Edge device resource scheduling method and system

Abstract

The application discloses a kind of edge device resource scheduling methods, including: obtaining edge device resource;The obtained edge device resource is divided into computing unit and management unit;According to the resource quantity of computing unit, establish computing resource pool;Algorithm service is split into operator, then operator is prepositioned into computing unit and is arranged as algorithm chain template;After matching, the algorithm task to be calculated and algorithm chain template are calculated;Through management unit, the computing resource use of each operator on algorithm chain template is monitored in real time, and according to use, resource in computing resource pool is scheduled, the utilization of edge device can be greatly improved by the application.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of edge computing, and particularly relates to an edge-end device resource scheduling method and system. BACKGROUND

[0002] Current artificial intelligence applications developed by power grid companies are mainly cloud-based applications, but computing power is widely distributed on the edge side, resulting in waste of computing resources such as computing power and network. In the future, with the large-scale deployment of edge-end devices, the demand for edge deployment and application of models will surge, and the realization of the connection of models between the cloud, edge and end has become an urgent need in the current stage. At the same time, the resource specifications of many device edges are limited, and the algorithm processing capability is weak, so it is difficult for the container orchestration system based on the cloud computing center to be parallelly migrated to the edge scene. Therefore, in the edge-end scene, the difference between the business demand for resources and the selection of microservices and the capability of the edge device need to be considered, but there is currently no good method to solve such problems. SUMMARY

[0003] In order to solve the problems existing in the prior art, the application provides an edge-end device resource scheduling method, which can improve the utilization rate of edge-end devices.

[0004] The technical problem to be solved by the application is solved by the following technical scheme:

[0005] In a first aspect, an edge-end device resource scheduling method is provided, comprising:

[0006] obtaining edge-end device resources;

[0007] dividing the obtained edge-end device resources into a computing unit and a management unit;

[0008] establishing a computing resource pool according to the number of resources of the computing unit;

[0009] splitting an algorithm service into an operator, and then prepositioning the operator into the computing unit to arrange an algorithm chain template;

[0010] calculating after matching the algorithm task to be calculated and the algorithm chain template;

[0011] real-time monitoring the computing resource usage of each operator on the algorithm chain template by the management unit, and scheduling the resources in the computing resource pool according to the usage.

[0012] In combination with the first aspect, further, the obtaining of the edge-end device resources comprises:

[0013] establishing a communication connection between edge-end devices;

[0014] deploying a cluster component on the edge-end devices to establish an edge-end device cluster;

[0015] Collect the number of CPU, GPU, memory, and storage resources of all edge devices used for computing based on the cluster.

[0016] In combination with the first aspect, further, deploying the cluster component on the edge device includes:

[0017] Deploy k3s-agent on edge devices divided into computing units, and deploy k3s-server on edge devices divided into management units.

[0018] In combination with the first aspect, further, establishing a computing resource pool according to the number of resources of the computing units includes:

[0019] The parameter standards of the edge device resources included in the computing unit are unified and statistically summarized to generate a computing resource pool, and the resources of the same edge device are marked as inseparable.

[0020] In combination with the first aspect, further, a configuration domain for adjusting operator logic and computing efficiency is provided on the operator.

[0021] In combination with the first aspect, further, when the operators are arranged into algorithm chain templates and pre-installed in the computing units, the locations of the operators and the resources required during operation are marked.

[0022] In combination with the first aspect, further, the real-time monitoring of computing resource usage of each operator on the algorithm chain template by the management unit and scheduling of resources in the computing resource pool according to the usage includes:

[0023] The management unit monitors in real time the computing resource usage of each operator in the algorithm chain template corresponding to the algorithm task to be calculated. If an instance of an algorithm chain template is completed, the computing resources of the operator corresponding to the algorithm chain template are released to the computing resource pool.

[0024] In a second aspect, a system for scheduling edge device resources is provided, including:

[0025] Resource acquisition module, used to obtain edge device resources;

[0026] A unit division module is used to divide the acquired edge device resources into computing units and management units;

[0027] A resource pool establishment module is used to establish a computing resource pool according to the number of resources of the computing unit;

[0028] The computing module is used to split the algorithm service into operators, and then pre-install the operators into computing units to compose them into algorithm chain templates;

[0029] Match the algorithm task to be calculated with the algorithm chain template and then calculate;

[0030] The resource scheduling module is used to monitor the computing resource usage of each operator on the algorithm chain template in real time through the management unit, and schedule the resources in the computing resource pool according to the usage.

[0031] In conjunction with the second aspect, further, the operations performed by the resource acquisition module include:

[0032] Establish communication connections between edge devices;

[0033] Deploy cluster components on edge devices to establish an edge device cluster;

[0034] Collect the number of CPU, GPU, memory, and storage resources of all edge devices used for computing based on the cluster.

[0035] In conjunction with the second aspect, further, the operations performed by the resource scheduling module include:

[0036] The management unit monitors in real time the computing resource usage of each operator in the algorithm chain template corresponding to the algorithm task to be calculated. If an instance of an algorithm chain template is completed, the computing resources of the operator corresponding to the algorithm chain template are released to the computing resource pool.

[0037] Beneficial effects of the present invention: The present invention coordinates the management of edge device resources by building an edge cluster network, greatly improves computing efficiency by splitting the algorithm into operators and presetting them into computing units for efficient business orchestration, while also improving the utilization rate of edge devices, reducing the imbalance in the use of edge device resources, waste of idle resources, and other phenomena, and effectively improving the computing power of edge devices in all aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the present invention;

[0039] Figure 2 This is a schematic diagram of the functional unit division in the present invention;

[0040] Figure 3 Schematic diagram of the operator, algorithm chain and business flow in the present invention;

[0041] Figure 4 This is a schematic diagram of the cluster network in the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0043] In order to better understand the present invention, the relevant technologies in the technical solution of the present invention are described below.

[0044] Example 1

[0045] like Figure 1 As shown, the present invention provides a method for scheduling edge device resources, comprising the following steps:

[0046] Step 1: Obtain edge device resources

[0047] To obtain edge device resources, we must first establish a communication connection between edge devices;

[0048] Then install cluster components on the edge devices to establish an edge device cluster;

[0049] Collect the resource quantity of all edge devices used for computing based on the cluster, including the quantity of CPU, GPU, memory, and storage resources.

[0050] Step 2: Functional unit division

[0051] Next, divide the edge devices into management units and computing units according to their functional positioning. The cluster established in step 1 is as follows: deploy k3s-agent on the edge devices divided into computing units, and deploy k3s-server on the edge devices divided into management units. (For example, Figure 2 , Figure 4 shown)

[0052] Step 3: Create a resource pool

[0053] After unifying the parameter standards of the edge device resources included in the computing unit (such as converting memory / disk specifications to MB units, CPU specifications to MHz, etc.), statistics are summarized to generate a computing resource pool, and the resources of the same edge device are marked as inseparable. Inseparable resources are the most efficient for data transmission on the same device. If the network transmission rate between devices does not reach 10GB / s, splitting the resources on the same device for different calculations will easily increase data transmission losses.

[0054] Step 4: Task calculation

[0055] First, we split the algorithm service (some standard commonly used algorithms) into operator 1 (a smaller algorithm unit). To facilitate the adjustment of the operator, we set up a "configuration domain" on the operator. Through the "configuration domain", we can control the operator's computing logic and efficiency.

[0056] We then pre-install operators into computing units using containers (Docker) to orchestrate them into a business-capable Algorithm Chain 2 template. During this process, we need to mark the location of the operators and the resources they require during runtime.

[0057] Operator 1 is an independent calculation logic, whose main task is to process the input parameters; algorithm chain 2 combines operators in sequence and is a template for business flow 3; business flow is an instance of the algorithm chain. Activating the algorithm chain allows data to be transferred between operators in the order of the chain, such as Figure 3 shown.

[0058] Step 5: Resource Scheduling

[0059] The management unit monitors the computing resource usage of each operator on the algorithm chain template in real time, and schedules the resources in the computing resource pool according to the usage. Specifically, the management unit monitors the computing resource usage of each operator in the algorithm chain template corresponding to the algorithm task to be calculated in real time. If an instance of an algorithm chain template is completed, the computing resources of the operator corresponding to the algorithm chain template are released to the computing resource pool.

[0060] Example 2

[0061] The present invention also provides an edge device resource scheduling system, comprising:

[0062] Resource acquisition module, used to obtain edge device resources;

[0063] A unit division module is used to divide the acquired edge device resources into computing units and management units;

[0064] A resource pool establishment module is used to establish a computing resource pool according to the number of resources of the computing unit;

[0065] The computing module is used to split the algorithm service into operators, and then pre-install the operators into computing units to compose them into algorithm chain templates;

[0066] Match the algorithm task to be calculated with the algorithm chain template and then calculate;

[0067] The resource scheduling module is used to monitor the computing resource usage of each operator on the algorithm chain template in real time through the management unit, and schedule the resources in the computing resource pool according to the usage.

[0068] The operations performed by the resource acquisition module include:

[0069] Establish communication connections between edge devices;

[0070] Deploy cluster components on edge devices to establish an edge device cluster;

[0071] Collect the number of CPU, GPU, memory, and storage resources of all edge devices used for computing based on the cluster.

[0072] The operations performed by the resource scheduling module include:

[0073] The management unit monitors in real time the computing resource usage of each operator in the algorithm chain template corresponding to the algorithm task to be calculated. If an instance of an algorithm chain template is completed, the computing resources of the operator corresponding to the algorithm chain template are released to the computing resource pool.

[0074] 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 application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the 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 produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. 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.

[0076] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0078] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for scheduling edge device resources, characterized in that: include: Obtain edge device resources; Divide the acquired edge device resources into computing units and management units; Establish a computing resource pool based on the number of computing unit resources; Split the algorithm service into operators, then pre-install the operators into computing units and orchestrate them into algorithm chain templates; Match the algorithm task to be calculated with the algorithm chain template and then calculate; The management unit monitors the computing resource usage of each operator on the algorithm chain template in real time and schedules the resources in the computing resource pool based on the usage; Acquiring edge device resources includes: Establish communication connections between edge devices; Deploy cluster components on edge devices to establish an edge device cluster; Collect the CPU, GPU, memory, and storage resources of all edge devices used for computing based on the cluster; Deploying cluster components on edge devices includes: Deploy k3s-agent on edge devices divided into computing units, and deploy k3s-server on edge devices divided into management units; Establishing a computing resource pool based on the number of computing unit resources includes: The parameter standards of the edge device resources included in the computing unit are unified and statistically summarized to generate a computing resource pool, and the resources of the same edge device are marked as inseparable; The operator also has a configuration field for adjusting the operator logic and computing efficiency; After arranging the operator into an algorithm chain template and pre-installing it in the computing unit, the operator's location and the resources required for runtime are marked; The real-time monitoring of the computing resource usage of each operator on the algorithm chain template by the management unit and scheduling of resources in the computing resource pool according to the usage includes: The management unit monitors in real time the computing resource usage of each operator in the algorithm chain template corresponding to the algorithm task to be calculated. If an instance of an algorithm chain template is completed, the computing resources of the operator corresponding to the algorithm chain template are released to the computing resource pool.

2. A system for scheduling edge device resources, characterized in that: include: Resource acquisition module, used to obtain edge device resources; A unit division module is used to divide the acquired edge device resources into computing units and management units; A resource pool establishment module is used to establish a computing resource pool according to the number of resources of the computing unit; The computing module is used to split the algorithm service into operators, and then pre-install the operators into computing units to compose them into algorithm chain templates; Match the algorithm task to be calculated with the algorithm chain template and then calculate; The resource scheduling module is used to monitor the computing resource usage of each operator on the algorithm chain template in real time through the management unit, and schedule the resources in the computing resource pool according to the usage; The operations performed by the resource acquisition module include: Establish communication connections between edge devices; Deploy cluster components on edge devices to establish an edge device cluster; Collect the CPU, GPU, memory, and storage resources of all edge devices used for computing based on the cluster; Deploying cluster components on edge devices includes: Deploy k3s-agent on edge devices divided into computing units, and deploy k3s-server on edge devices divided into management units; Establishing a computing resource pool based on the number of computing unit resources includes: The parameter standards of the edge device resources included in the computing unit are unified and statistically summarized to generate a computing resource pool, and the resources of the same edge device are marked as inseparable; The operator also has a configuration field for adjusting the operator logic and computing efficiency; After arranging the operator into an algorithm chain template and pre-installing it in the computing unit, the operator's location and the resources required for runtime are marked; The real-time monitoring of the computing resource usage of each operator on the algorithm chain template by the management unit and scheduling of resources in the computing resource pool according to the usage includes: The management unit monitors in real time the computing resource usage of each operator in the algorithm chain template corresponding to the algorithm task to be calculated. If an instance of an algorithm chain template is completed, the computing resources of the operator corresponding to the algorithm chain template are released to the computing resource pool.

Images