Multi-data center load collaborative control method, device and computer equipment based on information gap decision theory

Abstract

The present application relates to a method, device and computer equipment for collaborative control of multi-data center loads based on information gap decision theory. The method includes: constructing a data center load optimization model; initializing a data center group for quasi-cooperative game, and constructing a multi-data center load collaborative optimization model; establishing a multi-data center cost sharing model based on improved Shapley value; and establishing a multi-data center collaborative control model based on information gap decision theory according to the data center load optimization model, the multi-data center load collaborative optimization model and the improved Shapley value multi-data center cost sharing model. This method can achieve efficient coordination of computing resources, ensure reasonable distribution of benefits, and have the ability to cope with sudden computing needs.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a method, device and computer equipment for coordinated load control of multiple data centers based on information gap decision theory. Background Art

[0002] With the implementation of the "Eastern Data, Western Computing" strategy, investment in data center infrastructure and other infrastructure has increased. This initiative aims to adjust the distribution of computing power, direct computing demand from the east to the west, and fully utilize the abundant renewable energy resources in the west, such as wind, hydropower, and solar energy. This strategy not only benefits western power generation companies but also reduces the reliance of eastern data centers on fossil fuels. Furthermore, by synergizing the transfer of computing and electricity demand from the east to the west, the high electricity demand generated by high-density data centers in the east can be effectively avoided, effectively avoiding the need to invest in expensive ultra-high voltage transmission lines, thereby saving construction and operation and maintenance costs. Furthermore, this computing-electricity synergy model helps improve the overall energy efficiency of data centers. By spatially shifting computing demand, it maximizes the utilization of various high-efficiency computing resources and further reduces energy consumption.

[0003] However, in practice, efficient collaboration is difficult to achieve because data centers are owned by different operators and computing resources are not shared. Although data center operators can promote resource sharing through collaboration, they still face issues such as profit distribution. In addition, the existence of sudden computing demands brings uncertainty to resource sharing. Summary of the Invention

[0004] Based on this, it is necessary to provide a multi-data center load collaborative control method, device and computer equipment based on information gap decision theory to address the above technical problems.

[0005] In a first aspect, the present application provides a multi-data center load collaborative control method based on information gap decision theory, the method comprising:

[0006] Construct a data center load optimization model; the data center load optimization model is used to characterize the energy consumption of various devices within each data center;

[0007] Initialize a cluster of data centers for a quasi-cooperative game and construct a multi-data center load collaborative optimization model. The quasi-cooperative game data center cluster is a cluster of data centers from each operator that participate in resource allocation and energy consumption. The multi-data center load collaborative optimization model represents a model used by multiple data centers to optimize resource allocation and energy consumption when working together.

[0008] Establish a multi-data center cost sharing model based on improved Shapley value;

[0009] According to the data center load optimization model, the multi-data center load collaborative optimization model and the improved Shapley value multi-data center cost sharing model, a multi-data center collaborative control model based on information gap decision theory is established; the multi-data center collaborative control model based on information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers work together.

[0010] In one embodiment, building a data center load optimization model includes:

[0011] Extract key parameters of data center load; key parameters of data center load include: regional data center set, idle power of servers in the data center, peak power of servers in the data center, set of all servers in the data center, time interval and time of day set;

[0012] Based on the key parameters of data center load, a data center load pressure drop model is constructed;

[0013] Based on the key parameters of data center load and the data center load pressure drop model, a data center load optimization model is constructed.

[0014] In one embodiment, a data center load pressure drop model is constructed based on key data center load parameters, including:

[0015] Based on the key parameters of data center load, build server energy consumption optimization model, temperature control equipment energy consumption optimization model and distribution system energy consumption model; the key parameters of data center load include: regional data center collection , Server idle power in data centers , peak server power in data centers , data center All servers in the collection , time interval and time of day collection ;

[0016] Among them, the server energy consumption optimization model is:

[0017]

[0018] in, For data centers In the period Energy consumption generated by all servers; Indicates data center No. Servers in the time period Whether it is in the open state, if it is open, it is 1, otherwise it is 0; and is an auxiliary variable; Indicates data center No. Servers in the time period The CPU utilization of the virtual machine; CPU utilization thresholds set for data center servers; It is the maximum CPU utilization rate allowed for overclocking of data center servers to reduce overall energy consumption. is a maximum value; Indicates data center A collection of servers;

[0019] The energy consumption optimization model of temperature control equipment is:

[0020]

[0021] in, For data centers In the period Energy consumption of temperature control equipment; is the cooling efficiency function of the data center; Indicates data center Refrigeration equipment in the period Supply air temperature; is the heat distribution coefficient of the data center room; It is a data center Total number of servers;

[0022] The energy consumption model of the power distribution system is:

[0023]

[0024] in, is a coefficient that represents the ratio between the energy consumption of the power distribution system and the energy consumption of the server;

[0025] Build a data center load drop model based on the server energy consumption optimization model, temperature control equipment energy consumption optimization model, and power distribution system energy consumption model;

[0026] The data center load pressure drop model is:

[0027]

[0028] in, Indicates data center In the period The energy consumption generated.

[0029] In one embodiment, initializing a data center group for a collaborative game and building a multi-data center load collaborative optimization model includes:

[0030] by represents the data center cluster of the quasi-cooperative game, where , For collection Including data centers All subsets of ;

[0031] Construct computing demand statistical models, virtual machine resource allocation models, virtual machine bandwidth demand constraint models, and data center energy cost models;

[0032] The calculation demand statistical model is:

[0033]

[0034] in, For data centers IT users In the period To time period The calculation delay sequence of IT users In the period The minimum delay of calculation; For time period Statistical intervals about IT users; Represents the hierarchical user group set; is the number of user groups after classification, ( ) indicates the Level users, also refers to user groups ; Indicates the The upper limit of computing delay for users of level 1, The computing delay of the user level shall not exceed ; The interval used to calculate latency ratings for IT users; Indicates data center A collection of IT users; Indicates user group In the period The CPU utilization of the virtual machine; Indicates user group In the period The CPU utilization of virtual machines with uncertain computing needs should be considered; Indicates time period Data Center IT users Whether to include it in the user group ; Indicates data center Users In the period The CPU utilization of the virtual machine; Indicates data center Users In the period The CPU utilization of virtual machines due to uncertain computing demands;

[0035] The virtual machine resource allocation model is:

[0036]

[0037] in, It is the time period Virtual machine resources temporarily occupied due to random computing needs of IT users; Indicates data center In the period Virtual machine resources allocated temporarily to meet random computing needs of IT users; Data Center In the period The number of servers enabled, It is a data center Total number of servers;

[0038] The virtual machine resource bandwidth demand constraint model is:

[0039]

[0040] in, For data centers IT users From the period To time period The bandwidth requirement sequence of IT users In the period Maximum bandwidth requirements; Indicates data center Server The maximum bandwidth that can be provided; Indicates user group bandwidth requirements; Indicates user group In the period Bandwidth requirements for random computing needs;

[0041] The energy cost model of a data center is:

[0042]

[0043] in, Indicates data center In the period The energy cost, Indicates data center In the period The electricity price, Indicates data center In the period The unit incentives that can be obtained by participating in demand response to reduce load, Indicates data center In the period baseline load; Represents a collection of data centers All data centers within the period Total energy cost;

[0044] Based on the computing demand statistical model, virtual machine resource allocation model, virtual machine bandwidth demand constraint model and data center energy cost model, a multi-data center load collaborative optimization model is constructed.

[0045] In one embodiment, a multi-data center cost allocation model based on an improved Shapley value includes: a data center optimal cost function, a multi-data center Shapley value model, an improved Shapley value model considering computing demand uncertainty, a data center energy cost model based on Shapley value allocation, a modified data center energy cost model, an optimal energy cost model for a single data center, and an adaptive cost correction model;

[0046] Among them, the optimal cost function of the data center is:

[0047]

[0048] in, is the optimal energy cost function, and the input is , the objective function is to minimize the set Data centers within the period Total cost;

[0049] The multi-data center Shapley value model is:

[0050]

[0051] in, For data centers In the period The Shapley value of is the auxiliary variable introduced; Representing a collection The number of data centers, Representing a collection Number of data centers;

[0052] The improved Shapley value model considering the uncertainty of computing requirements is:

[0053]

[0054] in, For data centers In the period The modified Shapley value of is an auxiliary variable;

[0055] The data center energy cost model based on Shapley value allocation is:

[0056]

[0057] in, Indicates data center In the period Energy cost after allocation based on Shapley value;

[0058] The revised data center energy cost model is:

[0059]

[0060] in, For data centers In the period Energy costs based on modified Shapley values;

[0061] The optimal energy cost model for a single data center is:

[0062]

[0063] in, Data Center In the period The optimal energy cost without participating in the collaborative game is ;

[0064] The adaptive cost correction model is:

[0065]

[0066] Determine the data center set Each data center in Corresponding Is it established? If it is established, each data center In the period The energy cost is If it does not hold true, then each data center In the period The energy cost is .

[0067] In one embodiment, a multi-data center collaborative control model based on information gap decision theory is established according to a data center load optimization model, a multi-data center load collaborative optimization model, and an improved Shapley value multi-data center cost sharing model, including:

[0068] Sure When it is 0 ;

[0069] based on Solving for data center sets All data centers within the period The total energy cost is calculated and a cost constraint model based on information gap decision theory is constructed. The cost constraint model is:

[0070]

[0071] in, is the information gap decision theory coefficient, the initial value of t is 0, and t is assigned according to t=t+1;

[0072] make As variables, the optimization objective of the data center under the information gap decision theory is constructed; the optimization objective of the data center is:

[0073]

[0074] Solve the optimal solution for the data center optimization objective to determine the parameters of the following models and functions corresponding to the optimal solution:

[0075] Server energy consumption optimization model, temperature control equipment energy consumption optimization model, power distribution system energy consumption model, computing demand statistical model, virtual machine resource allocation model, virtual machine bandwidth demand constraint model, data center energy cost model, data center optimal cost function, multi-data center Shapley value model, improved Shapley value model considering computing demand uncertainty, data center energy cost model based on Shapley value allocation, revised data center energy cost model, single data center optimal energy cost model, adaptive cost correction model;

[0076] Output the coordinated control strategy of each data center based on the parameters of the model and function. The strategy includes 、 、 、 、 and .

[0077] In a second aspect, the present application further provides a multi-data center load collaborative control device based on information gap decision theory, the device comprising:

[0078] Data center load optimization model building module, used to build a data center load optimization model; the data center load optimization model is used to characterize the energy consumption of various devices within each data center;

[0079] The multi-data center load collaborative optimization model construction module is used to initialize a cluster of data centers that are quasi-collaborative and construct a multi-data center load collaborative optimization model. The cluster of data centers that are quasi-collaborative is a cluster of data centers from each operator that participate in resource allocation and energy consumption. The multi-data center load collaborative optimization model represents a model used by multiple data centers to optimize resource allocation and energy consumption when working together.

[0080] Multi-data center cost sharing model building module, used to establish a multi-data center cost sharing model based on improved Shapley value;

[0081] The multi-data center collaborative control model construction module is used to establish a multi-data center collaborative control model based on the information gap decision theory according to the data center load optimization model, the multi-data center load collaborative optimization model and the improved Shapley value multi-data center cost sharing model; the multi-data center collaborative control model based on the information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers work together.

[0082] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in the above embodiment when executing the computer program.

[0083] In a fourth aspect, the present application further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method in the above embodiment are implemented.

[0084] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements the steps of the method in the above embodiment when the computer program is executed by a processor.

[0085] The multi-data center load collaborative control method, device, and computer equipment based on information gap decision theory have at least the following beneficial effects:

[0086] By building a data center load optimization model, we determine the energy consumption of each data center. Based on this, we then construct a multi-data center load coordination optimization model based on computing requirements and energy consumption. This model determines the allocation ratio of computing resources to each data center. Furthermore, we establish a multi-data center cost allocation model based on the improved Shapley value to determine the cost of each data center corresponding to the computing resource allocation ratio, ensuring a reasonable distribution of benefits. Combining these three models and building a multi-data center coordination control model based on information gap decision theory, we can address sudden computing demands, achieve efficient coordination of computing resources, and ensure a reasonable distribution of benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0088] Figure 1 This is a diagram illustrating an application environment of a multi-data center load collaborative control method based on information gap decision theory in one embodiment;

[0089] Figure 2 1 is a flow chart of a method for coordinated load control of multiple data centers based on information gap decision theory in one embodiment;

[0090] Figure 3 A schematic diagram of a flow chart of steps for constructing a data center load optimization model in one embodiment;

[0091] Figure 4 A diagram of a multi-data center load coordination and control architecture in one embodiment;

[0092] Figure 5 1. A structural block diagram of a multi-data center load collaborative control device based on information gap decision theory in one embodiment;

[0093] Figure 6 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0094] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0095] The multi-data center 102 load coordination control method based on information gap decision theory provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, each data center 102 is connected to the data center 102 load collaborative control system 104, and the data center 102 load collaborative control system 104 is used to build a data center 102 load optimization model. The data center 102 load optimization model is used to characterize the energy consumption of various devices inside each data center 102; it is also used to initialize the data center 102 group that is quasi-cooperative game, and build a multi-data center 102 load collaborative optimization model. The data center 102 group that is quasi-cooperative game is a cluster composed of data centers 102 participating in resource allocation and energy consumption in each operator. The multi-data center 102 load collaborative optimization model characterizes multiple A model for optimizing resource allocation and energy consumption when data centers 102 work collaboratively; in addition, it is also used to establish a multi-data center 102 cost sharing model based on the improved Shapley value; the data center 102 load collaborative control system 104 establishes a multi-data center 102 collaborative regulation model based on the information gap decision theory based on the data center 102 load optimization model, the multi-data center 102 load collaborative optimization model and the improved Shapley value multi-data center 102 cost sharing model; the multi-data center 102 collaborative regulation model based on the information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers 102 work collaboratively.

[0096] In an exemplary embodiment, Figure 2 As shown in the figure, a multi-data center load coordination method based on information gap decision theory is provided. Figure 1 The data center load collaborative control system in FIG. 1 is described as an example, and includes the following steps S202 to S206. Among them:

[0097] S202, constructing a data center load optimization model; the data center load optimization model is used to characterize the energy consumption of various devices in each data center.

[0098] For example, operational data from each data center's internal equipment (such as servers, temperature control equipment, and power distribution systems) is collected and analyzed, including but not limited to CPU utilization, memory usage, network traffic, and power consumption. Based on this data, a data center load optimization model is constructed to determine the energy consumption of each data center. This model can be a mathematical model, such as a linear programming model or a mixed integer programming model, using data center energy consumption as the objective function and considering factors such as device load and task scheduling as constraints.

[0099] S204, initialize the data center group for the quasi-cooperative game and construct a multi-data center load collaborative optimization model; the data center group for the quasi-cooperative game is a cluster composed of data centers of each operator that participate in resource allocation and energy consumption; the multi-data center load collaborative optimization model represents a model used by multiple data centers to optimize resource allocation and energy consumption when working together.

[0100] For example, data centers participating in computing collaboration are identified from each operator and treated as a data center cluster. The data centers' operational data, including but not limited to each data center's current load level, power consumption, cooling system efficiency, network latency, and any other key performance-related metrics, are collected to serve as the basis for a multi-data center load collaborative optimization model. Based on the collected data and actual computing needs, a multi-data center load collaborative optimization model is constructed to determine virtual machine resource allocation, virtual machine bandwidth requirements, and data center energy costs.

[0101] S206: Establish a multi-data center cost sharing model based on the improved Shapley value.

[0102] For example, the Shapley value is a method for fairly distributing benefits or costs in a cooperative game. In this step, an improved Shapley value algorithm is designed to fairly distribute the cost savings or increased costs of each data center due to participating in collaborative optimization. Specifically, a multi-data center cost sharing model based on the improved Shapley value can be constructed based on models and parameters such as the data center optimal cost function, a multi-data center Shapley value model, an improved Shapley value model that considers computing demand uncertainty, a data center energy cost model based on Shapley value allocation, a modified data center energy cost model, a single data center optimal energy cost model, and an adaptive cost correction model.

[0103] S208, based on the data center load optimization model, the multi-data center load collaborative optimization model and the improved Shapley value multi-data center cost sharing model, establish a multi-data center collaborative control model based on the information gap decision theory; the multi-data center collaborative control model based on the information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers work together.

[0104] For example, based on the data center load optimization model, multi-data center load collaborative optimization model and improved Shapley value multi-data center cost sharing model determined in the above steps, the three are combined to establish a multi-data center collaborative control model based on information gap decision theory, so as to achieve efficient collaboration of computing resources of different operators.

[0105] The multi-data center load coordination and control method based on information gap decision theory constructs a data center load optimization model to determine the energy consumption of each data center. Based on this, a multi-data center load coordination optimization model is constructed based on computing requirements and energy consumption. This model determines the allocation ratio of computing resources to each data center. Furthermore, a multi-data center cost allocation model based on the improved Shapley value is established to determine the cost of each data center corresponding to the computing resource allocation ratio, ensuring a reasonable distribution of benefits. Combining the above three models and building a multi-data center coordination and control model based on information gap decision theory can help address sudden computing demands, thereby achieving efficient coordination of computing resources and ensuring a reasonable distribution of benefits.

[0106] In an exemplary embodiment, Figure 3 As shown, a data center load optimization model is constructed, including:

[0107] S302, extracting key parameters of data center load; key parameters of data center load include: regional data center set, idle power of data center servers, peak power of data center servers, set of all servers in the data center, time interval and time of day set.

[0108] S304: Build a data center load pressure drop model based on key data center load parameters.

[0109] S306: Building a data center load optimization model based on the data center load key parameters and the data center load pressure drop model.

[0110] Exemplarily, key parameters of data center load are extracted, including the set of regional data centers, the idle power of servers in the data center, the peak power of the servers, the set of all servers, the time interval, and the time set of the day; and based on these key parameters, a data center load drop model is constructed to simulate the load changes of the data center at different time points to help predict and control the energy consumption of the data center; further, based on the previously extracted key parameters and the constructed load drop model, a data center load optimization model is further constructed to achieve effective management of internal resources of the data center and optimization of energy consumption by comprehensively considering the energy consumption of the data center, the working status of the server, and time factors.

[0111] In this embodiment, by finely managing the load of the data center, unnecessary energy consumption is reduced and resource utilization is improved, thereby achieving the goal of energy conservation and emission reduction. In addition, by optimizing the load of the data center, the service quality and operating efficiency of the data center can also be improved, ensuring that while meeting business needs, costs are saved to the greatest extent and the reliability and stability of the system are improved.

[0112] In an exemplary embodiment, a data center load pressure drop model is constructed based on key data center load parameters, including:

[0113] Based on the key parameters of data center load, build server energy consumption optimization model, temperature control equipment energy consumption optimization model and distribution system energy consumption model; the key parameters of data center load include: regional data center collection , Server idle power in data centers , peak server power in data centers , data center All servers in the collection , time interval and time of day collection ;

[0114] Among them, the server energy consumption optimization model is:

[0115]

[0116] in, For data centers In the period Energy consumption generated by all servers; Indicates data center No. Servers in the time period Whether it is in the open state, if it is open, it is 1, otherwise it is 0; and is an auxiliary variable; Indicates data center No. Servers in the time period The CPU utilization of the virtual machine; CPU utilization thresholds set for data center servers; It is the maximum CPU utilization rate allowed for overclocking of data center servers to reduce overall energy consumption. is a maximum value; Indicates data center A collection of servers;

[0117] The energy consumption optimization model of temperature control equipment is:

[0118]

[0119] in, For data centers In the period Energy consumption of temperature control equipment; is the cooling efficiency function of the data center; Indicates data center Refrigeration equipment in the period Supply air temperature; is the heat distribution coefficient of the data center room; It is a data center Total number of servers;

[0120] The energy consumption model of the power distribution system is:

[0121]

[0122] in, is a coefficient that represents the ratio between the energy consumption of the power distribution system and the energy consumption of the server;

[0123] Build a data center load drop model based on the server energy consumption optimization model, temperature control equipment energy consumption optimization model, and power distribution system energy consumption model;

[0124] The data center load pressure drop model is:

[0125]

[0126] in, Indicates data center In the period The energy consumption generated.

[0127] In this embodiment, by constructing a server energy consumption optimization model, a temperature control equipment energy consumption optimization model and a power distribution system energy consumption model based on the key parameters of the data center load, and further integrating these models to construct a data center load pressure drop model, it is possible to achieve refined management and optimization of the energy consumption of the data center in different time periods. This not only helps to accurately control the energy consumption of servers within the data center, ensuring that energy consumption is minimized while meeting computing needs, but also can further reduce overall energy consumption and improve the energy efficiency ratio of the data center by optimizing the working efficiency of the temperature control equipment and the energy consumption ratio of the power distribution system, thereby achieving the purpose of energy conservation and emission reduction. Moreover, through the dynamic adjustment mechanism of the model, it can better cope with changes in computing needs, ensure the efficient operation and service quality of the data center in different time periods, and ultimately achieve significant optimization of the data center in resource utilization and energy consumption management.

[0128] In an exemplary embodiment, initializing a data center group for a simulated collaborative game and constructing a multi-data center load collaborative optimization model includes:

[0129] by represents the data center cluster of the quasi-cooperative game, where , For collection Including data centers All subsets of ;

[0130] Construct computing demand statistical models, virtual machine resource allocation models, virtual machine bandwidth demand constraint models, and data center energy cost models;

[0131] The calculation demand statistical model is:

[0132]

[0133] in, For data centers IT users In the period To time period The calculation delay sequence of IT users In the period The minimum delay of calculation; For time period Statistical intervals about IT users; Represents the hierarchical user group set; is the number of user groups after classification, ( ) indicates the Level users, also refers to user groups ; Indicates the The upper limit of computing delay for users of level 1, The computing delay of the user level shall not exceed ; The interval used to calculate latency ratings for IT users; Indicates data center A collection of IT users; Indicates user group In the period The CPU utilization of the virtual machine; Indicates user group In the period The CPU utilization of virtual machines with uncertain computing needs should be considered; Indicates time period Data Center IT users Whether to include it in the user group ; Indicates data center Users In the period The CPU utilization of the virtual machine; Indicates data center Users In the period The CPU utilization of virtual machines due to uncertain computing demands;

[0134] The virtual machine resource allocation model is:

[0135]

[0136] in, It is the time period Virtual machine resources temporarily occupied due to random computing needs of IT users; Indicates data center In the period Virtual machine resources allocated temporarily to meet random computing needs of IT users; Data Center In the period The number of servers enabled, It is a data center Total number of servers;

[0137] The virtual machine resource bandwidth demand constraint model is:

[0138]

[0139] in, For data centers IT users From the period To time period The bandwidth requirement sequence of IT users In the period Maximum bandwidth requirements; Indicates data center Server The maximum bandwidth that can be provided; Indicates user group bandwidth requirements; Indicates user group In the period Bandwidth requirements for random computing needs;

[0140] The energy cost model of a data center is:

[0141]

[0142] in, Indicates data center In the period The energy cost, Indicates data center In the period The electricity price, Indicates data center In the period The unit incentives that can be obtained by participating in demand response to reduce load, Indicates data center In the period baseline load; Represents a collection of data centers All data centers within the period Total energy cost;

[0143] Based on the computing demand statistical model, virtual machine resource allocation model, virtual machine bandwidth demand constraint model and data center energy cost model, a multi-data center load collaborative optimization model is constructed.

[0144] In this embodiment, by initializing a data center group for collaborative gaming and constructing a computing demand statistical model, a virtual machine resource allocation model, a virtual machine bandwidth demand constraint model, and a data center energy cost model, a multi-data center load collaborative optimization model is formed. This model can accurately record the computing demand of the data center in different time periods, and optimize according to the computing latency requirements and virtual machine resource allocation of IT users. At the same time, it takes into account the constraints of the virtual machine bandwidth requirements to ensure that resources are effectively allocated while meeting the service quality. In addition, by constructing a data center energy cost model, it can comprehensively consider factors such as electricity prices, demand response incentives, and baseline loads to optimize the energy consumption management of the data center. This ensures that the data center cluster can work efficiently and collaboratively in the face of dynamically changing computing demands while improving resource utilization and reducing operating costs, thereby achieving a dual improvement in energy conservation and emission reduction and economic benefits.

[0145] In an exemplary embodiment, a multi-data center cost allocation model based on an improved Shapley value includes: a data center optimal cost function, a multi-data center Shapley value model, an improved Shapley value model considering computing demand uncertainty, a data center energy cost model based on Shapley value allocation, a modified data center energy cost model, an optimal energy cost model for a single data center, and an adaptive cost correction model;

[0146] Among them, the optimal cost function of the data center is:

[0147]

[0148] in, is the optimal energy cost function, and the input is , the objective function is to minimize the set Data centers within the period Total cost;

[0149] The multi-data center Shapley value model is:

[0150]

[0151] in, For data centers In the period The Shapley value of is the auxiliary variable introduced; Representing a collection The number of data centers, Representing a collection Number of data centers;

[0152] The improved Shapley value model considering the uncertainty of computing requirements is:

[0153]

[0154] in, For data centers In the period The modified Shapley value of is an auxiliary variable;

[0155] The data center energy cost model based on Shapley value allocation is:

[0156]

[0157] in, Indicates data center In the period Energy cost after allocation based on Shapley value;

[0158] The revised data center energy cost model is:

[0159]

[0160] in, For data centers In the period Energy costs based on modified Shapley values;

[0161] The optimal energy cost model for a single data center is:

[0162]

[0163] in, Data Center In the period The optimal energy cost without participating in the collaborative game is ;

[0164] The adaptive cost correction model is:

[0165]

[0166] Determine the data center set Each data center in Corresponding Is it established? If it is established, each data center In the period The energy cost is If it does not hold true, then each data center In the period The energy cost is .

[0167] In this embodiment, by constructing a multi-data center cost sharing model based on an improved Shapley value, including a data center optimal cost function, a multi-data center Shapley value model, an improved Shapley value model considering computing demand uncertainty, a data center energy cost model based on Shapley value sharing, a corrected data center energy cost model, a single data center optimal energy cost model and an adaptive cost correction model, it is possible to achieve fair and reasonable sharing of the costs incurred by the data centers in the collaborative optimization process, ensuring that each data center shares the energy-saving effect or bears additional costs according to the degree of its contribution to the overall optimization, thereby promoting efficient collaboration between data centers and enhancing the stability and flexibility of the system. At the same time, by introducing an improved model of computing demand uncertainty and an adaptive cost correction mechanism, the model's adaptability to actual application scenarios is further improved, ensuring that optimal resource allocation and minimization of energy consumption can be achieved even in a dynamically changing environment, and ultimately achieving significant optimization of the data center cluster in energy conservation, emission reduction and cost control.

[0168] In an exemplary embodiment, a multi-data center collaborative control model based on information gap decision theory is established based on a data center load optimization model, a multi-data center load collaborative optimization model, and an improved Shapley value multi-data center cost sharing model, including:

[0169] Sure When it is 0 ;

[0170] based on Solving for data center sets All data centers within the period The total energy cost is calculated and a cost constraint model based on information gap decision theory is constructed. The cost constraint model is:

[0171]

[0172] in, is the information gap decision theory coefficient, the initial value of t is 0, and t is assigned according to t=t+1;

[0173] make As variables, the optimization objective of the data center under the information gap decision theory is constructed; the optimization objective of the data center is:

[0174]

[0175] Solve the optimal solution for the data center optimization objective to determine the parameters of the following models and functions corresponding to the optimal solution:

[0176] Server energy consumption optimization model, temperature control equipment energy consumption optimization model, power distribution system energy consumption model, computing demand statistical model, virtual machine resource allocation model, virtual machine bandwidth demand constraint model, data center energy cost model, data center optimal cost function, multi-data center Shapley value model, improved Shapley value model considering computing demand uncertainty, data center energy cost model based on Shapley value allocation, revised data center energy cost model, single data center optimal energy cost model, adaptive cost correction model;

[0177] Output the coordinated control strategy of each data center based on the parameters of the model and function. The strategy includes 、 、 、 、 and .

[0178] In this embodiment, a multi-data center collaborative control model is established based on the information gap decision theory, and the data center load optimization model, the multi-data center load collaborative optimization model and the improved Shapley value multi-data center cost sharing model are comprehensively considered to determine the information gap decision theory coefficient under the initial conditions, and the total energy consumption cost of the data center set in different time periods is solved by constructing a cost constraint model. Then, based on the information gap decision theory, a data center optimization target is constructed, and the optimal solution of the optimization target is solved to determine the parameters of each model and function. Finally, the collaborative control strategy of each data center is output, thereby realizing the optimization of the data center cluster in resource allocation and energy consumption management, improving the overall energy efficiency ratio, reducing operating costs, and promoting efficient collaboration and sustainable development among data centers.

[0179] In order to more intuitively illustrate the series of model construction involved in the multi-data center load coordination control method based on information gap decision theory of this application, please refer to the following Figure 4 shown.

[0180] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0181] Based on the same inventive concept, an embodiment of the present application also provides a multi-data center load collaborative control device based on information gap decision theory for implementing the multi-data center load collaborative control method based on information gap decision theory involved above. The implementation solution provided by the device is similar to the implementation solution described in the above method. Therefore, the specific limitations in the embodiments of one or more multi-data center load collaborative control devices based on information gap decision theory provided below can be found in the above limitations on the multi-data center load collaborative control method based on information gap decision theory, and will not be repeated here.

[0182] In an exemplary embodiment, Figure 5 As shown, a multi-data center load collaborative control device based on information gap decision theory is provided, including: a data center load optimization model construction module 502, a multi-data center load collaborative optimization model construction module 504, a multi-data center cost sharing model construction module 506 and a multi-data center collaborative control model construction module 508, wherein:

[0183] The data center load optimization model construction module 502 is used to construct a data center load optimization model; the data center load optimization model is used to characterize the energy consumption of various devices in each data center.

[0184] The multi-data center load collaborative optimization model construction module 504 is used to initialize the data center group for the quasi-collaborative game and construct a multi-data center load collaborative optimization model; the data center group for the quasi-collaborative game is a cluster composed of data centers participating in resource allocation and energy consumption among various operators; the multi-data center load collaborative optimization model represents a model used by multiple data centers to optimize resource allocation and energy consumption when working collaboratively.

[0185] The multi-data center cost sharing model building module 506 is used to build a multi-data center cost sharing model based on the improved Shapley value.

[0186] The multi-data center collaborative control model construction module 508 is used to establish a multi-data center collaborative control model based on the information gap decision theory according to the data center load optimization model, the multi-data center load collaborative optimization model and the improved Shapley value multi-data center cost sharing model; the multi-data center collaborative control model based on the information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers work together.

[0187] In an exemplary embodiment, the data center load optimization model building module 502 further includes:

[0188] The parameter extraction unit is used to extract key parameters of the data center load; the key parameters of the data center load include: regional data center set, idle power of the server in the data center, peak power of the server in the data center, set of all servers in the data center, time interval and time of day set.

[0189] The data center load pressure drop model building unit is used to build a data center load pressure drop model based on key data center load parameters.

[0190] The data center load optimization model building unit is used to build a data center load optimization model based on the data center load key parameters and the data center load pressure drop model.

[0191] Each module in the aforementioned multi-data center load coordination and control device based on information gap decision theory can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor within a computer device in the form of hardware, or can be stored in a computer device's memory in the form of software, allowing the processor to call and execute the corresponding operations of each module.

[0192] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 6As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store load data and model data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a multi-data center load collaborative control method based on information gap decision theory is implemented.

[0193] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0194] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0195] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0196] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0197] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0198] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0199] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A multi-data center load coordination and control method based on information gap decision theory, characterized in that: The method comprises: Constructing a data center load optimization model; the data center load optimization model is used to characterize the energy consumption of various devices within each data center; Initializing a data center cluster for a quasi-cooperative game and constructing a multi-data center load collaborative optimization model; the data center cluster for the quasi-cooperative game is a cluster consisting of data centers from each operator that participate in resource allocation and energy consumption; the multi-data center load collaborative optimization model represents a model for optimizing resource allocation and energy consumption when multiple data centers work collaboratively; Establish a multi-data center cost sharing model based on improved Shapley value; A multi-data center collaborative control model based on information gap decision theory is established based on the data center load optimization model, the multi-data center load collaborative optimization model, and the improved Shapley value multi-data center cost sharing model; the multi-data center collaborative control model based on information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers work collaboratively; The process of initializing a data center group for a quasi-cooperative game and constructing a multi-data center load collaborative optimization model includes: Is represents the data center group of the proposed collaborative game, where Is is all subsets of set I that include data center i; Construct computing demand statistical models, virtual machine resource allocation models, virtual machine bandwidth demand constraint models, and data center energy cost models; The computing demand statistical model is: in, is the computational delay sequence of IT user j in data center i from period t-1 to period t; represents the minimum computational delay of IT user j in time period t; Δτ is the statistical interval of IT users in time period t; M represents the set of hierarchical user groups; |M| is the number of hierarchical user groups, m (m∈M) represents the mth level user, also representing user group m; Indicates the upper limit of the computing delay for the mth level user. The computing delay of the mth level user shall not exceed ΔT cal The interval used to calculate the latency rating for IT users; i represents the set of IT users in data center i; u m,t represents the CPU utilization of the virtual machines of user group m in time period t; represents the CPU utilization of virtual machines with uncertain computing demands for user group m in period t; z i,j,m,t Indicates whether IT user j in data center i during period t is included in user group m; u i,j,t represents the CPU utilization of the virtual machine of user j in data center i during time period t; represents the CPU utilization of the virtual machine generated by the uncertain computing demand of user j in data center i during time period t; The virtual machine resource allocation model is: Among them, u i,k,t represents the CPU utilization of the virtual machine of the kth server in data center i during time period t; n i,t The number of servers enabled in data center i during period t, is the total number of servers in data center i; The virtual machine resource bandwidth requirement constraint model is: in, is the bandwidth demand sequence of IT user j in data center i from period t-1 to period t; represents the maximum bandwidth demand of IT user j in time period t; represents the maximum bandwidth that server k in data center i can provide; b m,t represents the bandwidth requirement of user group m; represents the bandwidth requirement of user group m for random computing needs in time period t; The data center energy cost model is: in, represents the energy cost of data center i in period t, q i,t represents the electricity price of data center i in time period t, e i,t represents the unit incentive that data center i can obtain by participating in demand response to reduce load in period t, represents the baseline load of data center i in time period t; represents the total energy cost of all data centers in the data center set Is in time period t; The multi-data center load collaborative optimization model is constructed based on the computing demand statistical model, the virtual machine resource allocation model, the virtual machine bandwidth demand constraint model and the data center energy cost model.

2. The multi-data center load coordinated control method based on information gap decision theory according to claim 1 is characterized in that: The constructing of the data center load optimization model includes: Extracting key data center load parameters; the key data center load parameters include: regional data center set, data center server idle power, data center server peak power, data center all server set, time interval and day time set; Based on the key data center load parameters, a data center load pressure drop model is constructed; The data center load optimization model is constructed based on the data center load key parameters and the data center load pressure drop model.

3. The multi-data center load coordinated control method based on information gap decision theory according to claim 2 is characterized in that: The step of constructing a data center load pressure drop model based on the data center load key parameters includes: Based on the key parameters of the data center load, a server energy consumption optimization model, a temperature control equipment energy consumption optimization model and a power distribution system energy consumption model are constructed; the key parameters of the data center load include: regional data center set I, data center server idle power P idle , the peak power of the server in the data center P dyn , the set of all servers K in data center i i , time interval Δt and time set T of a day; The server energy consumption optimization model is: in, is the energy consumption of all servers in data center i during time period t; i,k,t Indicates whether the kth server in data center i is in the on state during time period t. If it is on, it is 1, otherwise it is 0; x i,k,t and s i,k,t is an auxiliary variable; u i,k,t represents the CPU utilization of the virtual machine of the k-th server in data center i during time period t; CPU utilization thresholds set for data center servers; It is the maximum CPU utilization rate that the data center server is allowed to overclock to reduce the overall energy consumption; O is a maximum value; K i represents the set of servers in data center i; The energy consumption optimization model of the temperature control equipment is: in, is the energy consumption of the temperature control equipment of data center i in time period t; COP(·) is the cooling efficiency function of the data center; represents the air supply temperature of the cooling equipment of data center i in time period t; Z is the heat distribution coefficient of the data center room; |K i | is the total number of servers in data center i; The energy consumption model of the power distribution system is: Among them, α is a coefficient that represents the ratio between the energy consumption of the power distribution system and the energy consumption of the server; Constructing the data center load voltage drop model based on the server energy consumption optimization model, the temperature control equipment energy consumption optimization model and the power distribution system energy consumption model; The data center load pressure drop model is: in, represents the energy consumption of data center i in time period t.

4. The multi-data center load coordinated control method based on information gap decision theory according to claim 1 is characterized in that: The multi-data center cost sharing model based on the improved Shapley value includes: a data center optimal cost function, a multi-data center Shapley value model, an improved Shapley value model considering computing demand uncertainty, a data center energy cost model based on Shapley value sharing, a modified data center energy cost model, an optimal energy cost model for a single data center, and an adaptive cost correction model; The optimal cost function of the data center is: Where f is the optimal energy cost function, the input is Is, and the objective function is to minimize the total cost of the data centers in the set Is in time period t; The multi-data center Shapley value model is: Among them, V i,t is the Shapley value of data center i in time period t; j is an auxiliary variable introduced; |Is| represents the number of data centers in set Is, and |I| represents the number of data centers in set I; The improved Shapley value model considering the uncertainty of computing requirements is: in, represents the virtual machine resources temporarily allocated by data center i in time period t to meet the random computing needs of IT users; is the modified Shapley value of data center i in period t; is an auxiliary variable; The data center energy cost model based on Shapley value allocation is: in, represents the energy cost of data center i in period t after amortization based on the Shapley value; The modified data center energy cost model is: in, is the energy cost of data center i in period t based on the modified Shapley value; The optimal energy cost model for a single data center is: in, The optimal energy cost of data center i in time period t when it does not participate in the collaborative game, at this time Is = i; The adaptive cost correction model is: Determine the corresponding Is it true? If it is true, the energy cost of each data center i in time period t is If this is not true, the energy cost of each data center i in time period t is 5. The method for coordinated load control of multiple data centers based on information gap decision theory according to claim 4 is characterized in that: The multi-data center collaborative control model based on the information gap decision theory is established according to the data center load optimization model, the multi-data center load collaborative optimization model, and the improved Shapley value multi-data center cost sharing model, including: Sure f(Is) when it is 0; Based on f(Is), the total energy cost of all data centers in the data center set Is in time period t is solved, and a cost constraint model based on information gap decision theory is constructed. The cost constraint model is: in, is the information gap decision theory coefficient, the initial value of t is 0, and t is assigned according to t=t+1; make As variables, the data center optimization objective under the information gap decision theory is constructed; the data center optimization objective is: The optimal solution of the data center optimization objective is solved to determine the parameters of the following models and functions corresponding to the optimal solution: Server energy consumption optimization model, temperature control equipment energy consumption optimization model, power distribution system energy consumption model, computing demand statistical model, virtual machine resource allocation model, virtual machine bandwidth demand constraint model, data center energy cost model, data center optimal cost function, multi-data center Shapley value model, improved Shapley value model considering computing demand uncertainty, data center energy cost model based on Shapley value allocation, revised data center energy cost model, single data center optimal energy cost model, adaptive cost correction model; Based on the parameters of the model and function, the coordinated control strategy of each data center is output. The strategy includes u i,k,t 、y i,k,t 、 n i,t 、 and 6. A multi-data center load collaborative control device based on information gap decision theory, characterized in that: The device comprises: A data center load optimization model building module is used to build a data center load optimization model; the data center load optimization model is used to characterize the energy consumption of various devices within each data center; A multi-data center load collaborative optimization model construction module is used to initialize a cluster of data centers that are slated for collaborative gaming and construct a multi-data center load collaborative optimization model; the cluster of data centers slated for collaborative gaming is a cluster of data centers from each operator that participate in resource allocation and energy consumption; the multi-data center load collaborative optimization model represents a model used by multiple data centers to optimize resource allocation and energy consumption when working collaboratively; Multi-data center cost sharing model building module, used to establish a multi-data center cost sharing model based on improved Shapley value; A multi-data center collaborative control model construction module is used to establish a multi-data center collaborative control model based on information gap decision theory based on the data center load optimization model, the multi-data center load collaborative optimization model, and the improved Shapley value multi-data center cost sharing model; the multi-data center collaborative control model based on information gap decision theory is used to optimize resource allocation and energy consumption management when multiple data centers work together; The process of initializing a data center group for a quasi-cooperative game and constructing a multi-data center load collaborative optimization model includes: Is represents the data center group of the proposed collaborative game, where Is is all subsets of set I that include data center i; Construct computing demand statistical models, virtual machine resource allocation models, virtual machine bandwidth demand constraint models, and data center energy cost models; The computing demand statistical model is: in, is the computational delay sequence of IT user j in data center i from period t-1 to period t; represents the minimum computational delay of IT user j in time period t; Δτ is the statistical interval of IT users in time period t; M represents the set of hierarchical user groups; |M| is the number of hierarchical user groups, m (m∈M) represents the mth level user, also representing user group m; Indicates the upper limit of computing delay for users at level m. The computing delay for users at level m shall not exceed ΔT cal The interval used to calculate the latency rating for IT users; i represents the set of IT users in data center i; u m,t represents the CPU utilization of the virtual machines of user group m in time period t; represents the CPU utilization of virtual machines with uncertain computing demands for user group m in period t; z i,j,m,t Indicates whether IT user j in data center i during period t is included in user group m; u i,j,t represents the CPU utilization of the virtual machine of user j in data center i during time period t; represents the CPU utilization of the virtual machine generated by the uncertain computing demand of user j in data center i during time period t; The virtual machine resource allocation model is: Among them, u i,k,t represents the CPU utilization of the virtual machine of the kth server in data center i during time period t; n i,t The number of servers enabled in data center i during period t, is the total number of servers in data center i; The virtual machine resource bandwidth requirement constraint model is: in, is the bandwidth demand sequence of IT user j in data center i from period t-1 to period t; represents the maximum bandwidth demand of IT user j in time period t; represents the maximum bandwidth that server k in data center i can provide; b m,t represents the bandwidth requirement of user group m; represents the bandwidth requirement of user group m for random computing needs in time period t; The data center energy cost model is: in, represents the energy cost of data center i in period t, q i,t represents the electricity price of data center i in time period t, e i,t represents the unit incentive that data center i can obtain by participating in demand response to reduce load in period t, represents the baseline load of data center i in time period t; represents the total energy cost of all data centers in the data center set Is in time period t; The multi-data center load collaborative optimization model is constructed based on the computing demand statistical model, the virtual machine resource allocation model, the virtual machine bandwidth demand constraint model and the data center energy cost model.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images