of historical shelf data are used as a test set. Among them, N>1 is an integer, and N>N1>1 is an integer.

[0076] The training set is used to build a decision tree model, and the validation set is used to verify the accuracy of the decision tree model.

[0077] S330.3: Build a decision tree model based on the historical shelf data of the training set.

[0078] The control platform 220 uses the historical shelf data of the training set to train the decision tree training model and construct a decision tree model, so that the constructed decision tree model can determine multiple candidate shelf positions for the equipment to be listed based on the optimal division features in the target features.

[0079] Among them, the decision tree training model can be a classification and regression tree (CART) model or a random forest decision tree model, which is not specifically limited in the embodiments of the present application.

[0080] As an embodiment, if the target features include free space, power capacity, cooling capacity, network bandwidth, maintenance and access, cost-effectiveness, compliance, scalability, redundancy strategy, security level, network delay and disaster recovery, the reference target features of the trained decision tree model include available space, power capacity, cooling capacity and network bandwidth. Among them, the nodes of available space, power capacity, cooling capacity and network bandwidth can be adjusted as needed. In the embodiment of the present application, these four core features are selected as reference target features, which will improve the efficiency of the decision-making process compared to using all target features. In addition, these four features are the most basic and critical resources in data center operations, which directly determine whether the equipment to be put on the shelves can operate normally, as well as the overall performance and stability of the data center. Therefore, the candidate shelf locations screened out using these four core features can ensure that they are reliable at the physical infrastructure level, thereby reducing subsequent problems caused by insufficient resources or improper configuration.

[0081] The decision tree model includes a connected root node, multiple child nodes and leaf nodes. In one example, the root node of the decision tree model is a node for whether the power capacity is sufficient, the multiple child nodes include a node for whether the cooling capacity is sufficient, a node for whether the available space is sufficient and a node for whether the network bandwidth is sufficient, and the leaf nodes include a leaf node that does not meet the conditions and a leaf node that meets all the conditions.

[0082] Example description: Attached Figure 5A schematic diagram of a decision tree model provided for an embodiment of the present application. The decision tree model is a decision tree model obtained after training based on historical shelf data including a training set. The root node of the decision tree model is the node of whether the power capacity is sufficient. If sufficient, it jumps to the sub-node of whether the cooling capacity is sufficient. If insufficient, it jumps to the sub-node of whether the available space is sufficient. If the cooling capacity is sufficient, jump to the sub-node of whether the network bandwidth is sufficient. If the cooling capacity is insufficient, jump to the leaf node that does not meet the conditions. If the available space is sufficient, jump to the sub-node of whether the network bandwidth is sufficient. If the available space is insufficient, jump to the leaf node that does not meet the conditions. If the network bandwidth is sufficient, jump to the leaf node that meets all conditions. If the network bandwidth is insufficient, jump to the leaf node that does not meet the conditions. Among them, the leaf node that does not meet the conditions will no longer perform subsequent operations. The leaf node that meets all conditions executes subsequent steps.

[0083] It is understandable that if the target feature also includes other features such as business features or redundancy, based on the attached Figure 5 The decision tree model shown considers power capacity, cooling capacity, available space, and network bandwidth among the target features when processing the target features.

[0084] The decision tree model varies with the historical shelf data, so the decision tree model can automatically adjust its structure and parameters according to the historical shelf data to adapt to different data sets and business scenarios. This adaptability and flexibility enables the decision tree model to maintain high prediction accuracy and robustness in a complex and changing environment. And as the amount of data increases and the data is constantly updated, the decision tree model can continue to learn and optimize its decision rules. This means that the model can gradually discover the rules and patterns hidden in the data, so as to more accurately predict future shelf positions or results.

[0085] S330.4: Evaluate the performance of the decision tree model based on the validation set, and adjust the model parameters of the decision tree model.

[0086] The control platform 220 can determine the performance of the trained decision tree model based on the validation set and the evaluation parameters, wherein the target evaluation parameters are used to evaluate the performance of the decision tree model, such as accuracy and / or recall rate.

[0087] The following is a schematic explanation using the accuracy as the target evaluation parameter as an example.

[0088] For example, there are 100 historical shelf data in the validation set. After the historical shelf data in the validation set is input into the constructed decision tree model, 95 of the historical shelf data have the same shelf positions as the actual shelf positions, that is, the accuracy of the constructed decision tree model is 95%. If the accuracy threshold preset by the control platform 220 is 90%, 95%>90%, it is determined that the trained decision tree model meets the requirements.

[0089] Furthermore, if the accuracy is less than the accuracy threshold, it is necessary to readjust the model parameters of the decision tree model, such as the maximum depth of the decision tree or the minimum number of samples required for a leaf node, and verify again.

[0090] S330.5: Obtain the validated decision tree model.

[0091] In an embodiment of the present application, the control platform 220 inputs the target features and the spatial resources of the data center into a verified decision tree model. The decision tree model makes decisions on the spatial resources of the data center based on the optimal partitioning features, and selects at least one candidate shelf location that meets the requirements from the spatial resources of the data center.

[0092] The embodiment of the present application adopts a decision tree model to screen out multiple candidate shelf locations from the spatial resources of the data center. Compared with manual screening of the shelf locations of the equipment to be listed, this model is highly efficient and will not be affected by human subjectivity or the risk of erroneous recommendations, thereby improving the acquisition efficiency and further improving the success rate of the listing of the equipment to be listed.

[0093] S340: Filter out a target shelf location from at least one candidate shelf location based on the target feature.

[0094] The control platform 220 may select a target racking location from at least one candidate racking location based on the target feature, wherein the target racking location is also called the best racking location, which refers to a racking location that meets some or all of the load feature requirements.

[0095] The control platform 220 can first determine the recommendation priority of multiple candidate shelf locations, and use the candidate shelf location with the highest recommendation priority, the candidate shelf location with the second highest recommendation priority, or the candidate shelf location with other recommendation priorities as the target shelf location. In an embodiment of the present application, the recommendation priority is used to describe whether the recommended data center location is recommended first. The higher the recommendation priority, the more the data center location is recommended first. Based on the recommendation priority method, the most suitable candidate shelf location is given priority, which helps to optimize the allocation of shelf location resources and improve the overall efficiency of the data center.

[0096] In addition, the control platform 220 can assign scores to the preset scores of the feature data of the target feature, that is, to different feature data. The control platform 220 determines the score of each target feature of the candidate shelf position based on the preset scores of the feature data of the target feature. The preset score of each feature data is related to the recommended priority of determining the shelf position of the device to be shelf-mounted. For example, the higher the recommended priority, the higher the preset score, or the lower the recommended priority, the higher the preset score. The following is an example of the higher the recommended priority, the higher the preset score.

[0097] The control platform 220 may determine the recommendation priority of the candidate listing location based on the score of the target feature of each candidate listing location. For example, the control platform 220 may add the score of the target feature of each candidate listing location to obtain a sum value, and use the sum value as the recommendation priority of the candidate listing location.

[0098] In one example, the control platform 220 may assign a score to the characteristic data of each target characteristic based on pre-stored experience values. For example, if the target characteristics include free location, power capacity, cooling capacity, network bandwidth, maintenance and access, cost-effectiveness, compliance, scalability, redundancy, security, network latency, and disaster recovery;

[0099] The preset score of the feature data of the target feature can be obtained based on the following rules and actual design requirements.

[0100] The preset score of the characteristic data of the idle position is positively correlated with the size of the available space of the idle position. The larger the available space is, the higher the preset score is.

[0101] The preset score of the characteristic data of the cooling capacity is positively correlated with the cooling efficiency. The higher the cooling efficiency, the higher the preset score.

[0102] The preset score of the characteristic data of the network bandwidth is positively correlated with the size of the available bandwidth. The larger the available bandwidth, the higher the preset score.

[0103] The preset scores of the maintenance and access feature data are positively correlated with the convenience of physical access. The more convenient the physical range is, the higher the preset score is.

[0104] The preset score of the characteristic data of cost efficiency is positively correlated with the sum of rent and fees. The higher the rent and fees, the higher the preset score.

[0105] The preset score of the compliance characteristic data is related to whether it complies with industry standards and regulatory requirements. If it complies with industry standards and regulatory requirements, it is copied as the first value; if it does not comply with industry standards and regulatory requirements, it is copied as the second value. The first value is higher than the second value, for example, the first value is 2 and the second value is 1.

[0106] The preset score of the scalability characteristic data is positively correlated with the expansion potential. The greater the expansion potential, the higher the preset score.

[0107] The preset score of the characteristic data of the redundancy strategy is related to the redundancy category. The redundancy categories include no redundancy, dual redundancy and single redundancy. For no redundancy, the preset score is set to a first value, for single redundancy, the preset score is set to a second value, and for dual redundancy, the preset score is set to a third value, and the third value is greater than the first value, for example, the first value is 2, the second value is 1, and the third value is 3.

[0108] The preset score of the security feature data is positively correlated with the security level. The security level includes high security level, low security level and medium security level. For the low security level, the preset score is set to a first value, for the medium security level, the preset score is set to a second value, and for the high security level, the preset score is set to a third value.

[0109] The preset score of the characteristic data of network delay is positively correlated with the level of network delay. Network delay includes low level, medium level and high level. In the embodiment of the present application, the level of network delay below the first time threshold can be low level, the level of network delay above or equal to the first time threshold and below or equal to the second time threshold can be medium level, and the level of network delay above the second time threshold can be high level. The first time threshold is less than the second time threshold, and the first time threshold and the second time threshold are time values ​​that meet the recommended optimal shelf position determined according to actual design needs.

[0110] The preset score of the characteristic data of disaster recovery is negatively correlated with the recovery time. The shorter the recovery time, the higher the preset score. The following is an explanation with a specific embodiment. Table 1 is a scoring table provided in an embodiment of the present application. Available space is explained as available U position.

[0111] Table 1

[0112] feature Scoring Criteria Available U-slots 1U:1,2U:2,4U:3 Power capacity 500W:1,750W:2,1000W:3 Cooling capacity 1500BTU:1,2250BTU:2,3000BTU:3 Network latency Low:1,Middle:2,High:3 redundancy None: 1, Single Redundancy: 2, Dual Redundancy: 3 Security Level Low:1,Middle:2,High:3 ……

[0113] Then the control platform 220 obtains the score of each feature in each candidate shelf position and obtains the total score of all features of each candidate shelf position. The control platform 220 compares the scores of each candidate shelf position and selects the candidate shelf position corresponding to the highest score (if the importance level is lower, the score is lower, and the lowest score is selected), and takes the candidate shelf position as the best shelf position.

[0114] For example, if there are three candidate listing positions obtained in step S330, namely position A, position B and position C, the control platform 220 can obtain the score of each candidate listing position in each feature based on Table 1, as shown in Table 2.

[0115] Table 2

[0116]

[0117] The control platform 220 may add up all feature scores of each position, and use the position with the highest score as the shelf position of the device to be shelf-mounted. Among them, the total score of position C is the highest, so position C is the best shelf position.

[0118] This approach fully considers target features such as available U-position, power capacity, cooling capacity, network latency, redundancy, and security level, ensuring that idle resources in the data center are used rationally, reducing waste and idleness of resources, and improving the space utilization of the data center. The rack location also fully considers the power supply, cooling, and network load of the region, reducing the occurrence of performance bottlenecks, and improving the overall service quality and response speed. Taking the security level as one of the important factors in selecting the rack location can ensure that key equipment and data are better protected physically and on the network, reducing security risks. By optimizing the layout, the redundant configuration of equipment, such as power redundancy and network redundancy, can be better achieved, improving the reliability and fault tolerance of the system.

[0119] Furthermore, by setting a preset score for the feature data of the target feature, the sum of the values ​​of each candidate shelf position is obtained, and the sum is used as the recommended priority of the candidate shelf position. This method of quantifying the recommended priority by score can reduce subjective judgment, improve the objectivity and consistency of the determination, and all candidate shelf positions are evaluated based on the same features and preset scores, so that the comparison between different positions is more open, the determination results are more accurate, and more in line with customer expectations.

[0120] In another example, since different target features have different importance levels in determining the shelf position, in order to make the obtained optimal shelf position more in line with actual needs, the control platform 220 also pre-assigns different weights to each feature based on different importance levels. As shown in Table 3, the control platform 220 assigns weights of 0.1 / 0.2 / 0.15 / 0.3 / 0.1 / 0.1 to available U-bit, power capacity, cooling capacity, network delay, redundancy, and security level, respectively.

[0121] Table 3

[0122]

[0123] The control platform 220 can obtain the total score of each candidate location based on the weights shown in Table 3 and the score of each feature of each candidate location:

[0124] Position A: (1*0.1)+(1*0.2)+(1*0.15)+(1*0.3)+(2*0.1)+(2*0.1)=1.15

[0125] Position B: (2*0.1)+(2*0.2)+(2*0.15)+(2*0.3)+(3*0.1)+(2*0.1)=1.55

[0126] Position C: (3*0.1)+(3*0.2)+(3*0.15)+(3*0.3)+(3*0.1)+(3*0.1)=2.1

[0127] Then sort by total score, and finally select the position with the highest total score. That is, the best listing position is position C.

[0128] Different data centers or different business needs may attach different importance to various features. For example, for applications that need to process large amounts of real-time data, network latency may be a critical factor and therefore needs to be given a higher weight. By assigning weights to each feature, you can ensure that the final selected shelf location best meets actual needs. By giving priority to features with higher weights, you can ensure that key resources in the data center (such as power, cooling, network, etc.) are more reasonably configured and used. This helps reduce resource waste, improve resource utilization, and reduce operating costs. Choosing a shelf location that meets actual needs and optimizes resource configuration can help improve the overall performance of the data center.

[0129] In another example, an embodiment of the present application provides a scoring card model. The scoring card model is a scoring model for candidate shelf locations based on target features, which is used to subsequently predict whether the candidate shelf locations meet actual needs.

[0130] The present application embodiment provides a method for constructing a scoring card model. Figure 6 Flowchart of the framework method of the scorecard model provided for the implementation of this application. The method includes the following contents:

[0131] S340.1: Obtain historical listing data.

[0132] The control platform 220 obtains a plurality of historical rack data. The historical rack data refers to the rack locations determined by the historical rack equipment, as well as the target characteristics based thereon and the space resources of the historical rack data center.

[0133] Furthermore, considering that the acquired historical shelf data may have problems such as missing data, data anomalies or errors that may affect the accuracy of the model, the control platform 220 may continue to execute S330.2 to pre-process the historical shelf data.

[0134] S340.2: Preprocess historical shelf data.

[0135] The control platform 220 pre-processes the historical shelf data to improve the data quality of the historical shelf data.

[0136] In one example, the control platform 220 can clean the historical data on the shelves, remove missing values, outliers or erroneous data in the historical data on the shelves, thereby avoiding the impact of missing values, outliers or erroneous data on the decision tree model and improving training efficiency and training accuracy.

[0137] In another example, the control platform 22 can also normalize the historical data after removing missing values, outliers or erroneous data in the historical data, convert data of different dimensions to the same dimension, enhance the generalization ability of the model, and improve the training effect.

[0138] Furthermore, in order to improve the training effect of the decision tree model, the control platform 220 can also divide the preprocessed historical shelf data into data sets, using part of the historical shelf data as a training set and another part of the data as a test set. For example, the preprocessed historical shelf data consists of N groups, of which N1 group of historical shelf data is used as a training set, and the remaining N-N1 groups of historical shelf data are used as a test set. Among them, N>1 is an integer, and N>N1>1 is an integer.

[0139] The training set is used to build the scorecard model, while the validation set is used to verify the accuracy of the scorecard model.

[0140] S340.3: Build a scorecard model based on the historical shelf data of the training set.

[0141] The control platform 220 performs training based on algorithms such as logistic regression, decision tree or its integrated method (such as gradient boosting tree) and historical shelf data of the training set to obtain a scorecard model. The scorecard model includes target features, weights of target features, and scoring criteria for specific situations of target features.

[0142] In one example, the constructed scoring card model includes a prediction algorithm for predicting a candidate listing position, and the prediction algorithm is:

[0143] Score=∑ i (P i ×Q i )

[0144] Among them, Score is the score of the candidate listing position, i represents the scoring items included in the scoring card model, i is the same as the number of target features, and each scoring item represents a target feature. i represents the preset score of the feature data of scoring item i, Q i It represents the weight value of rating item i. The weight value of rating item i is positively correlated with the importance of rating item i in determining the listing position.

[0145] S340.4: Evaluate the performance of the scoring model based on the validation set and adjust the model parameters of the scorecard model.

[0146] The control platform 220 may determine the performance of the trained scorecard based on the validation set and the evaluation parameters.

[0147] If the scorecard model does not meet the evaluation parameters, the model parameters such as the weight of the scorecard model can be adjusted.

[0148] S340.5: Obtain the trained scorecard model.

[0149] Then the control platform 220 can input the candidate shelf positions into the scoring card model, and obtain the best shelf position through the scoring card model processing. The best shelf position is the shelf position of the device to be shelf-mounted.

[0150] Determining the best shelf position through the scorecard model has the following advantages: The scorecard model is a model created based on existing historical shelf data, so the target features and target feature weights used in the scorecard model can reflect the statistical management between each feature and the best shelf position. This method reduces the impact of subjective judgment and improves the accuracy of decision-making. In addition, the scorecard model can capture hidden patterns and rules in the data, thereby accurately predicting new candidate shelf positions. This predictive ability helps to discover and avoid potential risks and problems in advance.

[0151] In summary, after the racking location is determined, the embodiment of the present application uses the spatial resources and load performance of the data center as candidate conditions, and based on the decision tree model, can identify the computer room or cabinet with a lower current load as the candidate racking location. This helps to balance the load distribution of the entire data center, avoid performance degradation or overheating in certain areas due to carrying too many devices, and thus improve overall performance and stability.

[0152] Furthermore, the embodiment of the present application adopts a decision tree model and a method of automatically screening out the best shelf location from multiple candidate shelf locations based on target features. Compared with manual screening of the shelf locations of the equipment to be listed, it is highly efficient and will not be affected by human subjectivity or the risk of false recommendations, thereby improving the acquisition efficiency and increasing the success rate of the equipment to be listed.

[0153] The following provides a specific example in combination with the above embodiment to explain in detail the method for obtaining the shelf location provided by the present application. In the following example, it is assumed that the electronic device is a server. Figure 7 As shown, attached Figure 7 A flowchart of another method for obtaining a shelf location provided in an embodiment of the present application includes the following specific contents:

[0154] S701: The terminal 210 applies to put a batch of servers on the shelves.

[0155] In a specific implementation, the user can determine the type and number of servers to be put on the shelf based on business requirements through the terminal 210 .

[0156] S702: The terminal 210 obtains target features.

[0157] In specific implementation, the user may input the target features through the terminal 210 .

[0158] Among them, users can enter target characteristics through strategy information, and specifically enter the U-position settings for listing. The listing strategies include room selection strategy, power selection strategy, and cabinet matching strategy.

[0159] S703: The terminal 210 sends the target characteristics to the control platform 220.

[0160] S704: The control platform 220 inputs the target feature into the decision tree model to obtain multiple candidate listing locations.

[0161] The decision tree model is shown in the attached Figure 4 The decision tree model is constructed in the manner shown.

[0162] S705: The control platform 220 inputs multiple candidate shelf positions into the scoring card model to obtain the best shelf position.

[0163] The scorecard model is attached Figure 6 The decision tree model is constructed in the manner shown.

[0164] S706: The control platform 220 sends the optimal shelf position to the terminal 210.

[0165] S707: The terminal 210 determines the listing plan.

[0166] S708: Terminal 210 shelf process approval.

[0167] In a specific implementation, the approval system can be accessed at the terminal 210 , and the listing plan can be approved through the approval system. The approval process needs to be implemented through the digital twin platform 220 .

[0168] S709: The terminal 210 determines that the process is finished and the approval is rejected.

[0169] S710: The terminal 210 determines that the approval is passed.

[0170] S711: The server is actually on the shelf.

[0171] The user puts the server on the data center 230 according to the putting plan determined in step 602 .

[0172] S712: The control platform 220 observes the actual operating status of the data center 230.

[0173] In a specific implementation, the operating data of the data center 230 can be collected through the control platform 220 to observe the actual operating status of the data center 230.

[0174] S713: The control platform 220 checks whether there is any abnormal situation. If not, execute step 716; if yes, execute step S714.

[0175] S714: The server is removed from the shelves.

[0176] The server put on the data center 230 is removed from the shelf.

[0177] S715: The terminal 210 analyzes the cause of the abnormality, adjusts the rack cabinet and the rack position, and executes step 702.

[0178] In a specific implementation, the control platform 220 may feed back the abnormal situation to the terminal 210, so that the terminal 210 can analyze the cause of the abnormality and adjust the listing plan.

[0179] S716: Shelf completed.

[0180] That is, when the embodiment of the present application recommends server shelf locations through decision trees and scoring card models, the decision tree can be used to screen out locations that meet specific conditions, and then the screened locations can be comprehensively scored using the scoring card model to assist in the manual process of quickly selecting the most suitable shelf location. The present invention improves the server shelf efficiency of large data centers, while reducing errors and omissions caused by human subjective factors, thereby avoiding waste of resources or safety accidents. It also improves the server shelf efficiency of data centers, and uses machine learning to assist humans in quickly and accurately finding the best shelf location for each server, thereby achieving the result of reducing costs and increasing efficiency. It increases the reliability of server shelf locations in data centers, reduces errors and omissions caused by human subjective factors, and avoids waste of resources or safety accidents; and the model itself will continue to evolve with learning iterations, and its accuracy will continue to improve.

[0181] In addition, the embodiment of the present application also provides a device for obtaining a shelf position. Figure 8 A schematic diagram of the structure of a device for obtaining a shelf position provided in an embodiment of the present application. The device includes:

[0182] An acquisition unit 801 is used to acquire target characteristics, where the target characteristics include idle position and load characteristics;

[0183] A feature determination unit 802, configured to determine a reference target feature from the target feature;

[0184] A screening unit 803 is used to screen at least one candidate shelf position from the free positions based on the reference target feature, and at least one candidate shelf position meets the requirement of the reference target feature;

[0185] The position determination unit 804 is used to determine a target racking position from a plurality of candidate racking positions based on the target characteristics, where the target racking position is an idle position that meets the requirements of the load characteristics.

[0186] Among them, the specific implementation of each unit refers to the shelf location acquisition method described in the first aspect and will not be discussed here.

[0187] In addition, an embodiment of the present application also provides a computing device, which includes a memory and a processor; the memory is used to store programs; the processor is used to execute the programs stored in the memory, and when the program stored in the memory is executed, the processor is used to execute any method as described in the first aspect.

[0188] In actual use, the computing device can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as basic cloud computing services such as large databases and artificial intelligence platforms. When the above-mentioned server is a server cluster composed of multiple physical servers, or a distributed system, multiple physical servers can form a blockchain, and each physical server is a node on the blockchain. The physical type of the service area can be a rack server, a high-density server, a GPU server, a tower server, and can also be a blade server, a whole cabinet server, etc., which is not specifically limited in this application.

[0189] In addition, an embodiment of the present application also provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on one or more computing devices, the one or more computing devices execute the shelf position processing method of the above embodiment.

[0190] In addition, the embodiment of the present application further provides a computer program product, when the computer program product is executed by one or more computing devices, the one or more computing devices execute any of the aforementioned methods for processing the shelf position. The computer program product may be a software installation package, and when any of the aforementioned methods for processing the shelf position is required, the computer program product may be downloaded and executed on a computer.

[0191] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the method of each embodiment of the present application.

[0192] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0193] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a training device or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device or data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integration. Available media can be magnetic media, (e.g., floppy disk, hard disk, tape), optical media (e.g., DVD), or semiconductor media (e.g., solid-state hard disk (SSD)), etc.

[0194] The system architecture and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. A person of ordinary skill in the art can appreciate that with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

Claims

1. A method for obtaining a shelf location, characterized in that: The method comprises: Acquiring target characteristics, wherein the target characteristics include idle position and load characteristics; determining a reference target feature from the target features; Based on the reference target feature, screening out at least one candidate shelf position from the idle positions, wherein the at least one candidate shelf position meets the requirement of the reference target feature; Based on the target feature, a target racking position is determined from the plurality of candidate racking positions, and the target racking position is the idle position that meets the requirements of the load feature.

2. The method according to claim 1, characterized in that: The step of determining a target shelf location from a plurality of candidate shelf locations based on the target feature includes: Determining a recommendation priority of the plurality of candidate listing locations, wherein a higher recommendation priority of the candidate listing location indicates a greater number of requirements of the target features that the candidate listing location satisfies; Based on the recommendation priority, the target shelf position is determined.

3. The method according to claim 2, characterized in that: The determining of the recommendation priorities of the plurality of candidate listing locations comprises: Processing each of the candidate shelf locations based on a preset determination method to determine a recommendation priority of the plurality of candidate shelf locations; The processing of each candidate shelf location based on a preset determination method includes: Based on the preset scores of the feature data of the target features, determining the scores of each of the target features of the candidate listing locations, wherein the target features include a plurality of feature data, and the higher the preset score of each feature data, the higher the recommendation priority of the target feature; Based on the score of each of the target features of the candidate listing locations, a recommendation priority of the candidate listing locations is determined.

4. The method according to claim 1, characterized in that: The target characteristics include the free location, power capacity, cooling capacity, network bandwidth, maintenance and access, cost-effectiveness, compliance, scalability, redundancy, security, network latency or disaster recovery, and the reference target characteristics include: available space, power capacity, cooling capacity or network bandwidth.

5. The method according to claim 1, characterized in that: The step of determining a target shelf location from a plurality of candidate shelf locations based on the target feature includes: Based on the target feature and a scorecard model, scores of the plurality of candidate listing locations are obtained, wherein the scorecard model is used to determine the scores of the candidate listing locations based on preset scores of feature data of the target feature; Based on the multiple scores, a target shelf location is determined from the multiple candidate shelf locations.

6. The method according to claim 5, characterized in that: The scorecard model includes a prediction algorithm for predicting the candidate listing location, and the prediction algorithm is: Score=∑ i (P i ×Q i ) Wherein, Score is the score of the candidate listing position, i represents the scoring items included in the scoring card model, i is the same as the number of the target features, and each scoring item represents one of the target features, P i represents the preset score of the feature data of scoring item i, Q i represents the weight value of the scoring item i, and the weight value of the scoring item i is positively correlated with the importance of the scoring item i in determining the shelf position.

7. The method according to claim 1, characterized in that: The determining of the reference target feature from the target feature comprises: determining the reference target feature from the target feature based on a decision tree model; The step of selecting at least one candidate shelf position from the idle positions based on the reference target feature includes: Based on the decision tree model and the reference target features, screening out at least one candidate shelf position from the idle positions; Among them, the decision tree model is a decision tree model trained based on historical shelving data, which is used to determine the reference target feature from the target feature and screen out at least one candidate shelving position from the idle position, and the historical shelving data includes the target feature and reference shelving position of historical shelving equipment.

8. The method according to claim 7, characterized in that: The decision tree model includes a root node, multiple child nodes and leaf nodes connected to each other: The root node of the decision tree model is a node for determining whether the power capacity is sufficient, the multiple child nodes include a child node for determining whether the cooling capacity is sufficient, a child node for determining whether the available space is sufficient, and a child node for determining whether the network bandwidth is sufficient, and the leaf nodes include leaf nodes that do not meet the conditions and leaf nodes that meet all the conditions.

9. The method according to claim 8, characterized in that: The step of selecting at least one candidate shelf position from the idle positions comprises: If the power supply capacity is sufficient, jump to the sub-node of whether the cooling capacity is sufficient; if the power supply capacity is insufficient, jump to the sub-node of whether the available space is sufficient; If the cooling capacity is sufficient, jump to the sub-node of whether the network bandwidth is sufficient; if the cooling capacity is insufficient, jump to the leaf node that does not meet the condition; If the available space is sufficient, jump to the child node of whether the network bandwidth is sufficient; if the available space is insufficient, jump to the leaf node of the condition not met; If the network bandwidth is sufficient, jump to the leaf node that meets all conditions; if the network bandwidth is insufficient, jump to the leaf node that does not meet the conditions; the shelf position of the leaf node that meets all conditions is the candidate shelf position.

10. A computing device, characterized in that: include: Memory, used to store programs; A processor, configured to execute the program stored in the memory; when the program stored in the memory is executed, the processor is configured to execute the method according to any one of claims 1 to 9.

Images