Edge cloud environment-oriented resource awareness scheduling method, system and device, and medium

Abstract

The invention discloses a resource awareness scheduling method, system and device for an edge cloud environment and a medium. The method comprises the steps that in response to a task scheduling request, various load index values corresponding to alternative end nodes are collected based on preset load indexes, and the load indexes comprise the CPU occupancy rate, the memory occupancy rate, the node task occupancy rate and the temporary storage margin; standardizing the load index values, and generating objective weights corresponding to the load indexes based on the standardized load index values; and calculating the score of each alternative end node based on the objective weight corresponding to each load index, selecting an optimal scheduling node based on the score of each alternative end node, and executing task scheduling. Compared with the prior art, task resource scheduling can be more comprehensive and fine. And meanwhile, the task resource scheduling accuracy is improved, and resource use inclination is prevented.

Description

Technical Field

[0001] The present invention belongs to the field of network communication technology, and specifically relates to a resource-aware scheduling method, system, device and medium for an edge cloud environment. Background Art

[0002] With the rapid development of technologies such as 5G, Internet of Things (IoT), and Industrial Internet, edge computing, as an emerging computing paradigm, is gradually becoming an important way to solve distributed computing and data processing. However, the complexity and heterogeneity of the edge computing environment have brought a series of challenges to resource management and task scheduling. In traditional cloud computing environments, resource management is mainly concentrated in large data centers, while the large number of devices in edge computing environments, different computing capabilities, and limited network bandwidth make task scheduling extremely complex. Edge devices usually include smartphones, sensors, smart home devices, etc. They exist at the edge of the network in a decentralized manner with limited resources, energy, and computing power, which puts higher requirements on the real-time performance and energy efficiency of tasks.

[0003] The default scheduling algorithm of the existing edge cloud environment can only process two resource indicators, CPU and memory, in extreme cases, and cannot meet the needs of different applications for multiple resources, resulting in incomplete and intricate resource scheduling. At the same time, the static weight allocation method cannot adapt to the dynamic changes in resource utilization in the edge cloud environment, which can easily cause excessive resource usage.

[0004] Therefore, in response to the above technical problems, it is necessary to provide a new task scheduling strategy for application in edge cloud environments.

[0005] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgment or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a resource-aware scheduling method, system, device and medium for edge cloud environments, which can improve the flexibility and adaptability of task resource scheduling, prevent excessive resource tilt, and meet more needs.

[0007] In order to achieve the above object, a technical solution provided by a specific embodiment of the present invention is as follows:

[0008] In a first aspect, the present invention provides a resource-aware scheduling method for an edge cloud environment, comprising:

[0009] In response to the task scheduling request, based on the preset load indicators, various load indicator values ​​corresponding to the candidate end nodes are collected, and the load indicators include CPU occupancy rate, memory occupancy rate, node task ratio and temporary storage margin;

[0010] Standardizing the load index value, and generating objective weights corresponding to various load indexes based on the standardized load index value;

[0011] Based on the objective weights corresponding to the various load indicators, the scores of the candidate end nodes are calculated, and the optimal scheduling node is selected to perform task scheduling.

[0012] In one or more embodiments of the present invention, the method further comprises:

[0013] Obtain the demand load index value of the task to be scheduled;

[0014] If the load index value of the end node meets the required load index value, the end node is listed as a candidate end node;

[0015] If the load index value of the end node does not meet the required load index value, the end node is filtered so that the end node that does not meet the required load index value does not participate in the current task scheduling.

[0016] In one or more embodiments of the present invention, the load index value is standardized as follows:

[0017]

[0018] Among them, x nj is the load index value of the jth load index corresponding to the nth candidate end node, max(x j ) is the maximum value of the jth load index among the candidate end nodes, min(x j ) is the minimum value of the j-th load index among all candidate end nodes.

[0019] In one or more embodiments of the present invention, the formula for generating the objective weight corresponding to each load data based on the standardized load data is:

[0020]

[0021] Among them, x nj is the load index value of the jth load index corresponding to the nth candidate end node; is the average value of the jth load index of each candidate end node; N is the total number of candidate end nodes; x′ ni Indicates the normalized load index value corresponding to the ith index of the nth candidate end node; represents the standardized average value of the i-th indicator, p is the number of indicators that make up the load data, and W j is the weight value corresponding to the jth load indicator, r ij is the correlation coefficient between the i-th load index and the j-th load index.

[0022] In one or more embodiments of the present invention, the method further comprises:

[0023] A monitoring component is deployed on the edge node to monitor and collect the load indicator values ​​corresponding to the candidate end nodes.

[0024] In one or more embodiments of the present invention, the current score of the node is calculated based on the objective weights corresponding to the load data of the candidate end node, including:

[0025] The load index values ​​of the candidate end node are multiplied by the weight values ​​corresponding to the load index respectively, and the sum of the products of the load index values ​​and the weight values ​​corresponding to the load index is taken as the current score of the candidate end node.

[0026] In a second aspect, the present invention provides a resource-aware scheduling system for an edge cloud environment, comprising:

[0027] A collection module is used to collect various load indicator values ​​corresponding to the candidate end nodes in response to the task scheduling request based on the preset load indicators, wherein the load indicators include CPU occupancy rate, memory occupancy rate, node task ratio and temporary storage margin;

[0028] A processing module, used to standardize the load index value, and generate an objective weight corresponding to each load index based on the standardized load index value;

[0029] The scheduling module is used to calculate the score of each candidate end node based on the objective weight corresponding to each load indicator, and select the optimal scheduling node to perform task scheduling.

[0030] In one or more embodiments of the present invention, the system is further used for:

[0031] Obtain the demand load index value of the task to be scheduled;

[0032] If the load index value of the end node meets the required load index value, the end node is listed as a candidate end node;

[0033] If the load index value of the end node does not meet the required load index value, the end node is filtered so that the end node that does not meet the required load index value does not participate in the current task scheduling.

[0034] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the resource-aware scheduling method for an edge cloud environment by executing the computer instructions.

[0035] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the resource-aware scheduling method for an edge cloud environment.

[0036] Compared with the prior art, the resource-aware scheduling method for edge cloud environment provided by the present invention can cover a variety of computing indicators in edge cloud environment, and at the same time has a high degree of flexibility in the selection of computing indicators. Different indicators are set with different processing strategies based on their action mechanisms to meet the needs of different applications for a variety of resources, making task resource scheduling more comprehensive and refined. At the same time, the present invention adopts a strategy of real-time collection of resource utilization of node performance during application deployment, and generates the objective weight of the round of task scheduling in real time. The objective weight changes dynamically based on the different real-time data corresponding to the node indicator, and more accurately reflects the dynamic changes of resources in the edge cloud environment. It is suitable for the implementation of various cloud environments, and can improve the accuracy of task resource scheduling and prevent resource usage tilt. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 is a schematic diagram of an edge cloud environment system in one embodiment of the present invention;

[0039] Figure 2 It is a flow chart of a resource-aware scheduling method for an edge cloud environment in one embodiment of the present invention;

[0040] Figure 3 It is a structural block diagram of a resource-aware scheduling system for an edge cloud environment in one embodiment of the present invention;

[0041] Figure 4 It is a structural block diagram of an electronic device in one embodiment of the present invention. DETAILED DESCRIPTION

[0042] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0043] Unless explicitly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising”, etc., will be understood to include the stated elements or components but not to exclude other elements or components.

[0044] In one technical solution, a method for scheduling task resources at an edge device node specifically includes: obtaining the CPU occupancy rate and memory occupancy rate corresponding to the end node, and calculating the candidate node score using a fixed unified weight based on the default scheduling algorithm in the edge cloud environment. Task resource scheduling is then performed based on the score.

[0045] For example, the preferred algorithms in the default scheduler are roughly the following two:

[0046] Least Request Priority Score:

[0047]

[0048] Resource Balanced Allocation Score:

[0049]

[0050] Among them, Score is the score value of the node, S cpu is the total CPU time, N cpu is the CPU idle time, S mem is the total memory size, N mem The size of the memory used.

[0051] Based on the above formula, it can be seen that this scheme quantifies the node status through CPU usage and memory usage, and then allocates task resources. The setting of its fixed weight value is only to control the value range of the final node score value, and does not play a regulatory role in reflecting the actual resource situation of the current node. This static weight allocation method cannot adapt to the dynamic changes in resource utilization in the edge cloud environment, and it is easy to cause excessive tilt in resource utilization. At the same time, only two isolated indicators, CPU occupancy rate and memory occupancy rate, cannot accurately reflect the status of each node in the current edge cloud computing environment. The algorithm itself does not take into account the variability and conflict of resource indicators at all. Direct allocation cannot meet the needs of different applications for multiple resources, and resource scheduling is not comprehensive and detailed enough.

[0052] The inventor of the present invention discovered the main shortcomings of the above technology, and based on the shortcomings of the technology, proposed a new technical implementation idea: multi-dimensionalize the load indicators used to calculate the equal division in the node, and dynamically calculate the weights of the resource indicators through multi-dimensional indicators. Fully consider the variability and conflict of resource indicators, and can assign a corresponding objective weight to each resource indicator according to the actual operation of the cluster. At the same time, the weight will be recalculated in each scheduling, and then the node score will be calculated to select the best scheduling node, which can better adapt to the environment with complex requirements. Satisfy the problems of poor real-time performance and insufficient resource perception in actual implementation scenarios.

[0053] like Figure 1 The figure shows a schematic diagram of an edge cloud environment system. The structure of the present invention mainly relates to an edge cloud computing environment, and its core components include: edge nodes: computing entities distributed at the edge of the network, close to the data source. Responsible for collecting resource and status information of edge nodes, as well as scheduling decisions for tasks; end devices: computing entities distributed at the very edge of the network, generating data. Running some basic tasks; cloud data centers: providing additional computing resources to handle large-scale or non-real-time high-demand tasks. These components together constitute a multi-level computing environment, and the present invention focuses on the resource allocation and management between edge nodes and end devices, that is, the process of scheduling tasks from edge nodes to end nodes.

[0054] Please refer to Figure 2 FIG. 1 is a flow chart of resource-aware scheduling for edge cloud environments in one embodiment of the present invention. The resource-aware scheduling method for edge cloud environments specifically includes the following steps:

[0055] S201: In response to a task scheduling request, based on a preset load indicator, collect various load indicator values ​​corresponding to the candidate end node;

[0056] It should be noted that the load indicator described in the present invention refers to the factors that affect the end node's ability to process task resources when scheduling task resources for the end node, which may include but are not limited to CPU occupancy, memory occupancy, node task ratio, and temporary storage margin, etc. The embodiment of the present invention does not limit the selection of load indicators or the number of load indicators.

[0057] Among them, the CPU occupancy rate is a measure of the amount of time the CPU spends executing various tasks and processes on the computer. In the embodiment of the present invention, it is calculated as the ratio of the time required for the CPU to execute and process various tasks and program execution to the total CPU time. The memory occupancy rate value is the ratio of used memory to total memory. Used memory refers to the sum of physical memory currently allocated to the application, and total memory refers to the total amount of physical memory available to the system. Task ratio refers to the ratio of the number of tasks running on each node to the number of all tasks running in the cluster. This indicator is used to measure the workload of a single node in a cluster. By calculating the proportion of the number of tasks running on each node to the total number of tasks, the load of the node can be evaluated, thereby helping the scheduling algorithm to more reasonably allocate new tasks and avoid overloading some nodes while other node resources are idle. Temporary storage margin refers to the local storage of the host node for storing temporary data generated when the task is running, which directly affects the operating environment and performance of the task. If the temporary storage margin resources of a node are over-consumed, it may cause the node to be unable to run a new Pod or affect the performance of an existing Pod.

[0058] In the embodiment of the present invention, the manner of collecting the corresponding load data of the end node is not limited. In an exemplary embodiment, a monitoring component for monitoring and collecting the load index values ​​corresponding to the candidate end nodes can be deployed at the edge node.

[0059] For example, the load index values ​​corresponding to the candidate end nodes can be collected by deploying Metrics components on the edge nodes, and then monitoring the data collected by the Proxy agent in the end nodes, and then monitoring the resource usage of the end nodes in the cluster. The Metrics provides a powerful toolkit for measuring the behavior of key components in the production environment, and can monitor and count some behaviors in the service. Proxy is an important network proxy component. Proxy can route requests to the appropriate node according to the decision of the scheduler, so a Proxy agent is deployed on each end node. For another example, prometheus can also be used to monitor load indicators such as end node memory utilization and CPU utilization.

[0060] It should also be noted that for a specific task to be scheduled, not all end nodes can meet the task requirements of the task to be scheduled. For example, the remaining memory of the end node corresponding to the task to be scheduled is 1G, and the remaining memory of node A is only 500M. If end node A is not processed at this time, end node A will still participate in the data collection and score calculation process, resulting in a waste of computing resources; if end node A is elected, it may even cause the task to be unable to be processed, data loss and other adverse situations. Therefore, end nodes that do not meet the task requirements should be filtered before the election of the scheduling node, so that they do not participate in the election of the task scheduling node, to ensure that only nodes that meet the task resource requirements will be considered for task scheduling.

[0061] In a specific embodiment of the present invention, the resource-aware scheduling for edge cloud environments provided by the present invention also includes: obtaining a demand load index value of the task to be scheduled; if the end node load index value meets the demand load index value, listing the end node as an alternative end node; if the end node load index value does not meet the demand load index value, filtering the end node so that the end node that does not meet the demand load index value does not participate in this task scheduling.

[0062] For example, for a task P that needs to be scheduled j , whose demand vector is D j , then for this group of nodes N, node n i The local resource vector is F j , when D is satisfied j <F j , then the node n i Add to the sequence that can be scheduled. Here, less than means for D i Every element in is less than or equal to the vector F j The corresponding elements in .

[0063] S202: Standardizing the load index value, and generating an objective weight corresponding to each load index based on the standardized load index value;

[0064] It should be noted that, since different indicators have different effects on the node's ability to process task resources, the embodiment of the present invention adopts a classification and standardization strategy for different load indicators. Specifically, the load indicators used to score the end nodes are divided into positive indicators and negative indicators. Among them, the larger the index value of the positive indicator, the higher the corresponding end node score value, and the more conducive it is for the node to process related task resources; the larger the index value of the negative indicator, the lower the corresponding end node score value, and the more unfavorable it is for the node to process related task resources.

[0065] In an exemplary embodiment of the present invention, the formula for normalizing the load data is:

[0066]

[0067] Among them, x nj is the load index value of the jth load index corresponding to the nth candidate end node, max(x j ) is the maximum value of the jth load index among the candidate end nodes, min(x j ) is the minimum value of the j-th load index among all candidate end nodes.

[0068] Based on the above formula, after the user analyzes and selects the load index based on the actual implementation scenario, the index is first classified based on its nature. The load index values ​​of different types on the candidate end nodes are further standardized in a corresponding manner. This processing method can keep the original characteristics of the load data to the greatest extent, and the standardized results are the most accurate.

[0069] The present invention also provides another set of implementation methods. After the load indicators are selected and classified, the reverse indicators in the selected load indicators can be converted into forward indicators based on a preset indicator forward conversion method, such as the inverse reciprocal conversion method, and then all indicators can be brought into the same set of standardized formulas. The purpose of data standardization required by the present invention can also be achieved. It is understandable that the forward indicators can also be uniformly converted into reverse indicators, and the present invention will not repeat this implementation method.

[0070] Furthermore, the formula for generating the objective weights corresponding to each load data based on the standardized load index value is:

[0071]

[0072] Among them, x nj is the load index value of the jth load index corresponding to the nth candidate end node; is the average value of the jth load index of each candidate end node; N is the total number of candidate end nodes; x′ ni Indicates the normalized load index value corresponding to the ith index of the nth candidate end node; represents the standardized average value of the i-th indicator, p is the number of indicators that make up the load data, and W j is the weight value corresponding to the jth load indicator, r ij is the correlation coefficient between the i-th load index and the j-th load index.

[0073] Standard Deviation jIt can measure the degree of dispersion of the load index j in each end node. The larger the standard deviation, the greater the difference in the values ​​of the load index corresponding to each end node, the more information it can reflect, the stronger the evaluation strength of the load index itself, and the more weight should be assigned to the load index.

[0074] r ij It represents the correlation coefficient between load indicators i and j. The correlation coefficient is used to represent the correlation between load indicators. The stronger the correlation with other load indicators, the less conflict this load indicator has with other load indicators. The more the same information is reflected, the more repetitive the evaluation content that can be reflected is. This weakens the evaluation strength of this load indicator to a certain extent, and the weight assigned to this indicator should be reduced.

[0075] Finally, by combining the correlation coefficient of the load index of each end node and the standard deviation of each load index value, the weight value corresponding to each load index in this round of task scheduling can be calculated. Since the size of the weight value depends on the load index value collected in real time at the beginning of this round of task scheduling, the real-time performance of the weight value is better. It can better reflect the dynamic performance of each end node in the current edge environment.

[0076] For example, after pre-screening in the cluster, there are three candidate end nodes left, namely node 1, node 2 and node 3. Based on the corresponding implementation environment, the user selects four load indicators. The standard deviation corresponding to each indicator can be calculated through the above formula: S1, S2, S3 and S4; after data collection and standardization of the collected data at each node, a 3×4 matrix is ​​formed. The rows of the matrix represent the different indicator collection amounts of the same node, and the columns of the matrix represent the indicator values ​​corresponding to different nodes for the same indicator:

[0077]

[0078] Further calculation based on the correlation coefficient formula can obtain the correlation coefficient matrix:

[0079]

[0080] It can be understood that when i is equal to j, it indicates that the two indicators involved in the calculation of the correlation coefficient are in fact the same indicator. Therefore, for the case where i is equal to j in the correlation coefficient, there is no need to calculate and the item can be directly set to 1.

[0081] Furthermore, the weight value corresponding to each indicator can be calculated based on the parameters obtained from the above extremes. Take the calculation of the weight W1 corresponding to the indicator x1 as an example:

[0082]

[0083] S203: Based on the objective weights corresponding to the various load indicators, the scores of the candidate end nodes are calculated, and based on the scores of the candidate end nodes, the optimal scheduling node is selected to perform task scheduling.

[0084] In an exemplary embodiment of the present invention, the current score of the node is calculated based on the objective weights corresponding to the load data of the candidate node, including: multiplying the load index values ​​of the candidate end node by the weight values ​​corresponding to the load index, and taking the sum of the products of the load index values ​​and the weight values ​​corresponding to the load index as the current score of the candidate end node.

[0085] The corresponding scoring formula can be expressed as:

[0086]

[0087] Among them, Score n represents the score value of the nth node in this round of scheduling, W i represents the calculated weight value of the corresponding indicator i, x ni Represents the load index value of the nth node for load index i.

[0088] For example, following the above embodiment, after pre-screening in the cluster, three candidate end nodes remain, namely node 1, node 2 and node 3. The user selects four load indicators based on the corresponding implementation environment. After calculation, it can be known that the weight values ​​corresponding to each load indicator are W1, W2, W3 and W4 respectively; in the data collection stage, the load indicator values ​​of node 1 for the four indicators are x 11 、x 12 、x 13 and x 14 . Taking node 1 as an example, its corresponding score should be:

[0089]

[0090] It should be noted that in order to ensure the real-time consideration of the resource utilization of the end node performance indicators during the application deployment process in the embodiment of the present invention, to avoid uneven cluster load and affect system performance, it is necessary to start a new round of load indicator value collection and load indicator weight value calculation before each round of task scheduling, so as to obtain the optimal scheduling node for the current task to be forwarded.

[0091] In order to further verify the significant effect of the resource-aware scheduling method for edge cloud environment provided by the present invention, the following is a set of control experiments:

[0092] This experiment sets up two clusters, one cluster uses the original scheduler; the other cluster uses the resource-aware scheduling method for edge cloud environment provided by the present invention. The clusters are deployed on Alibaba Cloud servers, the basic system environment is centos7, and for each cluster size, excluding the master node, the number of slave nodes is 4.

[0093] In order to achieve periodic task creation, choose to create a task based on time interval repetitive scheduling through CronJob (periodic task), record the scheduling process time, scheduler space occupancy and scheduling results. In order to measure whether the cluster's resource utilization is balanced, the experimental evaluation index is displayed by the variance of the utilization rate of a certain resource in each node of the cluster. The calculation process is as follows:

[0094]

[0095] U ij represents the utilization of resource j on node i, n represents the total number of slave nodes, represents the average utilization of resource j.

[0096]

[0097] S j It represents the variance of the utilization of resource j in the cluster, which can reflect whether the utilization of resource j is balanced.

[0098] At the same time, based on the overall resource situation, the average value of the variance is defined to reflect the overall resource imbalance of the cluster:

[0099]

[0100] S avg The mean of the variance of the utilization of all resources in the cluster, where p represents the number of indicators. This metric comprehensively considers the fluctuation of each resource in the cluster and provides a global resource balance assessment. The calculated resource imbalance value reflects the overall status of the entire cluster in terms of resource utilization.

[0101] Through experimental verification in a real edge cloud environment, the scheduling strategy of the present invention improves the standard deviation of the comprehensive load by 57% compared with the default scheduling algorithm when the cluster load is large, showing a better resource scheduling effect.

[0102] Please refer to Figure 3 As shown, based on the same inventive concept as the aforementioned resource-aware scheduling method for edge cloud environments, an embodiment of the present invention provides a resource-aware scheduling system 300 for edge cloud environments, which includes: a collection module 301, a processing module 302 and a scheduling module 303.

[0103] Specifically, the collection module 301 is used to respond to the task scheduling request and, based on the preset load indicators, collect the load indicator values ​​corresponding to the candidate end nodes, wherein the load indicators include CPU occupancy, memory occupancy, node task ratio and temporary storage margin; the processing module 302 is used to standardize the load indicator values ​​and generate objective weights corresponding to the load indicators based on the standardized load indicator values; the scheduling module 303 is used to calculate the scores of the candidate end nodes based on the objective weights corresponding to the load indicators, and select the optimal scheduling node based on the scores of the candidate end nodes to perform task scheduling.

[0104] Please refer to Figure 4 As shown, an embodiment of the present invention further provides an electronic device 400, which includes at least one processor 401, a memory 402 (for example, a non-volatile memory), a memory 403, and a communication interface 404, and the at least one processor 401, the memory 402, the memory 403, and the communication interface 404 are connected together via a bus 405. At least one processor 401 is used to call at least one program instruction stored or encoded in the memory 402, so that the at least one processor 401 performs various operations and functions of the resource-aware scheduling method for the edge cloud environment described in various embodiments of this specification.

[0105] In an embodiment of the present specification, the electronic device 400 may include, but is not limited to, a personal computer, a server computer, a workstation, a desktop computer, a laptop computer, a notebook computer, a mobile electronic device, a smart phone, a tablet computer, a cellular phone, a personal digital assistant (PDA), a handheld device, a messaging device, a wearable electronic device, a consumer electronic device, and the like.

[0106] An embodiment of the present invention also provides a computer-readable medium, which carries computer execution instructions. When the computer execution instructions are executed by a processor, they can be used to implement various operations and functions of the resource-aware scheduling method for an edge cloud environment described in the various embodiments of this specification.

[0107] The computer-readable medium in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0108] In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0109] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may 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.

[0110] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatuses, systems, and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0111] The foregoing description of specific exemplary embodiments of the present invention is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present invention to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present invention and various different selections and changes. The scope of the present invention is intended to be limited by the claims and their equivalents.

[0112] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0113] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A resource-aware scheduling method for edge cloud environment, characterized in that: include: In response to the task scheduling request, based on the preset load indicators, various load indicator values ​​corresponding to the candidate end nodes are collected, and the load indicators include CPU occupancy rate, memory occupancy rate, node task ratio and temporary storage margin; Standardizing the load index value, and generating objective weights corresponding to various load indexes based on the standardized load index value; Based on the objective weights corresponding to the various load indicators, the scores of the candidate end nodes are calculated, and the optimal scheduling node is selected to perform task scheduling.

2. The resource-aware scheduling method for edge cloud environment according to claim 1, characterized in that: The method further comprises: Obtain the demand load index value of the task to be scheduled; If the load index value of the end node meets the required load index value, the end node is listed as a candidate end node; If the load index value of the end node does not meet the required load index value, the end node is filtered so that the end node that does not meet the required load index value does not participate in the current task scheduling.

3. The resource-aware scheduling method for edge cloud environment according to claim 1, characterized in that: The formula for normalizing the load index value is: Among them, x nj is the load index value of the jth load index corresponding to the nth candidate end node, max(x j ) is the maximum value of the jth load index among the candidate end nodes, min(x j ) is the minimum value of the j-th load index among all candidate end nodes.

4. The resource-aware scheduling method for edge cloud environment according to claim 1, characterized in that: The formula for generating the objective weights of each load data based on the standardized load data is: Among them, x nj is the load index value of the jth load index corresponding to the nth candidate end node; is the average value of the jth load index of each candidate end node; N is the total number of candidate end nodes; x′ ni Indicates the normalized load index value corresponding to the ith index of the nth candidate end node; represents the standardized average value of the i-th indicator, p is the number of indicators that make up the load data, and W j is the weight value corresponding to the jth load indicator, r ij is the correlation coefficient between the i-th load index and the j-th load index.

5. The resource-aware scheduling method for edge cloud environment according to claim 1, characterized in that: The method further comprises: A monitoring component is deployed on the edge node to monitor and collect the load indicator values ​​corresponding to the candidate end nodes.

6. The resource-aware scheduling method for edge cloud environment according to claim 1, characterized in that: Based on the objective weights corresponding to the load data of the candidate end nodes, the current score of the node is calculated, including: The load index values ​​of the candidate end node are multiplied by the weight values ​​corresponding to the load index respectively, and the sum of the products of the load index values ​​and the weight values ​​corresponding to the load index is taken as the current score of the candidate end node.

7. A resource-aware scheduling system for edge cloud environments, characterized in that: include: A collection module is used to collect various load indicator values ​​corresponding to the candidate end nodes in response to the task scheduling request based on the preset load indicators, wherein the load indicators include CPU occupancy rate, memory occupancy rate, node task ratio and temporary storage margin; A processing module, used to standardize the load index value, and generate an objective weight corresponding to each load index based on the standardized load index value; The scheduling module is used to calculate the score of each candidate end node based on the objective weight corresponding to each load indicator, and select the optimal scheduling node to perform task scheduling.

8. The resource-aware scheduling system for edge cloud environments according to claim 7, characterized in that: The system is also used to: Obtain the demand load index value of the task to be scheduled; If the load index value of the end node meets the required load index value, the end node is listed as a candidate end node; If the load index value of the end node does not meet the required load index value, the end node is filtered so that the end node that does not meet the required load index value does not participate in the current task scheduling.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the resource-aware scheduling method for an edge cloud environment according to any one of claims 1 to 6 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the resource-aware scheduling method for an edge cloud environment according to any one of claims 1 to 6.

Images