A Data Center Demand Response Optimization Method Considering Network Channel Capacity and Node Uncertainty

Abstract

This invention relates to the field of data center demand response optimization technology, solving the technical problem of data centers being affected by network uncertainties when participating in demand response. In particular, it relates to a data center demand response optimization method that considers network channel capacity and node uncertainties. This method includes the following steps: constructing a network transmission path planning model for data center m to transfer workload to data center n; constructing a first data center energy consumption model to reduce the load on data center m; and a second data center energy consumption model for data center n to absorb renewable energy. This invention can avoid the impact of numerous uncertainties in the network on data center participation in demand response, effectively modeling the uncertainties of network capacity and node transmission workload, analyzing the most extreme cases, and formulating corresponding data center demand response strategies to ensure stable operation and cost control of the data center.

Description

Technical Field

[0001] This invention relates to the field of data center demand response optimization technology, and in particular to a data center demand response optimization method that takes into account network channel capacity and node uncertainty. Background Technology

[0002] As a key component of new infrastructure, data centers are rapidly increasing in number and scale, exacerbating the imbalance between electricity supply and demand in certain regions. In the context of data sharing between eastern and western regions, data centers are strengthening the synergy between computing power and electricity, hoping to achieve spatial transfer of computing power and tasks through networks, thereby shifting electricity demand and maximizing the utilization of power resources in different areas. To achieve this synergy, power grid companies are attempting a demand response approach, providing subsidies to data center operators to incentivize them to alter the spatial distribution of data center workloads (computing tasks), using networks to transfer workloads from one region to data centers in another.

[0003] While existing technologies and research consider the network impact on data center workload transmission, they rarely take into account the capacity of network channels and the uncertainty of latency caused by network nodes processing workloads. These uncertainties exist within the network and are all factors that data centers must consider when participating in demand response. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a data center demand response optimization method that considers network channel capacity and node uncertainty, thus solving the technical problem of data centers being affected by network uncertainty when participating in demand response.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a data center demand response optimization method considering network channel capacity and node uncertainty, the method comprising the following steps:

[0006] Set up a data center and The network between them has capacity constraints, limiting the amount of workload that can be transmitted at different times, and the construction of data centers... Migrate workloads to the data center Network transmission path planning model;

[0007] Build to reduce data center The first data center energy consumption model for load, and the data center Energy consumption model for a second data center that incorporates new energy sources;

[0008] A multi-data center power cost optimization model is established based on the network transmission path planning model, the first data center energy consumption model, and the second data center energy consumption model.

[0009] Data center operators during time periods With minimizing energy costs as the optimization objective, a multi-data center electricity cost optimization model is solved under the first constraint. This yields the data center operator's energy cost during the time period when the data center participates in demand response and is not affected by network uncertainties. Optimal energy cost ;

[0010] Based on optimal energy cost Build maximum energy cost An uncertain optimization model is proposed, and the uncertain optimization strategy is obtained by solving the uncertain optimization model.

[0011] By employing the above technical solution, the present invention provides a data center demand response optimization method that considers network channel capacity and node uncertainty, which has at least the following beneficial effects:

[0012] 1. This invention can avoid the impact of many uncertainties in the network on the participation of data centers in demand response. It can effectively model the uncertainty of network capacity and node transmission workload, analyze the most extreme cases, and formulate corresponding data center demand response strategies to ensure the stable operation and cost control of data centers.

[0013] 2. This invention can reduce network latency and processing delay, enabling data centers to provide a faster and smoother service experience, meeting users' demands for high-performance, low-latency services. Furthermore, it can adapt more quickly to business changes and market demands by dynamically adjusting resource allocation and service models to meet the needs of different scenarios.

[0014] 3. This invention can more accurately predict and address potential bottlenecks and failure points, thereby taking proactive measures to avoid service interruptions or performance degradation. It enables data centers to allocate and schedule resources more rationally, avoiding excessive idleness or waste, thus improving resource utilization efficiency and reducing energy costs. Attached Figure Description

[0015] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0016] Figure 1 This is a flowchart of the data center demand response optimization method of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This will allow for a full understanding of how the present application uses technical means to solve technical problems and achieve technical effects, and to facilitate its implementation.

[0018] In this embodiment, the data center Migrate workloads to data center n to reduce data center costs Electricity load participates in peak shaving to improve data center performance. The electricity load participates in valley filling. Setting up a data center. and The network between them has capacity constraints, limiting the amount of workload that can be transferred at different times. (Data Center) and The set of network nodes between This indicates that the network link is... .

[0019] Please refer to Figure 1 This embodiment proposes a data center demand response optimization method considering network channel capacity and node uncertainty. By modeling the uncertainty of network capacity and node transmission workload, and under the condition of known maximum energy cost of the data center, it analyzes the most extreme network channel capacity and network node processing latency, and formulates a data center demand response strategy. The method includes the following steps:

[0020] S1, Configure Data Center and The network between them has capacity constraints, limiting the amount of workload that can be transmitted at different times, and the construction of data centers... Migrate workloads to the data center Network transmission path planning model;

[0021] The network transmission path planning model includes network link constraints, network link capacity constraints, and network node uncertainty constraints, among which:

[0022] Network link constraints are:

[0023] (1)

[0024] In the above formula, , They represent data centers respectively. and The set of network nodes and network links between them; An integer variable representing a network node. During the period Transmitted to network node The workload; Represents network nodes To network node The link, similarly Represents network nodes To network node The link; It is an integer variable representing the data center. During the period The amount of workload transferred out; An integer variable, representing the time period. Migrate to data center The workload; This is a function used to analyze the bandwidth capacity required for workload transmission in the network;

[0025] function and Satisfy the following formula:

[0026] (2)

[0027] In the above formula, , Data Center and The set of sets The data center within the organization offloads workloads to reduce the load, and aggregates... The data centers within the facility are shifting their workloads to absorb new energy sources; Indicates data center During the period The workload that has been moved out; Indicates data center During the period Handle workloads transferred from other data centers.

[0028] The network link capacity constraint is:

[0029] The workload of network transmission Due to network link capacity limitations, the following constraints must be met:

[0030] (3)

[0031] In the above formula, For uncertain variables, representing network nodes To network node The link during the time period The upper limit of transmission capacity; The value is uncertain, and it is defined as follows:

[0032]

[0033] In the above formula, A constant represents a network node. To network node The link during the time period Theoretical transmission capacity; Let be a variable representing a network node. To network node The link during the time period Actual transmission capacity vs. theoretical transmission capacity The error;

[0034] The network node uncertainty constraint is:

[0035] The workload of network transmission It will also be subject to the latency of network node information processing, and must meet the following constraints:

[0036] (4)

[0037] In the above formula, Indicates time period Does the workload pass through network nodes during transmission? The value is 1 if the condition is met, and 0 otherwise. The parameter represents a maximum value; it is used to determine whether a node has been passed. Then, the latency of the transmission workload can be estimated.

[0038] Time delay This represents an uncertain variable, which is a network node. The latency caused by transmitting workloads, i.e.:

[0039] (5)

[0040] In the above formula, A constant represents a network node. During the period The theoretical delay generated by the transmission workload; among which It is a constant; Let be a variable representing a network node. During the period Theoretical delay of transmitting workload With actual delay The error.

[0041] This embodiment enables data centers to provide a faster and smoother service experience by reducing network latency and processing delays, meeting users' demands for high-performance, low-latency services. Furthermore, it allows for faster adaptation to business changes and market demands, dynamically adjusting resource allocation and service models to meet the needs of different scenarios.

[0042] S2, Build to reduce data center The first data center energy consumption model for load, and the data center The second data center energy consumption model for absorbing new energy sources; the first data center energy consumption model includes data centers. First energy consumption model and data center The first workload constraint and the expression for the first energy consumption model are:

[0043] The energy consumption calculation formula for a data center is as follows:

[0044] (6)

[0045] In the above formula, Indicates data center During the period Energy consumption; The parameter represents the ratio of total data center energy consumption to server energy consumption; Let be a variable representing the data center. During the period The planned number of servers to be used; The parameter represents the data center. Server idle power; The parameter represents the data center. The server's peak power; The parameter represents the data center before participating in demand response. During the period The workload to be processed; Indicates data center During the period The workload that has been moved out; The parameter represents the data center. The rate at which the server processes workloads; Indicates a time interval.

[0046] The first workload constraint is:

[0047] Data Center During the period The offloaded workload It should not exceed the workload it was originally planned to handle. It satisfies the following constraints:

[0048] (7)

[0049] To ensure data center The activated servers are able to complete the workload processing within the specified time, and the data center workload and the planned number of servers meet the following constraints:

[0050] (8)

[0051] In the above formula, The parameter represents the data center. The maximum processing latency that the workload can tolerate, and the servers ultimately planned to be used, must meet the following constraints:

[0052] (9)

[0053] In the above formula, The parameter represents the data center. The total number of servers.

[0054] The second data center energy consumption model includes data centers The second energy consumption model and data centers The second workload constraint; the expression for the second energy consumption model is:

[0055] (10)

[0056] In the above formula, Indicates data center During the period Energy consumption; Let be a variable representing the data center. During the period The planned number of servers to be used; The parameter represents the data center. Server idle power; The parameter represents the data center. The server's peak power; The parameter represents the data center before participating in demand response. During the period The workload to be processed; Indicates data center During the period Handling workloads transferred from other data centers; Indicates data center The rate at which the server processes workloads; Indicates a time interval.

[0057] The second workload constraint is:

[0058] To ensure data center The activated servers are able to complete the workload processing within the specified time, and the data center workload and the planned number of servers meet the following constraints:

[0059] (11)

[0060] (12)

[0061] In the above formula, The parameter represents the data center. Handling workloads The maximum tolerable processing latency; Let be a variable, representing the data center before participating in demand response. Handling workloads Number of servers required; Let be a variable representing the data center's participation in demand response. Handling incoming workloads Number of servers required; A constant, representing the data center Handling incoming workloads Time limit; For uncertain variables, representing network nodes The latency caused by transmitting workload;

[0062] (13)

[0063] Data Center Number of servers to be started The following constraints must be met:

[0064] (14)

[0065] (15)

[0066] In the above formula, Indicates data center The total number of servers.

[0067] This embodiment can more accurately predict and respond to potential bottlenecks and failure points, thereby taking proactive measures to avoid service interruptions or performance degradation. It enables data centers to allocate and schedule resources more rationally, avoiding excessive idleness or waste, thus improving resource utilization efficiency and reducing energy costs.

[0068] S3. Based on the network transmission path planning model, the energy consumption model of the first data center, and the energy consumption model of the second data center, a multi-data center power cost optimization model is established. The objective function for optimizing the total energy cost of data center operators is:

[0069] (16)

[0070] (17)

[0071] In the above formula, Let be a variable, representing the time period of the data center. Energy costs; The parameter represents the data center. During the period Electricity price; The parameter represents the data center. During the period Subsidies for units participating in the response to renewable energy consumption; The parameter represents the data center. During the period Electricity price; The parameter represents the data center. During the period Subsidies for units participating in peak shaving response; Indicates data center During the period Energy consumption; Indicates data center During the period The original energy consumption before participating in demand response; Indicates data center During the period Energy consumption; Indicates data center During the period The original energy consumption before participating in demand response; , Data Center and A set; Indicates data center During the period The workload that has been moved out; Indicates data center During the period Handle workloads transferred from other data centers.

[0072] This embodiment helps reduce energy consumption and carbon emissions in data centers, aligning with the principles of green and sustainable development. By optimizing resource utilization and reducing waste, data centers contribute to environmental protection while providing efficient services.

[0073] S4, based on the time period of the data center operator With minimizing energy costs as the optimization objective, a multi-data center electricity cost optimization model is solved under the first constraint. This yields the data center operator's energy cost during the time period when the data center participates in demand response and is not affected by network uncertainties. Optimal energy cost Specifically, this means: Determine the theoretical time delay. With actual delay error Actual transmission capacity vs. theoretical transmission capacity error Taking the total energy cost of the data center operator as the optimization objective in equation (16), the goal is to minimize the energy consumption of the data center during the specified time period. The energy cost is given by equations (1) to (15) and (17) as the first constraint. The solution is to find the optimal total energy cost for the data center operator under the condition of uncertainty, which is the energy cost of the data center during the time period. Optimal energy cost .

[0074] S5, Based on optimal energy cost Build maximum energy cost An uncertainty optimization model is used to solve for the uncertainty optimization model and obtain the uncertainty optimization strategy. In this embodiment, the uncertainty optimization model consists of a network transmission path planning model, a first data center energy consumption model, a second data center energy consumption model, and a multi-data center power cost optimization model. The uncertainty optimization strategy includes: , , , , , , , ,in:

[0075] Let be a variable, representing the data center before participating in demand response. Handling workloads Number of servers required; Let be a variable representing the data center's participation in demand response. Handling incoming workloads Number of servers required; Indicates data center During the period Energy consumption; Indicates data center During the period Energy consumption; Indicates data center During the period The workload that has been moved out; Indicates data center During the period Handling workloads transferred from other data centers; An integer variable representing a network node. During the period Transmitted to network node The workload; Indicates time period Does the workload pass through network nodes during transmission? The value is 1 if the condition is met, and 0 otherwise. Let be a variable representing the data center. During the period The planned number of servers to be used; in step S5, the specific process includes the following steps:

[0076] S51. Initialize theoretical delay With actual delay error Actual transmission capacity vs. theoretical transmission capacity error As a variable;

[0077] S52. Construct the uncertainty optimization objective function for the uncertainty optimization model, with the following expression:

[0078] (18)

[0079] In the above formula, Let be a variable, representing the total utility under uncertainty;

[0080] S53. Cost constraints for constructing an uncertain optimization model, namely:

[0081] (19)

[0082] In the above formula, It is a constant, set by the data center operator, to limit the maximum energy cost that the data center can accept; The constant represents the time period of the data center. The maximum acceptable energy cost;

[0083] S54. Taking the uncertainty optimization objective function as the optimization objective, the uncertainty optimization model is solved under the second constraint to obtain the uncertainty optimization strategy; in this implementation, the second constraint is Equation (1) to (17) and Equation (19).

[0084] This embodiment models the uncertainties of network capacity and node transmission workload. Under the condition of known maximum energy cost of data center, it analyzes the most extreme network channel capacity and network node processing latency, and formulates data center demand response strategies. This avoids the impact of many uncertainties in the network on the data center's participation in demand response. It can effectively model the uncertainties of network capacity and node transmission workload, analyze the most extreme cases, and formulate corresponding data center demand response strategies to ensure the stable operation and cost control of data center.

[0085] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Since the above embodiments are substantially similar to the method embodiments, their descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0087] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A data center demand response optimization method considering network channel capacity and node uncertainty, characterized in that, The method includes the following steps: Set up a data center and The network between them has capacity constraints, limiting the amount of workload that can be transmitted at different times, and the construction of data centers... Migrate workloads to the data center Network transmission path planning model; Build to reduce data center The first data center energy consumption model for load, and the data center Energy consumption model for a second data center that incorporates new energy sources; A multi-data center power cost optimization model is established based on the network transmission path planning model, the first data center energy consumption model, and the second data center energy consumption model. Data center operators during time periods With minimizing energy costs as the optimization objective, a multi-data center electricity cost optimization model is solved under the first constraint. This yields the data center operator's energy cost during the time period when the data center participates in demand response and is not affected by network uncertainties. Optimal energy cost ; Based on optimal energy cost Build maximum energy cost An uncertain optimization model is established, and the uncertain optimization strategy is obtained by solving the uncertain optimization model. The specific process is as follows: Initialize theoretical delay With actual delay error Actual transmission capacity vs. theoretical transmission capacity error As a variable; The objective function for constructing the uncertainty optimization model is expressed as: In the above formula, Let be a variable, representing the total utility under uncertainty; The cost constraint for constructing an uncertain optimization model is as follows: In the above formula, It is a constant, set by the data center operator, to limit the maximum energy cost that the data center can accept; The constant represents the time period of the data center. The maximum acceptable energy cost; Using the uncertainty optimization objective function as the optimization objective, the uncertainty optimization model is solved under the second constraint to obtain the uncertainty optimization strategy.

2. The data center demand response optimization method according to claim 1, characterized in that, The network transmission path planning model includes network link constraints, network link capacity constraints, and network node uncertainty constraints, wherein: Network link constraints are: （1） In the above formula, , They represent data centers respectively. and The set of network nodes and network links between them; An integer variable representing a network node. During the period Transmitted to network node The workload; Represents network nodes To network node The link, similarly Represents network nodes To network node The link; It is an integer variable representing the data center. During the period The amount of workload transferred out; An integer variable, representing the time period. Migrate to data center The workload; This is a function used to analyze the bandwidth capacity required for workload transmission in the network; function and Satisfy the following formula: （2） In the above formula, , Data Center and A set; Indicates data center During the period The workload that has been moved out; Indicates data center During the period Handling workloads transferred from other data centers; The network link capacity constraint is: The workload of network transmission Due to network link capacity limitations, the following constraints must be met: （3） In the above formula, For uncertain variables, representing network nodes To network node The link during the time period The upper limit of transmission capacity; The value is uncertain, and it is defined as follows: In the above formula, A constant represents a network node. To network node The link during the time period Theoretical transmission capacity; Let be a variable representing a network node. To network node The link during the time period Actual transmission capacity vs. theoretical transmission capacity The error; The network node uncertainty constraint is: The workload of network transmission It will also be subject to the latency of network node information processing, and must meet the following constraints: （4） In the above formula, Indicates time period Does the workload pass through network nodes during transmission? The value is 1 if the condition is met, and 0 otherwise. The parameter represents a maximum value; it is used to determine whether a node has been passed. Then, the latency of the transmission workload can be estimated; Time delay This represents an uncertain variable, which is a network node. The latency caused by transmitting workloads, i.e.: （5） In the above formula, A constant represents a network node. During the period The theoretical delay generated by the transmission workload; among which It is a constant; Let be a variable representing a network node. During the period Theoretical delay of transmitting workload With actual delay The error.

3. The data center demand response optimization method according to claim 1, characterized in that, The first data center energy consumption model includes data centers First energy consumption model and data center The first workload constraint and the expression for the first energy consumption model are: The energy consumption calculation formula for a data center is as follows: （6） In the above formula, Indicates data center During the period Energy consumption; The parameter represents the ratio of total data center energy consumption to server energy consumption; Let be a variable representing the data center. During the period The planned number of servers to be used; The parameter represents the data center. Server idle power; The parameter represents the data center. The server's peak power; The parameter represents the data center before participating in demand response. During the period The workload to be processed; Indicates data center During the period The workload that has been moved out; The parameter represents the data center. The rate at which the server processes workloads; Indicates a time interval; The first workload constraint is: Data Center During the period The offloaded workload It should not exceed the workload it was originally planned to handle. It satisfies the following constraints: （7） To ensure data center The activated servers are able to complete the workload processing within the specified time, and the data center workload and the planned number of servers meet the following constraints: （8） In the above formula, The parameter represents the data center. The maximum processing latency that the workload can tolerate, and the servers ultimately planned to be used, must meet the following constraints: （9） In the above formula, The parameter represents the data center. The total number of servers.

4. The data center demand response optimization method according to claim 1, characterized in that, The second data center energy consumption model includes data centers The second energy consumption model and data centers The second workload constraint; the expression for the second energy consumption model is: （10） In the above formula, Indicates data center During the period Energy consumption; Let be a variable representing the data center. During the period The planned number of servers to be used; The parameter represents the data center. Server idle power; The parameter represents the data center. The server's peak power; The parameter represents the data center before participating in demand response. During the period The workload to be processed; Indicates data center During the period Handling workloads transferred from other data centers; Indicates data center The rate at which the server processes workloads; Indicates a time interval; The second workload constraint is: To ensure data center The activated servers are able to complete the workload processing within the specified time, and the data center workload and the planned number of servers meet the following constraints: （11） （12） In the above formula, The parameter represents the data center. Handling workloads The maximum tolerable processing latency; Let be a variable, representing the data center before participating in demand response. Handling workloads Number of servers required; Let be a variable representing the data center's participation in demand response. Handling incoming workloads Number of servers required; A constant, representing the data center Handling incoming workloads Time limit; For uncertain variables, representing network nodes The latency caused by transmitting workload; （13） Data Center Number of servers to be started The following constraints must be met: （14） （15） In the above formula, Indicates data center The total number of servers.

5. The data center demand response optimization method according to claim 1, characterized in that, The expression for the multi-data center power cost optimization model is as follows: （16） （17） In the above formula, Let be a variable, representing the time period of the data center. Energy costs; The parameter represents the data center. During the period Electricity price; The parameter represents the data center. During the period Subsidies for units participating in the response to renewable energy consumption; The parameter represents the data center. During the period Electricity price; The parameter represents the data center. During the period Subsidies for units participating in peak shaving response; Indicates data center During the period Energy consumption; Indicates data center During the period The original energy consumption before participating in demand response; Indicates data center During the period Energy consumption; Indicates data center During the period The original energy consumption before participating in demand response; , Data Center and A set; Indicates data center During the period The workload that has been moved out; Indicates data center During the period Handle workloads transferred from other data centers.

6. The data center demand response optimization method according to claim 1, characterized in that, The uncertainty optimization model consists of a network transmission path planning model, a first data center energy consumption model, a second data center energy consumption model, and a multi-data center electricity cost optimization model.

7. The data center demand response optimization method according to claim 1, characterized in that, The uncertainty optimization strategy includes: , , , , , , , ,in: Let be a variable, representing the data center before participating in demand response. Handling workloads Number of servers required; Let be a variable representing the data center's participation in demand response. Handling incoming workloads Number of servers required; Indicates data center During the period Energy consumption; Indicates data center During the period Energy consumption; Indicates data center During the period The workload that has been moved out; Indicates data center During the period Handling workloads transferred from other data centers; An integer variable representing a network node. During the period Transmitted to network node The workload; Indicates time period Does the workload pass through network nodes during transmission? The value is 1 if the condition is met, and 0 otherwise. Let be a variable representing the data center. During the period The number of servers to be used.

8. A computer-readable storage medium, characterized in that, It stores computer-executable instructions that, when executed by a processor, implement the data center demand response optimization method according to any one of claims 1 to 7.

9. A computer device, characterized in that, It includes a processor and a memory, the memory being used to store a computer program that, when executed by the processor, implements the data center demand response optimization method as described in any one of claims 1 to 7.

Images