A big data driven dynamic resource scheduling method for IPv6 edge computing nodes

Abstract

The present invention discloses a big data-driven dynamic resource scheduling method for IPv6 edge computing nodes, which relates to the field of edge computing technology. The specific steps of the scheduling method are: S100, resource data collection and integration: deploying monitoring equipment at the IPv6 edge computing node to collect data on computing, network, storage and energy resources. The present invention realizes comprehensive monitoring and quantitative integration of computing, network, storage and energy resource data through multi-dimensional resource data collection and storage. This comprehensive data collection method makes resource scheduling more accurate and efficient. By constructing an energy consumption performance model, the present invention can accurately calculate the performance value and use big data to train and optimize the model, thereby achieving a balance between energy consumption and performance in the resource scheduling process, which not only improves resource utilization efficiency, but also reduces energy consumption costs, providing strong support for the green and sustainable development of edge computing nodes.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to a big data-driven dynamic resource scheduling method for IPv6 edge computing nodes. Background Art

[0002] With the rapid development of information technology, edge computing, as a new computing model, is gradually changing the way data is processed. By deploying computing, storage, and application core resources at the edge of the network, edge computing enables local processing and analysis of data, greatly improving data processing efficiency and response speed. Especially in the IPv6 network environment, the number of edge computing nodes and the amount of data have shown explosive growth, which has put higher requirements on resource scheduling and management. Therefore, how to efficiently schedule and manage the resources of IPv6 edge computing nodes has become a technical problem that needs to be solved urgently.

[0003] Traditional resource scheduling methods mainly rely on static configuration and manual management, and cannot be dynamically adjusted according to real-time load and resource conditions. This method is not only inefficient, but also prone to resource waste and overload. In addition, traditional resource scheduling methods often ignore the balance between energy consumption and performance, resulting in excessively high energy costs, which is not conducive to the green and sustainable development of edge computing nodes. Therefore, there is an urgent need for a method that can adapt to the IPv6 network environment and can dynamically schedule and manage edge computing node resources in real time to improve resource utilization efficiency, reduce energy costs, and enhance system stability and reliability.

[0004] Therefore, developing a big data-driven dynamic resource scheduling method for IPv6 edge computing nodes will promote the rapid development of IPv6 edge computing technology in the fields of cloud computing and the Internet of Things, and bring a more stable and efficient service experience to related industries. Summary of the Invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide a big data-driven dynamic resource scheduling method for IPv6 edge computing nodes. By comprehensively monitoring and quantitatively integrating computing, network, storage and energy resource data, an energy consumption performance model is constructed to achieve a balance between energy consumption and performance in the resource scheduling process. At the same time, the method can monitor and predict business loads in real time, dynamically adjust resource allocation, and effectively avoid resource idleness and overload. In addition, the introduction of multi-dimensional resource collaborative scheduling decisions and real-time migration and dynamic resource reorganization mechanisms further improves the overall performance and flexibility of the system.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a big data driven dynamic resource scheduling method for IPv6 edge computing nodes,

[0007] S100, Resource Data Collection and Integration: Deploy monitoring equipment at IPv6 edge computing nodes to collect data on computing, network, storage, and energy resources. The data is transmitted to the data center via the IPv6 network for preprocessing and standardization, and then integrated into comprehensive quantitative values ​​for storage in the resource information database.

[0008] S200, energy consumption performance model construction: extract relevant data from the resource information database, construct an energy consumption performance model to calculate performance values, and use the extracted data to train and optimize the model;

[0009] S300, Load Monitoring, Forecasting and Warning: This collects real-time load data using a distributed message queue, applies a service load prediction algorithm to predict future service loads based on the real-time collected service load values, and compares the future service loads with warning thresholds.

[0010] S400, Multi-resource Cooperative Scheduling: When a new computing task arrives, based on the business load monitoring and prediction results, a set of candidate nodes is screened from both load and resource conditions. The performance and energy consumption values ​​of each candidate node in the set are combined, and a multi-dimensional resource cooperative scheduling decision formula is used to calculate the optimal node for task allocation. The optimal node is the node with the highest comprehensive score calculated by the multi-dimensional resource cooperative scheduling decision formula.

[0011] S500, real-time migration and resource reorganization: During the operation of the node, the business load and energy consumption are monitored in real time to determine whether to trigger the real-time migration and dynamic resource reorganization mechanism. When the real-time migration and dynamic resource reorganization mechanism is triggered, the optimal node is recalculated, and some or all tasks on the current node are migrated to the optimal node. The computing, storage, and network resources are reallocated according to the new load and system status.

[0012] Furthermore, in the S100, resource data collection and integration, the collection of various resources:

[0013] For computing resources: Use system monitoring software to obtain real-time data on CPU usage, core temperature, and memory usage;

[0014] For network resources: Use a network traffic monitor to collect data on network bandwidth usage, data packet transmission rate, and network latency;

[0015] For storage resources: Use the storage system's built-in management interface to obtain data on disk read and write rates and storage capacity usage;

[0016] For energy resources: install smart meters and energy consumption sensors to monitor node power consumption and energy efficiency indicators in real time.

[0017] Furthermore, in the above S100, resource data collection and integration, the data is integrated into a comprehensive quantitative value through a multi-dimensional resource quantitative integration algorithm, and the formula is: ,in, Represents the comprehensive quantitative value of multi-dimensional resources, is the number of resource types, It is The weight coefficient of the resource, For the Actual quantitative indicators of this resource.

[0018] Furthermore, in the energy consumption performance model construction in S200, the energy consumption performance model is constructed using the formula: ,in, Represents performance indicators, is the energy consumption value, 、 、 is the constant coefficient obtained by fitting large data. is the number of impact factors associated with multidimensional resources, It is The coefficient of the impact factor, No. An instantiation indicator of the comprehensive quantitative value of multi-dimensional resources.

[0019] Furthermore, in said S200, the energy consumption value in the energy consumption performance model in the energy consumption performance model construction The calculation formula is: ,in Indicates the energy consumption value of the edge computing node, 、 、 They are the energy consumption coefficients of computing resource usage, network traffic, and storage access, It is the resource utilization rate, the value range is 0-1, Network traffic refers to the amount of data transmitted per unit time. is the memory access frequency, is the energy consumption of the node when it is idle.

[0020] Furthermore, in the step of training the energy consumption performance model in the energy consumption performance model construction, the multi-dimensional resource comprehensive quantitative value, energy consumption value and performance index data of the resource information database are divided into a training set and a test set in proportion, and the parameters α, β, γ and , , initialize, use the training set to train the model, and adjust the parameters by using the mean square error as the loss function. The calculation formula is: ,in, is the number of samples in the training set, is the actual performance index value of the kth sample in the training set, It is the prediction performance index value calculated based on the current model parameters. The gradient descent method is used to update the parameters. For the parameter α, the update formula is: , where η is the learning rate, which controls the step size of parameter updates, is a parameter The updated value, is a parameter The values ​​before update, for β, γ and Similarly, the trained model is updated and evaluated using the test set. The model is optimized based on the evaluation results and then applied in practice.

[0021] Furthermore, in the load monitoring, prediction and early warning step S300, a service load prediction algorithm is used to predict the future service load based on the service load value collected in real time. The formula is: ,in, The next moment The business load forecast value, For the current moment The actual monitoring value of the business load, is the load trend continuation coefficient, is the coefficient of resource change on load, is the number of resource dimensions that have a significant impact on the load among multi-dimensional resources. It is The weight coefficient of the impact of resource dimension changes on load, It is The change in the comprehensive quantitative value of a multi-dimensional resource.

[0022] Furthermore, in the load monitoring, forecasting and warning step S300, the future business load is compared with the warning threshold, and the warning upper threshold of the business load is set according to the historical business load data and business needs. and lower threshold , when the predicted business load value Greater than or equal to or less than or equal to the threshold When the business load is determined to be abnormal, the early warning mechanism is automatically triggered to notify the system administrator to adjust resource allocation or expand nodes.

[0023] Furthermore, in the S400, the optimal node is calculated using a multi-dimensional resource collaborative scheduling decision formula in the multi-resource collaborative scheduling, and the formula is: ,in represents the optimal scheduling decision result, is the set of all candidate nodes, Represents a candidate node in the set, and Node performance values ​​and energy consumption values, is a node The number of network topology hops or physical distance to the data storage location, 、 、 It is a weight coefficient set based on the business's emphasis on performance, energy consumption, and data transmission costs, and is preset through historical data analysis.

[0024] Furthermore, in the above S500, the real-time migration and dynamic resource reorganization judgment formula is used to judge whether to trigger the real-time migration and dynamic resource reorganization mechanism. The formula is: ,in, A flag that determines whether to trigger real-time migration and dynamic resource reorganization. is the business load of the current node, is the historical average business load of the node, is the load overload trigger threshold coefficient, is the energy consumption of the current node, is the safe upper limit of node energy consumption, It is the energy consumption too high trigger threshold coefficient. When the load is too high or the energy consumption is too high, , triggering the real-time migration and dynamic resource reorganization mechanism; otherwise .

[0025] Compared with the existing technology, this big data-driven dynamic resource scheduling method for IPv6 edge computing nodes has the following beneficial effects:

[0026] 1. The present invention realizes comprehensive monitoring and quantitative integration of computing, network, storage and energy resource data through multi-dimensional resource data collection and storage. This comprehensive data collection method makes resource scheduling more accurate and efficient. By constructing an energy consumption performance model, the present invention can accurately calculate the performance value and use big data to train and optimize the model, thereby achieving a balance between energy consumption and performance in the resource scheduling process. It not only improves resource utilization efficiency, but also reduces energy consumption costs, providing strong support for the green and sustainable development of edge computing nodes. In addition, the method can also monitor and predict business loads in real time, dynamically adjust resource allocation according to load conditions, effectively avoid the occurrence of resource idleness and overload, and further improve the stability and reliability of the system.

[0027] 2. The present invention introduces multi-dimensional resource collaborative scheduling decision-making and real-time migration and dynamic resource reorganization mechanism. The multi-dimensional resource collaborative scheduling decision formula comprehensively considers multiple factors such as performance value, energy consumption value and data storage location distance, making task allocation more reasonable and effectively improving the overall performance of the system. The real-time migration and dynamic resource reorganization mechanism can adjust resource distribution in real time according to business load and energy consumption to ensure that the system is always in the optimal state. This dynamic adjustment capability not only improves the flexibility and adaptability of the system, but also provides users with a more stable and efficient service experience. At the same time, the implementation of this method has also promoted the widespread application of IPv6 edge computing technology in cloud computing and the Internet of Things, and promoted the rapid development of related industries.

[0028] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0030] Figure 1 A flowchart of a big data driven dynamic resource scheduling method for IPv6 edge computing nodes;

[0031] Figure 2 A framework diagram of a big data-driven dynamic resource scheduling method for IPv6 edge computing nodes. DETAILED DESCRIPTION

[0032] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0033] Example 1:

[0034] Application in intelligent transportation systems.

[0035] In a busy urban transportation hub area, multiple IPv6 edge computing nodes are deployed to process the large amount of data generated by the intelligent transportation system, such as video streams from traffic cameras and vehicle sensor data.

[0036] Resource data collection and integration: Monitoring equipment is installed on each edge computing node. System monitoring software is used to obtain real-time CPU usage, core temperature, and memory usage to understand the usage status of computing resources. Network traffic monitors are used to collect network bandwidth usage, packet transmission rate, and network latency data to understand network resource status. The storage system's built-in management interface is used to obtain disk read and write rates and storage capacity usage data to evaluate storage resources. Smart meters and energy consumption sensors are installed to monitor node power consumption and energy efficiency indicators in real time. This data is transmitted to the data center via the IPv6 network. After preprocessing and standardization, it is integrated into a comprehensive quantitative value based on a multi-dimensional resource quantification integration algorithm. The formula is: ,in, Represents the comprehensive quantitative value of multi-dimensional resources, is the number of resource types, It is The weight coefficient of the resource, For the The actual quantitative indicators of the resources are stored in the resource information database.

[0037] Energy consumption performance model construction: Extract relevant data from the resource information database, build an energy consumption performance model to calculate the performance value, the formula is: ,in, Represents performance indicators, is the energy consumption value, and the formula is: ,in Indicates the energy consumption value of the edge computing node, 、 、 They are the energy consumption coefficients of computing resource usage, network traffic, and storage access, It is the resource utilization rate, the value range is 0-1, Network traffic refers to the amount of data transmitted per unit time. is the memory access frequency, is the energy consumption of the node when it is idle, 、 、 is the constant coefficient obtained by fitting large data. is the number of impact factors associated with multidimensional resources, It is The coefficient of the impact factor, For the The instantiation index of the multi-dimensional resource comprehensive quantitative value is used to train and optimize the model. The multi-dimensional resource comprehensive quantitative value, energy consumption value and performance index data are divided into training set and test set in proportion. The model parameters α, β, γ and , , use the training set to train the model, and adjust the parameters by using the mean square error as the loss function. The calculation formula is: ,in, is the number of samples in the training set, is the actual performance index value of the kth sample in the training set, It is the prediction performance index value calculated based on the current model parameters. The gradient descent method is used to update the parameters. For the parameter α, the update formula is: , where η is the learning rate, which controls the step size of parameter updates, is a parameter The updated value, is a parameter The values ​​before update, for β, γ and Similarly, the trained model is updated and evaluated using the test set. The model is optimized based on the evaluation results and applied in practice.

[0038] Load monitoring, prediction and early warning: Use distributed message queues to collect real-time load data, and use the business load prediction algorithm to predict future business load based on the real-time collected business load values. The formula is: ,in, The next moment The business load forecast value, For the current moment The actual monitoring value of the business load, is the load trend continuation coefficient, is the coefficient of resource change on load, is the number of resource dimensions that have a significant impact on the load among multi-dimensional resources. It is The weight coefficient of the impact of resource dimension changes on load, It is The change in the comprehensive quantitative value of multi-dimensional resources is used to set the upper threshold of business load warning based on historical business load data and business needs. and lower threshold , when the predicted business load value Greater than or equal to or less than or equal to the threshold When the business load is determined to be abnormal, the early warning mechanism is automatically triggered to notify the system administrator to adjust resource allocation or expand nodes.

[0039] Multi-resource collaborative scheduling: When a new traffic data analysis task arrives, based on the business load monitoring and prediction results, a set of candidate nodes is screened from both load and resource conditions. The performance value and energy consumption value of each candidate node in the set are combined, and the multi-dimensional resource collaborative scheduling decision formula is used to calculate the optimal node for task allocation. The formula is: ,in represents the optimal scheduling decision result, is the set of all candidate nodes, Represents a candidate node in the set, and Node performance values ​​and energy consumption values, is a node The number of network topology hops or physical distance to the data storage location, 、 、 The weight coefficient is set according to the business's emphasis on performance, energy consumption, and data transmission cost. For example, if the current traffic flow monitoring task has high real-time requirements, that is, more emphasis is placed on performance, the performance weight coefficient can be appropriately increased. , select a network with high performance value and high network topology hop count or physical distance The closer node executes the task.

[0040] Real-time migration and resource reorganization: During node operation, the service load and energy consumption are monitored in real time. The real-time migration and dynamic resource reorganization judgment formula is used to determine whether to trigger the real-time migration and dynamic resource reorganization mechanism. The formula is: ,in, A flag that determines whether to trigger real-time migration and dynamic resource reorganization. is the business load of the current node, is the historical average business load of the node, is the load overload trigger threshold coefficient, is the energy consumption of the current node, is the safe upper limit of node energy consumption, The energy consumption is too high to trigger the threshold coefficient. When the business load of a node (such as a large number of sudden traffic incidents causing a sharp increase in video analysis tasks) is greater than Multiply by the historical average business load, or energy consumption is greater than When multiplied by the energy consumption safety limit, , triggering this mechanism, recalculating the optimal node, migrating some or all tasks on the current node to the optimal node, and reallocating computing, storage, and network resources according to the new load and system status to ensure efficient and stable operation of the system.

[0041] To sum up, in the intelligent transportation system, this invention ensures the efficient operation of the system from many aspects. Resource data collection and integration comprehensively obtains various types of resource data to provide a basis for subsequent decision-making; the energy consumption performance model accurately analyzes the relationship between energy consumption and performance, helping to optimize resource utilization; load monitoring prediction and early warning can timely detect load anomalies to ensure system stability; multi-resource collaborative scheduling accurately allocates tasks based on multiple factors; real-time migration and resource reorganization dynamically adjust resources to adapt to business changes. These links work closely together to enable the intelligent transportation system to quickly process large amounts of data, effectively alleviate traffic congestion, improve traffic efficiency, and ensure traffic safety, demonstrating the significant advantages and application value of this invention in the field of intelligent transportation.

[0042] Example 2:

[0043] Application in industrial manufacturing production lines.

[0044] In a large automobile manufacturing plant, IPv6 edge computing nodes are widely deployed on its production lines. These nodes are responsible for processing production equipment operation data and quality inspection data, and play a key role in ensuring the efficient and stable operation of the production line. The specific implementation process of the invention in this scenario will be described in detail below.

[0045] Resource data collection and integration: Various monitoring devices are carefully deployed on each edge computing node to comprehensively and accurately obtain the resource usage status of the node.

[0046] Computing resource monitoring: System monitoring software can collect real-time information about CPU usage, core temperature, and memory usage. For example, when a production line is operating at full capacity, CPU usage may rise sharply, and core temperature will also increase accordingly. This data can help us understand the real-time usage status of computing resources and promptly identify potential performance bottlenecks.

[0047] Network resource monitoring: Using a network traffic monitor, you can collect data on network bandwidth usage, data packet transmission rates, and network latency. During the automotive manufacturing process, a large amount of production equipment data needs to be transmitted to edge computing nodes in real time for processing. Therefore, network stability and bandwidth adequacy are crucial. If the network bandwidth is insufficient or there is high latency, data transmission may be untimely, affecting production efficiency.

[0048] Storage resource monitoring: Through the storage system's built-in management interface, you can obtain disk read and write rates and storage capacity usage data. As production data continues to accumulate, storage resource usage needs to be closely monitored. If storage capacity is close to saturation, data cleanup or expansion may be required in a timely manner to ensure normal data storage and access.

[0049] Energy consumption resource monitoring: Installing smart meters and energy consumption sensors can monitor the power consumption and energy efficiency indicators of nodes in real time. In industrial production, energy cost is an important consideration. By monitoring energy consumption, the operation strategy of nodes can be optimized, energy consumption can be reduced, and energy utilization efficiency can be improved.

[0050] The collected data is transmitted to the data center via the IPv6 network. In the data center, the data is first preprocessed, including data cleaning, noise removal and outlier removal to ensure data accuracy and reliability. Then, the data is standardized to convert different types of data into a unified format and scale. Finally, the data is integrated into a comprehensive quantitative value based on a multi-dimensional resource quantitative integration algorithm. The formula is: , and stored in the resource information database. This comprehensive quantitative value can fully reflect the resource usage status of the node and provide an important basis for subsequent resource scheduling.

[0051] Energy consumption and performance model construction, extracts relevant data from the resource information database to build an energy consumption and performance model. The construction of this model aims to reveal the relationship between the energy consumption and performance of the node, so as to achieve a balance between energy consumption and performance in the resource scheduling process. The formula is: , the model is trained and optimized using the extracted data, and the multi-dimensional resource comprehensive quantitative value, energy consumption value and performance index data from the resource information database are divided into training set and test set in proportion, and the parameters α, β, γ and , , initialize, use the training set to train the model, and adjust the parameters by using the mean square error as the loss function. The calculation formula is: , the gradient descent method is used to update the parameters. For the parameter α, the update formula is: , for β, γ and Similarly, updates are performed and the trained model is evaluated using the test set. The model is optimized based on the evaluation results. If the model's prediction error is large, the model parameters can be adjusted or training data can be added to improve the model's accuracy and generalization ability. The optimized model will be applied to the actual resource scheduling process.

[0052] Load monitoring, prediction and early warning: With the help of distributed message queues, load data of edge computing nodes is collected in real time. Distributed message queues have the characteristics of high throughput, low latency and strong reliability, which can ensure the timely and accurate transmission of load data. The formula is: , predicts future business load based on the business load values ​​collected in real time. This algorithm comprehensively considers the changes in the current business load and the comprehensive quantitative values ​​of multi-dimensional resources, and can more accurately predict future business load trends. It sets the upper threshold of business load warning based on historical business load data and production needs. and lower threshold For example, if historical data shows that when the business load exceeds When the node performance will be significantly reduced, the warning upper threshold can be set to ; When the business load is lower than There may be a waste of resources when the warning threshold is set to , when the predicted business load value Greater than or equal to or less than or equal to the threshold When the business load is determined to be abnormal, the early warning mechanism is automatically triggered. The early warning information will be notified to the system administrator via SMS, email or system prompts, so that the administrator can take timely measures to adjust resource allocation or expand nodes to ensure the stable operation of the production line.

[0053] Multi-resource collaborative scheduling: When a new production data analysis task arrives, reasonable resource scheduling is required to ensure that the task can be executed efficiently and stably. Based on the business load monitoring and prediction results, a set of candidate nodes is screened from the two aspects of load and resource conditions. First, nodes with excessively high business loads or insufficient resources are excluded; then, nodes with good performance, low energy consumption, and stable network connections are selected as candidate nodes. Combining the performance value and energy consumption value of each candidate node in the candidate node set, the multi-dimensional resource collaborative scheduling decision formula is used to calculate the optimal node for task allocation. The formula is: For example, for production equipment failure prediction tasks with high real-time requirements, the performance weight coefficient can be appropriately increased. , select nodes with high performance values ​​and short network topology hops or physical distances to perform tasks; for tasks that are sensitive to energy consumption, the energy consumption weight coefficient can be appropriately increased , select nodes with lower energy consumption to execute tasks.

[0054] Real-time migration and resource reorganization: During node operation, it is necessary to monitor business load and energy consumption in real time to adjust and optimize resources in a timely manner. The real-time migration and dynamic resource reorganization judgment formula is used to determine whether to trigger the real-time migration and dynamic resource reorganization mechanism. The formula is: When the business load of a node (such as during the peak production period, a large number of production equipment generates data at the same time, causing the business load of the node to increase sharply) is greater than Multiply by the historical average business load, or energy consumption is greater than When multiplied by the energy consumption safety limit, , triggering this mechanism, once the real-time migration and dynamic resource reorganization mechanism is triggered, the system will recalculate the optimal node and migrate some or all tasks on the current node to the optimal node. At the same time, according to the new load and system status, it will reallocate computing, storage, and network resources. For example, if the computing resources of a node are insufficient, while the computing resources of another node are surplus, some computing tasks can be migrated to the node with more remaining computing resources; if the storage capacity of a node is close to saturation, some data can be migrated to the node with larger storage capacity. In this way, the efficient and stable operation of the system is ensured, and the production efficiency and quality of the production line are improved.

[0055] To sum up, in the scenario of industrial manufacturing production lines, this invention plays a key role. Resource data collection and integration realizes all-round monitoring of edge computing node resources on the production line; energy consumption performance model construction provides a quantitative basis for energy consumption management and performance optimization; load monitoring prediction and early warning can detect load anomalies in advance and prevent production interruptions; multi-resource collaborative scheduling reasonably allocates resources according to task requirements and improves production efficiency; real-time migration and resource reorganization ensures timely adjustment when equipment load or energy consumption is abnormal. This series of operations ensures stable operation of the production line, reduces production costs, improves product quality, and effectively promotes the intelligent upgrading of industrial manufacturing. It has broad application prospects in the field of industrial production.

[0056] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes, characterized in that: The specific steps of this scheduling method are: S100, Resource Data Collection and Integration: Deploy monitoring equipment at IPv6 edge computing nodes to collect data on computing, network, storage, and energy resources. The data is transmitted to the data center via the IPv6 network for preprocessing and standardization, and then integrated into comprehensive quantitative values ​​for storage in the resource information database. S200, energy consumption performance model construction: extract relevant data from the resource information database, construct an energy consumption performance model to calculate performance values, and use the extracted data to train and optimize the model; S300, Load Monitoring, Forecasting and Warning: This collects real-time load data using a distributed message queue, applies a service load prediction algorithm to predict future service loads based on the real-time collected service load values, and compares the future service loads with warning thresholds. S400, Multi-resource Cooperative Scheduling: When a new computing task arrives, based on the business load monitoring and prediction results, a set of candidate nodes is screened from both load and resource conditions. The performance and energy consumption values ​​of each candidate node in the set are combined, and a multi-dimensional resource cooperative scheduling decision formula is used to calculate the optimal node for task allocation. The optimal node is the node with the highest comprehensive score calculated by the multi-dimensional resource cooperative scheduling decision formula. S500, real-time migration and resource reorganization: During the operation of the node, the business load and energy consumption are monitored in real time to determine whether to trigger the real-time migration and dynamic resource reorganization mechanism. When the real-time migration and dynamic resource reorganization mechanism is triggered, the optimal node is recalculated, and some or all tasks on the current node are migrated to the optimal node. The computing, storage, and network resources are reallocated according to the new load and system status.

2. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 1, characterized in that: S100, the collection of various resources in resource data collection and integration: For computing resources: Use system monitoring software to obtain real-time data on CPU usage, core temperature, and memory usage; For network resources: Use a network traffic monitor to collect data on network bandwidth usage, data packet transmission rate, and network latency; For storage resources: Use the storage system's built-in management interface to obtain data on disk read and write rates and storage capacity usage; For energy resources: install smart meters and energy consumption sensors to monitor node power consumption and energy efficiency indicators in real time.

3. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 1, characterized in that: In the above S100, in resource data collection and integration, the data is integrated into a comprehensive quantitative value through a multi-dimensional resource quantitative integration algorithm, and the formula is: ,in, Represents the comprehensive quantitative value of multi-dimensional resources, is the number of resource types, It is The weight coefficient of the resource, For the Actual quantitative indicators of this resource.

4. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 3, characterized in that: The energy consumption performance model is constructed in the energy consumption performance model construction in S200, and the formula is: ,in, Represents performance indicators, is the energy consumption value, 、 、 is the constant coefficient obtained by fitting large data. is the number of impact factors associated with multidimensional resources, It is The coefficient of the impact factor, No. An instantiation indicator of the comprehensive quantitative value of multi-dimensional resources.

5. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 4, characterized in that: In the S200, the energy consumption value in the energy consumption performance model is constructed. The calculation formula is: ,in Indicates the energy consumption value of the edge computing node, 、 、 They are the energy consumption coefficients of computing resource usage, network traffic, and storage access, It is the resource utilization rate, the value range is 0-1, Network traffic refers to the amount of data transmitted per unit time. is the memory access frequency, is the energy consumption of the node when it is idle.

6. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 1, characterized in that: In the step of training the energy consumption performance model in the energy consumption performance model construction, the multi-dimensional resource comprehensive quantitative value, energy consumption value and performance index data of the resource information database are divided into a training set and a test set in proportion, and the parameters α, β, γ and , , initialize, use the training set to train the model, and adjust the parameters by using the mean square error as the loss function. The calculation formula is: ,in, is the number of samples in the training set, is the actual performance index value of the kth sample in the training set, It is the prediction performance index value calculated based on the current model parameters. The gradient descent method is used to update the parameters. For the parameter α, the update formula is: , where η is the learning rate, which controls the step size of parameter updates, is a parameter The updated value, is a parameter The values ​​before update, for β, γ and Similarly, the trained model is updated and evaluated using the test set. The model is optimized based on the evaluation results and then applied in practice.

7. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 1, characterized in that: In the load monitoring, prediction and early warning, the service load prediction algorithm is used to predict the future service load based on the service load value collected in real time. The formula is: ,in, The next moment The business load forecast value, For the current moment The actual monitoring value of the business load, is the load trend continuation coefficient, is the coefficient of resource change on load, is the number of resource dimensions that have a significant impact on the load among multi-dimensional resources. It is The weight coefficient of the impact of resource dimension changes on load, It is The change in the comprehensive quantitative value of a multi-dimensional resource.

8. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 7, characterized in that: S300, comparing the future business load with the warning threshold in load monitoring, prediction and warning, and setting the warning upper threshold of the business load based on historical business load data and business needs and lower threshold , when the predicted business load value Greater than or equal to or less than or equal to the threshold When the business load is determined to be abnormal, the early warning mechanism is automatically triggered to notify the system administrator to adjust resource allocation or expand nodes.

9. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 1, characterized in that: In the above S400, the optimal node is calculated by using a multi-dimensional resource collaborative scheduling decision formula in the multi-resource collaborative scheduling, and the formula is: ,in represents the optimal scheduling decision result, is the set of all candidate nodes, Represents a candidate node in the set, and Node performance values ​​and energy consumption values, is a node The number of network topology hops or physical distance to the data storage location, 、 、 It is a weight coefficient set based on the business's emphasis on performance, energy consumption, and data transmission costs, and is preset through historical data analysis.

10. A big data driven dynamic resource scheduling method for IPv6 edge computing nodes according to claim 1, characterized in that: In the above S500, the real-time migration and dynamic resource reorganization judgment formula is used to judge whether to trigger the real-time migration and dynamic resource reorganization mechanism. The formula is: ,in, A flag that determines whether to trigger real-time migration and dynamic resource reorganization. is the business load of the current node, is the historical average business load of the node, is the load overload trigger threshold coefficient, is the energy consumption of the current node, is the safe upper limit of node energy consumption, It is the energy consumption too high trigger threshold coefficient. When the load is too high or the energy consumption is too high, , triggering the real-time migration and dynamic resource reorganization mechanism; otherwise .

Images