Adaptive edge computing resource optimization method
By implementing adaptive resource optimization methods in edge computing and using data collection and real-time monitoring to automatically calculate and adjust the amount of resources, the problem of insufficient resource utilization and untimely allocation in the existing technology is solved, and efficient resource management and load are achieved. adapt.
Patent Information
- Application Number
- CN202411429255.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-14
- Publication Date
- 2025-01-24
AI Technical Summary
The prior art cannot accurately determine whether edge computing resources are overloaded for a long time, and cannot calculate the expanded and reduced resources in detail, resulting in insufficient resource utilization and untimely allocation.
Through data collection, defining resource optimization scope, increasing or reduction of computing resources, closure of resource and distribution resources and real -time monitoring and feedback to achieve adaptive optimization of resources. Specific steps include using tools such as Loadrunner to collect the performance data of marginal servers, define the threshold of CPU, memory, and disk usage, automatically calculate the resource volume according to work load prediction and real -time requirements, and dynamically adjust resources through the virtual machine server.
The automatic adjustment of the edge computing resources is realized, the resource utilization rate is improved, the changing workload is adapted, and the energy consumption and delay are reduced.
Abstract
Description
Technical Field
[0001] The present invention relates to the field of edge computing data processing technology, and to an adaptive edge computing resource optimization method. By adaptively optimizing edge computing resources, the method can automatically increase and decrease edge computing resources. Background Art
[0002] An important problem in edge computing resource allocation is that resources cannot be automatically allocated on demand, resource requirements do not match existing resources, resource allocation is not timely, and the required resource calculation is not detailed. The present invention can solve the above problems in response to this situation. Most current methods cannot accurately determine whether resources are overloaded for a long time, and cannot calculate in detail the amount of expanded and reduced resources, resulting in insufficient resource utilization. Problems such as untimely allocation. This patent is an adaptive edge computing resource optimization method and system, which aims to improve the efficiency of edge computing nodes through intelligent resource allocation, adapt to variable workloads, and reduce energy consumption and delays. Summary of the invention
[0003] This patent is mainly completed through closed-loop actions such as data collection, definition of resource optimization range, calculation of resource increase or decrease, allocation of resources, real-time monitoring and feedback. The main invention point is the calculation of resource increase or decrease and the entire process. This patent is an adaptive edge computing resource optimization method and system, which aims to improve the efficiency of edge computing nodes through intelligent resource allocation, adapt to variable workloads, and reduce energy consumption and latency.
[0004] An adaptive edge computing resource optimization method comprises the following steps: (1) Data collection The LoadRunner server performance test tool is used to collect the configuration and usage of the edge server's CPU usage, memory usage, and disk usage. Samples are drawn according to the requirements, and the interval time is customized for analysis. The sample size is defined as n, n>=10, and the interval between two tests is defined as s, which can be customized. The real number set A for monitoring load values is A, and all its elements are a[i], where i is the element subscript.
[0005] (2) Define the scope of resource optimization 1. Define the thresholds of CPU usage, memory usage, and disk usage as max=80% and min=20%; 2. Define the number of times exceeding max as overstepMax. If overstepMax accounts for more than 50% of the total number of detections, it is determined that resources are scarce and the mechanism for increasing resources needs to be activated; If the number of times exceeding the threshold in the sample volume is set to account for more than 50%, that is, if the CPU, memory, and disk usage on the edge device is detected 10 times, and the usage rate exceeds 80% for 5 or more times, it is determined that the system resources are overloaded and resources need to be automatically increased; if the number of times below the threshold is set to account for more than 50%, that is, if the CPU, memory, and disk usage on the edge device is detected 10 times, and the usage rate is less than 20% for 5 times, it is determined that the system resources are not fully utilized and some resources need to be automatically released; 3. Define the number of times below the minimum as underMin. If underMin accounts for more than 50% of the total number of detections, it is considered a waste of resources and the resource reduction mechanism needs to be activated; (3) Calculate the amount of resources that are reduced or increased Based on workload forecasts and real-time demand, the system automatically calculates the amount of resources that need to be expanded or reduced; 1. The average value of the usage / occupancy rate that exceeds the threshold max within the number of detections minus the threshold max is the usage / occupancy rate of the resource that exceeds the max. The original resource amount is multiplied by the usage / occupancy rate of the resource that exceeds the max to indicate the amount of resources that exceed the max, that is, the amount of resources that need to be expanded. This is defined as add and rounded up. This definition indicates that resources need to be expanded by the amount of add to ensure that the device performance is always kept below the resource threshold max and there will be no resource shortage. 2. The average value of the usage / occupancy rate below the threshold min within the number of detections minus the absolute value of the threshold min indicates the usage / occupancy rate of the resource below min. Multiply the original resource amount original by the usage / occupancy rate of the resource below min to indicate the amount of resources below min, that is, the amount of resources that need to be reduced, which is defined as reduce and rounded up. This definition indicates that the amount of resources needs to be reduced to ensure that the device performance always remains above the resource threshold min, and there will be no waste of resources; (4) Allocate resources After receiving the instruction to increase or decrease resources, the system automatically calculates the amount of resources that need to be increased or decreased, and then adds a virtual machine server to the edge computing node based on the amount of resources; (5) Real-time monitoring and feedback The system monitors resource utilization and task execution performance in real time, and continuously optimizes resource management strategies based on feedback to adapt to environmental changes and changes in demand.
[0006] Beneficial effects of the present invention: By monitoring the CPU, memory, and disk usage on edge devices in real time, we can analyze resource usage trends within a certain range, define resource usage thresholds, and use machine learning algorithms to predict future workload trends.
[0007] When the resource usage within the monitoring range exceeds the threshold multiple times, it is determined that resources need to be increased. After receiving the resource increase information, the increase amount is calculated, and then edge computing resources are increased as needed; When resource usage is lower than the threshold for multiple times, it is determined that resources need to be reduced. After receiving the resource reduction information, the reduction amount is calculated, and then the edge computing resources are reduced as needed. DETAILED DESCRIPTION
[0008] The present invention is further described as follows in conjunction with the embodiments: An adaptive edge computing resource optimization method, the specific steps are as follows: (1) Data collection Use some mature tools such as LoadRunner and other server performance testing tools to collect the configuration and usage of edge server CPU usage, memory usage, and disk usage. Draw samples according to the requirements and customize the interval time for analysis. The sample size is defined as n, n>=10, and the interval between two detections is defined as s, which can be customized. The real number set A for monitoring load values, all of which are a[i], where i is the element subscript.
[0009] (2) Define the scope of resource optimization If the number of times exceeding the threshold within the sample volume is set to account for more than 50%, that is, if the CPU, memory, and disk usage on the edge device is detected 10 times, and the usage rate exceeds 80% for 5 or more times, it is judged that the system resources are overloaded and resources need to be automatically increased; if the number of times below the threshold is set to account for more than 50%, that is, if the CPU, memory, and disk usage on the edge device is detected 10 times, and the usage rate is less than 20% for 5 times, it is judged that the system resources are not fully utilized and some resources need to be automatically released.
[0010] The existing technology mostly uses the moving average method to predict the value of the next period SMA = (a[1]+ a[2]+ ... + a[n]) / n, where a represents the monitored data value and n represents the number of recent items, i.e. the sample size (which can be n months, n days, n hours, etc.). The disadvantage of the moving average method is that when the data difference between the monitored values is large, the average value will be lowered, which cannot reflect the actual load situation well, and the resource usage will be too high at a certain moment, causing the edge server node to crash.
[0011] 1. Define the thresholds of CPU usage, memory usage, and disk usage as max=80% and min=20%.
[0012] 2. Define the number of times exceeding max as overstepMax. If overstepMax accounts for more than 50% of the total number of detections, it is determined that resources are scarce and the mechanism for increasing resources needs to be activated.
[0013] The expression is: (1)overstepMax =∑1, if a[i]>= 80%; (2)overstepMax / n*100>= 50%; Example: define n=10, s is per hour, then the CPU usage set A={50,60,80,89,78,95,78,67,88,90} According to the existing moving average method, the value of the next period is predicted to be SMA = (a[1]+ a[2]+ ... + a[n]) / n=(50+60+80+89+78+95+78+67+88+90) / 10=75.5. This means that there is no resource load on the edge server node, but the actual situation is that resources are in short supply more than half of the time.
[0014] According to the algorithm of the present invention, overstepMax=∑1, if a[i]>= 80%=5 times; 5 / 10*100>= 50%; it means that within the 10 monitoring ranges, there are 5 times when the CPU usage rate exceeds 80%, indicating that there is a risk of resource shortage, and it is determined to be resource shortage, and it is necessary to start the mechanism of increasing resources.
[0015] 3. Define the number of times below the min as underMin. If underMin accounts for more than 50% of the total detection times, it is judged as a waste of resources and the resource reduction mechanism needs to be activated.
[0016] The expression is: (1) underMin=∑1, if a[i]<=20%; (2) underMin / n*100>=50%; (3) Calculate the amount of resources that are reduced or increased Based on workload forecasts and real-time demands, the system automatically calculates the amount of resources that need to be expanded or reduced, ensuring that the system will neither experience resource shortages nor waste.
[0017] 1. The average value of the usage / occupancy rate that exceeds the threshold max within the number of detections minus the threshold max represents the usage / occupancy rate of the resource that exceeds the max. Multiply the original resource amount by the usage / occupancy rate that exceeds the max to represent the amount of resources that exceed the max, that is, the amount of resources that need to be expanded, which is defined as add and rounded up. This definition means that the amount of resources that need to be expanded by add can ensure that the device performance always remains below the resource threshold max and there will be no resource shortage.
[0018] Expression: add = ceil(original*(AVG(∑a[i], if a[i]>80%)-80%)) 2. The average value of the usage / occupancy rate below the threshold min within the number of detections minus the absolute value of the threshold min indicates the usage / occupancy rate of the resource below min. Multiply the original resource amount original by the usage / occupancy rate of the resource below min to indicate the amount of resources below min, that is, the amount of resources that need to be reduced, which is defined as reduce and rounded up. This definition indicates that the amount of resources needs to be reduced to ensure that the device performance always remains above the resource threshold min, and there is no waste of resources.
[0019] Expression: reduce = ceil(original*(|AVG(∑a[i], if a[i]<20%)-20%|)) (4) Allocate resources After receiving instructions to increase or decrease resources, the system automatically calculates the amount of resources that need to be increased or decreased, and then adds virtual machine servers to the edge computing nodes based on the amount of resources.
[0020] (5) Real-time monitoring and feedback The system monitors resource utilization and task execution performance in real time, and continuously optimizes resource management strategies based on feedback to adapt to environmental changes and changes in demand.
[0021] Glossary Edge computing: refers to the use of an open platform that integrates network, computing, storage, and application core capabilities to provide the nearest service on the side close to the source of objects or data. Its application is initiated on the edge side, resulting in faster network service response, meeting the basic needs of the industry in terms of real-time business, application intelligence, security and privacy protection. Edge computing is between physical entities and industrial connections, or on top of physical entities. Cloud computing can still access historical data of edge computing.
Claims
1. An adaptive edge computing resource optimization method, characterized in that The steps include: (1) Data collection Use the LoadRunner server performance test tool to collect the configuration and usage of the edge server's CPU usage, memory usage, and disk usage. Draw samples according to the requirements and analyze them at a custom interval. The sample size is defined as n, where n>=10, and the interval between two tests is defined as s, which can be customized. The real number set A for obtaining the monitored load value is a[i], where i is the element subscript. (2) Define the scope of resource optimization 1. Define the thresholds of CPU usage, memory usage, and disk usage as max=80% and min=20%; Define the number of times exceeding max as overstepMax. If overstepMax accounts for more than 50% of the total number of detections, it is determined that resources are scarce and the mechanism for increasing resources needs to be activated. If the number of times exceeding the threshold in the sample volume is set to account for more than 50%, that is, if the CPU, memory, and disk usage on the edge device is detected 10 times, and the usage rate exceeds 80% for 5 or more times, it is determined that the system resources are overloaded and resources need to be automatically increased; if the number of times below the threshold is set to account for more than 50%, that is, if the CPU, memory, and disk usage on the edge device is detected 10 times, and the usage rate is less than 20% for 5 times, it is determined that the system resources are not fully utilized and some resources need to be automatically released; Define the number of times below the minimum as underMin. If underMin accounts for more than 50% of the total number of detections, it is considered a waste of resources and the resource reduction mechanism needs to be activated; (3) Calculate the amount of resources that are reduced or increased Based on workload forecasts and real-time demand, the system automatically calculates the amount of resources that need to be expanded or reduced; 1. The average value of the usage / occupancy rate that exceeds the threshold max within the number of detections minus the threshold max is the usage / occupancy rate of the resource that exceeds the max. The original resource amount is multiplied by the usage / occupancy rate of the resource that exceeds the max to indicate the amount of resources that exceed the max, that is, the amount of resources that need to be expanded. This is defined as add and rounded up. This definition indicates that resources need to be expanded by the amount of add to ensure that the device performance is always kept below the resource threshold max and there will be no resource shortage.
2. The average value of the usage rate / occupancy rate below the threshold min within the number of detections minus the absolute value of the threshold min indicates the usage rate / occupancy rate of the resource below min. The original resource amount is multiplied by the usage rate / occupancy rate of the resource below min to indicate the amount of resources below min, that is, the amount of resources that need to be reduced, which is defined as reduce, and rounded up. This definition indicates that the amount of resources needs to be reduced to ensure that the device performance always remains above the resource threshold min, and there will be no resource waste; (4) Allocate resources After receiving the instruction to increase or decrease resources, the system automatically calculates the amount of resources that need to be increased or decreased, and then adds a virtual machine server to the edge computing node based on the amount of resources; (5) Real-time monitoring and feedback The system monitors resource utilization and task execution performance in real time, and continuously optimizes resource management strategies based on feedback to adapt to environmental changes and changes in demand.