CN116010070A

CN116010070A - Deployment method, system, equipment and storage medium of edge cloud platform virtual machine

Info

Publication number: CN116010070A
Application number: CN202111229495.1A
Authority: CN
Inventors: 刘冠思; 廖德甫
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Hangzhou Information Technology Co Ltd
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Hangzhou Information Technology Co Ltd
Priority date: 2021-10-21
Filing date: 2021-10-21
Publication date: 2023-04-25

Abstract

The embodiment of the invention discloses a deployment method, a system, equipment and a storage medium of an edge cloud platform virtual machine. The method comprises the following steps: receiving a virtual machine deployment request, and acquiring resource information of at least one edge node from a storage module according to the virtual machine deployment request; determining a target edge node from at least one edge node according to the resource information of the at least one edge node, wherein the resource information comprises Graphic Processor (GPU) resource information corresponding to each edge node; sending a GPU resource allocation request to a target edge node; receiving target virtual machine and target GPU resource information fed back by a target edge node based on a GPU resource allocation request; according to the target virtual machine and the target GPU resource information, a virtual machine deployment instruction is sent to the target edge node, and the method and the device realize flexible and effective use of the GPU resources by the deployment virtual machine.

Description

Deployment method, system, equipment and storage medium of edge cloud platform virtual machine

Technical Field

The present invention relates to the field of resource management of an edge cloud platform, and in particular, to a method, a system, an apparatus, and a storage medium for deploying a virtual machine of an edge cloud platform.

Background

With the acceleration and transformation of fields such as education, medical treatment, enterprise office and the like driven by the development of cloud computing and artificial intelligence, a graphics processor (Graphics Processing Unit, GPU) is widely focused on the advantages of floating point operation and parallel operation performance, and graphics processing tasks such as image classification, video analysis and the like are involved when large-scale software is developed and designed on a virtualized desktop, so that GPU resources are required to be used on a virtualized platform to improve the concurrent processing capacity of a virtual machine.

When the existing virtual machine system is used for graphic processing, a graphics card direct connection mode is generally adopted, one virtual machine is deployed on a host machine of the virtual machine system, independent binding of the virtual machine and a host machine physical GPU is achieved, only the virtual machine has the right of using GPU resources, the deployment method of the virtual machine leads to lower flexibility of using the GPU resources, the GPU resources can only be used by one virtual machine, and waste of the GPU resources is caused.

Disclosure of Invention

The embodiment of the invention provides a deployment method, a system, equipment and a storage medium of an edge cloud platform virtual machine, which can deploy the virtual machine on the edge cloud platform to realize flexible and effective use of GPU resources.

The technical scheme of the invention is realized as follows:

the embodiment of the invention provides a deployment method of an edge cloud platform virtual machine, which is applied to an edge center node and comprises the following steps:

receiving a virtual machine deployment request, and acquiring resource information of at least one edge node from a storage module according to the virtual machine deployment request;

determining a target edge node from the at least one edge node according to the resource information of the at least one edge node, wherein the resource information comprises Graphic Processor (GPU) resource information corresponding to each edge node;

sending a GPU resource allocation request to the target edge node;

receiving target virtual machine and target GPU resource information fed back by the target edge node based on the GPU resource allocation request;

and sending a virtual machine deployment instruction to the target edge node according to the target virtual machine and the target GPU resource information.

The embodiment of the invention provides a deployment method of an edge cloud platform virtual machine, which is applied to a target edge node and comprises the following steps:

transmitting own resource information to an edge center node, wherein the resource information comprises graphic processor GPU resource information;

receiving a GPU resource allocation request fed back by an edge center node based on the resource information of the target edge node;

Acquiring a target virtual machine according to the GPU resource allocation request, determining target GPU resource information, and sending the target virtual machine and the target GPU resource to the edge center node;

and receiving a virtual machine deployment instruction fed back by the edge center node based on the target virtual machine and the target GPU resource, and deploying the target virtual machine according to the virtual machine deployment instruction.

The embodiment of the invention provides an edge center node device, which comprises:

the storage module is used for storing the resource information of at least one edge node;

the edge cloud platform UI module is used for receiving a virtual machine deployment request and sending the virtual machine deployment request to the edge application management module;

the edge application management module is used for receiving the virtual machine deployment request and acquiring resource information of at least one edge node from the storage module according to the virtual machine deployment request; determining a target edge node from the at least one edge node according to the resource information of the at least one edge node, wherein the resource information comprises Graphic Processor (GPU) resource information corresponding to each edge node; sending a GPU resource allocation request to the target edge node; receiving target virtual machine and target GPU resource information fed back by the target edge node based on the GPU resource allocation request; and sending a virtual machine deployment instruction to the target edge node according to the target virtual machine and the target GPU resource information.

An embodiment of the present invention provides a target edge node device, which is characterized by comprising:

the proxy module is used for sending the self resource information to the edge center node, wherein the resource information comprises graphic processor GPU resource information; receiving a GPU resource allocation request fed back by an edge center node based on the resource information of the target edge node; acquiring a target virtual machine according to the GPU resource allocation request, determining target GPU resource information, and sending the target virtual machine and the target GPU resource to the edge center node;

the arrangement module is used for receiving a virtual machine deployment instruction fed back by the edge center node based on the target virtual machine and the target GPU resource, and deploying the target virtual machine according to the virtual machine deployment instruction.

The embodiment of the invention provides edge center node equipment, which comprises the following components: a first memory and a first processor;

the first memory stores a computer program capable of running on the first processor, and when the first processor executes the computer program, the method for deploying the edge cloud platform virtual machine applied to the edge center node is realized.

The embodiment of the invention provides target edge node equipment, which comprises the following components: a second memory and a second processor;

The second memory stores a computer program capable of running on the second processor, and when the second processor executes the computer program, the deployment method of the edge cloud platform virtual machine applied to the target edge node is realized.

The embodiment of the invention provides a computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and the computer program realizes a deployment method of an edge cloud platform virtual machine applied to an edge center node when being executed by a first processor, or realizes the deployment method of the edge cloud platform virtual machine applied to a target edge node when being executed by a second processor.

The embodiment of the invention provides a deployment method, a system, equipment and a storage medium of an edge cloud platform virtual machine, wherein a virtual machine deployment request is received, resource information of at least one edge node is acquired from a storage module according to the virtual machine deployment request, a target edge node is determined from the at least one edge node according to the resource information of the at least one edge node, so that the target node is determined to be the edge node to be deployed of the virtual machine, the resource information comprises Graphic Processor (GPU) resource information corresponding to each edge node, then, a GPU resource allocation request is sent to the target edge node, a GPU resource which can be used by the virtual machine to be deployed is determined according to the target virtual machine and the target GPU resource information, and a virtual machine deployment instruction is sent to the target edge node according to the target virtual machine and the target GPU resource information. Therefore, the optimal edge node which can be deployed by the virtual machine can be determined according to the resource information of the edge node, and the GPU resource allocation request is sent to the optimal edge node, so that the information corresponding to the GPU resources which can be used by the virtual machine to be deployed is obtained, the virtual machine is deployed according to the information, and the flexible and effective use of the GPU resources by the virtual machine to be deployed is realized.

Drawings

Fig. 1 is a schematic architecture diagram of a distributed system of an edge cloud platform according to an embodiment of the present invention;

fig. 2 is a flow chart of a deployment method of an edge cloud platform virtual machine according to an embodiment of the present invention;

fig. 3 is a flow chart of a method for determining a target edge node according to an embodiment of the present invention;

fig. 4 is a flow chart of another method for deploying an edge cloud platform virtual machine according to an embodiment of the present invention;

fig. 5 is a flowchart of another method for deploying an edge cloud platform virtual machine according to an embodiment of the present invention;

fig. 6 is a flowchart of a deployment method of an edge cloud platform virtual machine according to an embodiment of the present invention;

fig. 7 is a flowchart of a deployment method of an edge cloud platform virtual machine according to an embodiment of the present invention;

fig. 8 is a schematic diagram of a capability distribution architecture of an edge cloud platform according to an embodiment of the present invention;

fig. 9 is a schematic flow chart of deploying and calling vGPU by an edge cloud platform virtual machine according to an embodiment of the present invention;

fig. 10 is a schematic structural diagram of an edge center node device according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of a target edge point device according to an embodiment of the present invention;

Fig. 12 is a schematic structural diagram of an edge center node device according to an embodiment of the present invention;

fig. 13 is a schematic structural diagram of a target edge node device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention, and the described embodiments should not be construed as limiting the present invention, and all other embodiments obtained by those skilled in the art without making any inventive effort are within the scope of the present invention.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is to be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with one another without conflict.

In the following description, references to "one\another\yet another" are merely to distinguish similar objects and do not represent a particular ordering for the objects, it being understood that "one\another\yet another" may be interchanged in a particular order or precedence where allowed, so that embodiments of the invention described herein can be practiced in an order other than that illustrated or described herein.

The development of cloud computing, artificial intelligence drives the acceleration and transformation of the fields of education, medical treatment, enterprise office and the like, the GPU has massive market demands in terms of excellent performance of floating point operation and parallel operation, the dependency of a new generation of operating system and software on the GPU is heavier and heavier, the GPU comprises thousands of computing units, the parallel computing aspect shows strong advantages, the GPU through instance is specially optimized for deep learning, and massive computation can be completed in a short time.

The cloud desktop used for the cloud computing can completely replace a common computer, brings great advantages in daily offices, and large-scale 3D design including image classification, video analysis, voice recognition and natural voice processing is needed by workers developing and designing on the cloud desktop.

In the prior virtual machine system, graphics processing is generally carried out in a graphics card direct connection mode, a virtual machine is deployed on a host machine of the virtual machine, independent binding of the virtual machine and a physical GPU of the host machine is realized, and only the virtual machine has the right of using the GPU and directly accesses the physical GPU through driving. Although the method can save the integrity and independence of the physical GPU, so that the virtual machine can perform general calculation, the deployment method for realizing the virtual machine has lower flexibility in using GPU resources, and the GPU resources can only be used by one virtual machine, so that the waste of GPU resources is caused.

The embodiment of the invention provides a deployment method of an edge cloud platform virtual machine, which can deploy the virtual machine on the edge cloud platform to realize flexible and effective use of GPU resources. The edge cloud is a cloud data center which is distributed on the edge side of the network and provides real-time data processing and analysis decision. The edge calculation is a distributed operation architecture, and the operation of application programs, data materials and services is processed by moving a network center node to an edge node on the network logic; cloud computing is a distributed system that links various services, such as software development platforms, servers, and other software, through the internet. The edge cloud platform is based on the cloud computing technology and the edge computing capability, the cloud computing platform is built on the basis of edge facilities, an elastic cloud platform with comprehensive computing, network, storage, safety and other capabilities of the edge end is formed, and low-delay, high-safety and schedulable distributed cloud services are provided for the user terminal.

The deployment method of the edge cloud platform virtual machine provided by the embodiment of the invention is based on the distributed system architecture of the edge cloud platform, a schematic diagram of the distributed system architecture of an optional edge cloud platform is shown in fig. 1, and the distributed system 100 of the edge cloud platform comprises an edge center node 200 and at least one edge node 300.

The edge center node 200 comprises an edge cloud platform User Interface (UI) module 201, an edge application management module 202 and a storage module 203, wherein the edge cloud platform UI module 201 is connected with the edge application management module 202, and the edge application management module 202 is connected with the storage module 203; the edge node 300 module comprises an agent module 301 and an orchestration module 302, the agent module 301 and the orchestration module 302 being connected. The edge application management module 202 in the edge center node 200 and the edge node 300 interact data through the application program interface (Representational State Transfer Ful Application program interface, RESTful API) for the representational state transfer.

It should be noted that, the edge center node may be an edge center node server, the edge node 300 may be an edge node server, the edge center node server and the edge node server are cloud servers disposed on an edge cloud platform, and at least one of the edge center node server and the edge node server is provided.

Next, a deployment method of the edge cloud platform virtual machine provided by the embodiment of the invention will be described. As shown in fig. 2, a flowchart of a method for deploying an edge cloud platform virtual machine according to an embodiment of the present invention is shown, where the method includes the following steps:

s101, receiving a virtual machine deployment request, and acquiring resource information of at least one edge node from a storage module according to the virtual machine deployment request.

The embodiment of the invention is a method for deploying the virtual machine on the edge cloud platform, and the edge cloud platform is arranged at a position close to the terminal equipment, so that the instantaneity and the safety of data transmission can be ensured. A Virtual Machine (VM) is a complete computer system that has a complete hardware system function through software simulation and operates in a completely isolated environment, and works that can be completed in a physical computer can be implemented in a Virtual Machine, for example, processing works involving image recognition, image classification, and the like of artificial intelligence. The virtual machine is deployed on the cloud server in the edge cloud platform, so that the resources of the cloud server can be shared to realize the processing of multiple tasks or complex tasks.

In some embodiments, the user sends the virtual machine deployment request to the edge cloud platform at the terminal device, where the manner of sending the virtual machine deployment request may be that the user clicks a virtual machine deployment instruction on the terminal device interface, or other operations capable of triggering sending the virtual machine deployment request to the edge cloud platform.

It should be noted that, when the edge cloud platform receives the virtual machine deployment request, specifically, the edge cloud platform UI module 201 in the edge center node 200 receives the virtual machine deployment request, and then the edge cloud platform UI module 201 forwards the deployment request to the edge application management module 202, and after the edge application management module 202 receives the virtual machine deployment request, the storage module 203 in the edge center node 200 queries to obtain the resource information of the edge node server where the edge node 300 is located.

S102, determining a target edge node from at least one edge node according to the resource information of the at least one edge node, wherein the resource information comprises the GPU resource information of the graphics processor corresponding to each edge node.

The edge cloud platform includes one or more edge nodes, and the edge application management module 202 may obtain resource information of all edge nodes, where the resource information of an edge node may be hardware resource information of an edge node server, for example, resource information of a GPU, a CPU, a memory, a hard disk, etc., specifically, may be the utilization ratio of the GPU and the CPU, the capacity of the memory and the hard disk, etc., which is not limited herein.

It should be noted that, the edge application management module 202 may obtain the resource information of each edge node, analyze the resource information of each edge node, and obtain the resource usage situation of each edge node, so as to determine the target edge node according to the resource usage situation of all edge nodes. The decision method for determining the target edge node by the edge application management module 202 may be to comprehensively compare all edge nodes, and select the edge node with the lowest utilization rate of GPU and CPU and the highest memory and hard disk capacity in all edge nodes, so as to use the selected edge node as the target edge node.

In some embodiments, the target edge node may be any one of all edge nodes, where the target edge node determined by the edge application management module 202 is the best one of all target edge nodes with the best resource information, where the resource information may be the GPU and CPU of the target edge node with the lowest utilization, and the most memory and hard disk capacity, and may also include other methods capable of indicating the best resource information, which is not limited herein.

In other embodiments, there may be at least two target edge nodes determined by the edge application management module 202, where the resource information conditions of the at least two target edge nodes may be the same or similar, and the resource information conditions may be similar determination results obtained after comprehensively comparing various information resources of the plurality of target edge nodes.

S103, sending a GPU resource allocation request to a target edge node;

the target edge node is an edge node determined by the edge application management module 202 to be capable of deploying a virtual machine, the edge node server corresponding to the target edge node is provided with a GPU resource, and the virtual machine deployed on the edge node server corresponding to the target node can execute a graphics processing task by using the GPU resource. After the edge application management module 202 determines the target edge node, in order to obtain the GPU resources of the target edge node, a GPU resource allocation request needs to be sent to the target edge node.

It should be noted that, the GPU resource allocation request may represent a request to acquire a GPU resource of the target edge node, and further may also represent a request to deploy a virtual machine on the target edge node, where the GPU resource allocation request may be a request sent by the edge application management module 202 to the target edge node.

S104, receiving target virtual machine and target GPU resource information fed back by the target edge node based on the GPU resource allocation request;

after sending the GPU resource allocation request to the target edge node, the edge application management module 202 receives information fed back by the target edge node, where the fed back information includes information required for deploying the virtual machine, and the information fed back by the target edge node may be the target virtual machine and the target GPU resource information. The target virtual machine is associated with the virtual machine to be deployed, and the target GPU resource information is associated with GPU resource information of the target edge node.

In some embodiments, the target virtual machine may characterize identity information of the virtual machine to be deployed, which may be an identity, such as a universally unique identification code (Universally Unique IDentifier, UUID) of the virtual machine. The target GPU resource information may be identification information of part or all of GPU resources of an edge node server where the target edge node is located, and specifically may be identification numbers (IDentity document, IDs) of part or all of GPU resources.

It can be understood that after the edge application management module 202 receives the target virtual machine and the target GPU resource information fed back by the target edge node, the identification information of the virtual machine and the identification information corresponding to the GPU resource that can be used by the virtual machine to be deployed can be clarified, that is, the key information of virtual machine deployment is obtained.

S105, sending a virtual machine deployment instruction to a target edge node according to the target virtual machine and the target GPU resource information.

After obtaining the target virtual machine information and the target GPU resource information fed back by the target edge node, the edge application management module 202 determines that the conditions for virtual machine deployment have been satisfied, and then the edge application management module 202 sends a virtual machine deployment instruction to the target virtual machine to complete the deployment of the virtual machine to be deployed.

In the embodiment of the invention, a virtual machine deployment request is received, resource information of at least one edge node is acquired from a storage module according to the virtual machine deployment request, a target edge node is determined from the at least one edge node according to the resource information of the at least one edge node, so that the target node is the edge node to be deployed of the virtual machine, wherein the resource information comprises Graphic Processor (GPU) resource information corresponding to each edge node, then, a GPU resource allocation request is sent to the target edge node, a target virtual machine and target GPU resource information fed back by the target edge node based on the GPU resource allocation request are received, GPU resources which can be used by the virtual machine to be deployed are determined, and a virtual machine deployment instruction is sent to the target edge node according to the target virtual machine and the target GPU resource information. Therefore, the optimal edge node which can be deployed by the virtual machine can be determined according to the resource information of the edge node, and the GPU resource allocation request is sent to the optimal edge node, so that the information corresponding to the GPU resources which can be used by the virtual machine to be deployed is obtained, the virtual machine is deployed according to the information, and the GPU resources are flexibly and effectively used.

Fig. 3 is a flow chart of a method for determining a target edge node according to an embodiment of the present invention, where in some embodiments of the present invention, resource information further includes: the specific implementation process of determining the target edge node from at least one edge node according to the CPU and the memory information of the central processing unit corresponding to each edge node and the resource information of at least one edge node, namely S102, may include:

s1021, determining at least one candidate edge node based on GPU resource information of the at least one edge node.

The edge application management module 202 may determine at least one candidate edge node, which is an edge node featuring GPU resources, based on resource information on an edge node server to which the at least one edge node corresponds.

In some embodiments, the edge node with the GPU resource may be an edge node with a low GPU resource utilization rate, or may be an edge node with a high GPU resource utilization rate, which is not specifically limited herein, and may be all candidate edge nodes as long as the edge node has the GPU resource.

It should be noted that, the GPU resource information includes: the number of virtualized graphics processor vGPU instances corresponding to each edge node. The GPU resources on the edge node may be multiple vGPU instances obtained by virtualizing the GPU, each vGPU instance has GPU resources, the vGPU instances are independent, the virtualized GPU resources may be used by multiple virtual machines to realize resource sharing, for example, one virtual machine uses one vGPU instance, all the vGPU instances become all the resources of the GPU on the edge node, and the GPU resource information of the edge node may also include the name of the edge node.

In some embodiments, determining the candidate edge node based on the GPU resource information of the at least one edge node, i.e., S1021, the above-described implementation may include S1021a, as follows:

s1021a, if at least one edge node has the edge node with the number of the corresponding vGPU instances being greater than zero, determining the edge node as at least one candidate edge node.

It can be understood that if there is an available GPU resource on the edge node, the GPU resource may be virtualized to obtain at least one vGPU instance, so when determining the candidate edge node according to the GPU resource information of the edge node, the number of vGPU instances in the edge node may be directly determined, and if the number of vGPU instances is greater than zero, it indicates that the edge node has GPU resources, where the number of vGPU instances is a positive integer greater than zero; if the number of vGPU instances is equal to zero, then this edge node is indicated as having no GPU resources. By using the method of determining candidate edge nodes by the number of corresponding vGPU instances present in the edge node, candidate edge nodes having GPU resources can be determined more quickly and conveniently.

S1022, determining the target edge node according to the CPU and the memory information of at least one candidate edge node.

After determining the candidate edge node with the GPU resource, other resource information such as a CPU, a memory, a hard disk, etc. of the candidate edge node needs to be considered, and it should be noted that the candidate edge node with the GPU resource can ensure that the virtual machine deployed on the candidate edge node invokes the GPU resource to execute the graphics processing task, however, other resource information such as the CPU, the memory, etc. of the candidate edge node is also an important parameter for reflecting the performance of the virtual machine. Therefore, other resource information such as a CPU and a memory of the candidate edge node needs to be further considered to determine an optimal edge node in the candidate edge nodes, and the optimal edge node is taken as a target edge node.

In some embodiments, the optimal edge node may be a candidate edge node with the lowest CPU utilization rate and the most memory and hard disk capacity, and specifically, the situations of resources such as CPU, memory, hard disk and the like in each candidate edge node may be comprehensively compared, and the candidate edge node with the optimal resource situation is determined as the target edge node.

It should be noted that, the GPU resource in the target edge node includes at least one vGPU instance, the target GPU resource information obtained in S104 includes vGPU instance information, and the target GPU resource information indicates any vGPU instance included in the GPU resource information of the target edge node.

In some embodiments of the present disclosure, before receiving the virtual machine deployment request, that is, before S101, the deployment method of the edge cloud platform virtual machine further includes:

s101a, receiving the resource information of at least one edge node, and storing the resource information of the at least one edge node into a storage module.

It should be noted that, the storage module 203 in the edge center node stores the resource information of all edge nodes in advance, after the edge cloud platform is established, all edge nodes correspond to respective edge node servers, and the resource information of the edge node servers is sent to the edge center node, specifically, may be the edge application management module 202, and after the edge application management module receives the resource information, the edge application management module stores the resource information to the storage module 203.

It can be understood that, the edge application management module 202 receives the resource information of all edge nodes, and stores the resource information of all edge nodes in the storage module 203, so that the edge center node does not need to send the resource information request to all edge nodes after receiving the virtual machine deployment request, thereby accelerating the deployment process of the virtual machine and shortening the deployment time of the virtual machine.

In some embodiments of the present disclosure, after receiving the target virtual machine and the target GPU resource information fed back by the target edge node based on the GPU resource allocation request, that is, after step S104, the deployment method of the edge cloud platform virtual machine further includes: s104a, updating, in the storage module 203, the number of vGPU instances available in the GPU resource information of the target edge node.

The method comprises the steps that a target virtual machine determines the identification of a virtual machine to be deployed, target GPU resource information determines partial GPU resources on a target edge node, the partial GPU resources correspond to available vGPU instances, so that the virtual machine to be deployed can call GPU resources corresponding to the vGPU instances, and when the vGPU instances are used by the virtual machine to be deployed, the vGPU instances become unavailable vGPU instances in the target edge node. Thus, after the edge application management module 202 receives the target virtual machine and the target GPU resource information fed back by the target edge node, the edge application management module 202 needs to update the GPU resource information in the storage module 203, where the updated GPU resource information includes the number of vGPU instances available in the target edge node.

It can be appreciated that by updating the number of vGPU instances available in the target edge node, the edge application management module 202 can be guaranteed to obtain the correct information about all edge nodes when the virtual machine deployment request is received next time, and normal operation of virtual machine deployment is facilitated.

In another embodiment of the disclosure, as shown in fig. 4, a flowchart of another method for deploying an edge cloud platform virtual machine provided by the present disclosure, after sending a virtual machine deployment instruction to a target edge node, that is, after S105, the method for deploying an edge cloud platform virtual machine further includes:

s106, receiving a virtual machine deployment result.

The virtual machine deployment instruction sent by the edge application management module 202 to the target edge node indicates that the virtual machine can be deployed, the specific process of deployment is executed by the target edge node, and after the virtual machine deployment is completed, the edge application management module 202 receives a virtual machine deployment result sent by the target edge node, and the virtual machine deployment result indicates whether the deployment is successful or failed.

Fig. 5 is a flow chart of a deployment method of an edge cloud platform virtual machine according to an embodiment of the present invention, where the method is applied to a target edge node, and the method includes the following steps:

s401, transmitting self resource information to an edge center node, wherein the resource information comprises graphic processor GPU resource information;

before the edge cloud platform receives the virtual machine deployment request sent by the end user, or after the edge cloud platform is established, all edge nodes in the edge cloud platform send their own resource information to the edge center node, specifically, the proxy module 301 in the edge node sends the GPU resource information of the node where the proxy module is located to the edge center node.

In some embodiments, the resource information further includes central processing unit CPU and memory information; the step of sending the resource information of the target edge node to the edge center node, that is, the implementation process of S401 may further include S401a, as follows:

s401a, GPU resource information, CPU and memory information of the target edge node are sent to the edge center node.

It should be noted that, the resource information may be information such as a hard disk, a video memory, etc., and all other resource information of the edge node except the GPU resource information is stored in the compiling module 302.

In some embodiments, when the edge node is determined to be a candidate edge node, the orchestration module 302 in the candidate edge node transmits the resource information to the proxy module 301, where the resource information is all other information excluding GPU resources in the candidate node, and after receiving the resource information, the proxy module sends the resource information to the edge application management module of the edge center node.

In some embodiments, the GPU resource information of the target edge node includes: and according to at least one virtualized graphics processor (vGPU) instance created by the GPU resources of the target edge node, the GPU resources on the target edge node can be a plurality of vGPU instances obtained by virtualization of the GPU, each vGPU instance is provided with GPU resources, the vGPU instances are mutually independent, and the virtualized GPU resources can be used by a plurality of virtual machines to realize resource sharing.

It can be understood that when the edge node is the target edge node, all the resource information of the target edge node needs to be sent to the edge center node, so that the edge center node performs the subsequent operation of virtual machine deployment according to the resource information of the target edge node.

S402, receiving a GPU resource allocation request fed back by an edge center node based on resource information of a target edge node;

after the target edge node sends the resource information to the edge center node, a GPU resource allocation request sent by the edge center node is received, where the GPU resource allocation request may indicate that the GPU resource of the target edge node is requested to be acquired, and further may also indicate that a request for deploying a virtual machine on the target edge node, and specifically may be that the proxy module 301 receives the GPU resource allocation request.

S403, acquiring a target virtual machine according to the GPU resource allocation request, determining target GPU resource information, and sending the target virtual machine and target GPU resources to an edge center node;

after receiving the GPU resource allocation request sent by the edge center node, the proxy module 301 will automatically acquire the target virtual machine, where the target virtual machine may represent identity information of the virtual machine to be deployed, and the identity information may be an identity identifier, for example, a UUID corresponding to the virtual machine. In some embodiments, a mirror interface may be provided by the proxy module 301 of the target edge node, and the virtual machine is started to obtain the identity of the virtual machine.

It should be noted that, after the target virtual machine is acquired, the proxy module 301 may determine the target GPU resource information on the target edge node.

In some embodiments, the target GPU resource information indicates any vGPU instance included in the GPU resource information of the target edge node. After the GPU resource of the target edge node is virtualized, at least one vGPU instance is obtained, and when the proxy module 301 obtains the identity of the virtual machine, one vGPU instance can be selected from all the vGPU instances of the target edge node, where the vGPU instance corresponds to unique identification information, and the identification information can be an ID of the vGPU instance. After the proxy module 301 acquires the target virtual machine and the target GPU resource, the target virtual machine and the target GPU resource are sent to the edge application management module 202 of the edge center node.

S404, receiving a virtual machine deployment instruction fed back by the edge center node based on the target virtual machine and the target GPU resource, and deploying the target virtual machine according to the virtual machine deployment instruction.

The proxy module 301 of the target edge node receives a virtual machine deployment instruction sent by the edge center node, where the virtual machine deployment instruction indicates that the target edge node already has a condition for deploying a virtual machine, and specifically, the condition may be that an identity of the target virtual machine is obtained, and an ID of a vcpu instance that can be used, and then deployment of the virtual machine may be performed.

In some embodiments, deploying a target virtual machine according to a virtual machine deployment instruction includes: and binding the target GPU resource information with the target virtual machine according to the virtual machine deployment instruction so as to realize the deployment of the target virtual machine. When the virtual machine is specifically deployed, the proxy module sends the previously obtained target virtual machine and target GPU resource information to the orchestration module 302, the orchestration module 302 binds the target virtual machine and the target GPU resource information, the virtual machine to be deployed after binding is associated with the target edge node, and GPU resources corresponding to the vGPU instance of the target edge node can be used.

It will be appreciated that after the orchestration module 302 binds the target virtual machine and the target GPU resource information, the virtual machine may be created and started up, until the virtual machine deployment is complete.

In still another embodiment of the present disclosure, as shown in fig. 6, a flowchart of a method for deploying an edge cloud platform virtual machine according to the present disclosure, after deploying a target virtual machine according to a virtual machine deployment instruction, the method for deploying an edge cloud platform virtual machine further includes:

s405, obtaining a virtual machine deployment result; and sending the virtual machine deployment result to the edge center node.

The deployment result of the virtual machine indicates that the virtual machine deployment succeeds or fails, and in particular, the deployment result may be sent by the orchestration module 302 to the edge application management module 202 of the edge center node.

In some embodiments, the successful deployment of the virtual machine may include successful binding of the target virtual machine and the target GPU resource information, and smooth creation and starting of the virtual machine; correspondingly, the virtual machine deployment failure may be any one of failure of successful binding of the target virtual machine and the target GPU resource information, failure of virtual machine creation, and failure of virtual machine startup.

In still another embodiment of the present disclosure, as shown in fig. 7, a flowchart of a method for deploying an edge cloud platform virtual machine according to the present disclosure is provided, where the method involves an interaction process of a terminal, an edge center node, and an edge node, and the method includes:

s501, the edge node sends the self resource information to the edge center node, wherein the resource information comprises GPU resource information of a graphic processor.

S502, the edge center node stores the resource information of at least one edge node into a storage module.

The storage module 203 in the edge center node stores the resource information of all edge nodes in advance, after the edge cloud platform is established, all edge nodes correspond to respective edge node servers, the resource information of the edge node servers is sent to the edge center node, specifically, the edge application management module 202, and after receiving the resource information, the edge application management module stores the resource information to the storage module 203.

S503, the terminal user sends a virtual machine deployment request to the edge center node.

According to the actual demand of virtual machine deployment, the end user sends a virtual machine deployment request to an edge center node on the edge cloud platform, specifically, the user clicks a virtual machine deployment instruction on an interface of the terminal equipment to send the virtual machine deployment request to an edge cloud platform UI module 201 in the edge center node, and the edge cloud platform UI module 201 can forward the virtual machine deployment request to an edge application management module 202.

S504, the edge center node acquires resource information of at least one edge node from the storage module according to the virtual machine deployment request.

It should be noted that, when the edge cloud platform receives the virtual machine deployment request, specifically, after the edge cloud platform UI module 201 in the edge center node 200 receives the virtual machine deployment request, the edge cloud platform UI module 201 forwards the deployment request to the edge application management module 202, and after the edge application management module 202 receives the virtual machine deployment request, the storage module 203 in the edge center node queries to obtain resource information of the edge node server where the edge node 300 is located.

S505, the edge center node determines a target edge node from at least one edge node according to the resource information of the at least one edge node, wherein the resource information comprises the GPU resource information of the graphics processor corresponding to each edge node.

It may be appreciated that the edge application management module 202 may obtain the resource information of each edge node, analyze the resource information of each edge node, and obtain the resource usage of each edge node, so as to determine the target edge node according to the resource usage of all edge nodes. The decision method for determining the target edge node by the edge application management module 202 may be to comprehensively compare all edge nodes, and select the edge node with the lowest utilization rate of GPU and CPU and the highest memory and hard disk capacity among all edge nodes, so as to use the selected edge node as the target edge node.

S506, the edge center node sends a GPU resource allocation request to the target edge node.

After the edge application management module 202 determines the target edge node, in order to obtain the GPU resources of the target edge node, a GPU resource allocation request needs to be sent to the target edge node. It should be noted that, the GPU resource allocation request may represent a request to acquire a GPU resource of the target edge node, and further may also represent a request to deploy a virtual machine on the target edge node, where the GPU resource allocation request may be a request sent by the edge application management module 202 to the target edge node.

S507, the target edge node obtains the target virtual machine according to the GPU resource allocation request, determines target GPU resource information and sends the target virtual machine and the target GPU resource to the edge center node.

After receiving the GPU resource allocation request sent by the edge center node, the proxy module 301 of the target edge node will automatically acquire the target virtual machine, where the target virtual machine may represent identity information of the virtual machine to be deployed, and the identity information may be an identity identifier, for example, a UUID corresponding to the virtual machine. In some embodiments, a mirror interface may be provided by the proxy module 301 of the target edge node, and the virtual machine is started to obtain the identity of the virtual machine.

S508, the edge center node sends a virtual machine deployment instruction to the target edge node according to the target virtual machine and the target GPU resource information.

After the edge application management module 202 of the edge center node obtains the target virtual machine information and the target GPU resource information fed back by the target edge node, it is determined that the conditions for virtual machine deployment are satisfied, and then the edge application management module 202 sends a virtual machine deployment instruction to the target virtual machine to complete deployment of the virtual machine to be deployed.

S509, the target edge node deploys the target virtual machine according to the virtual machine deployment instruction.

S510, the target edge node obtains a virtual machine deployment result and sends the virtual machine deployment result to the edge center node.

The deployment result of the virtual machine indicates that the virtual machine deployment succeeds or fails, and specifically, the deployment result may be sent by the orchestration module 302 of the target edge node to the edge application management module 202 of the edge center node.

S511, the edge center node returns the deployment result of the virtual machine to the end user.

The specific process of virtual machine deployment is executed by the target edge node, after the virtual machine deployment is completed, the edge application management module 202 of the edge center node receives the virtual machine deployment result sent by the target edge node, the virtual machine deployment result indicates that the deployment is successful or failed, and then the edge application management module 202 forwards the virtual machine deployment result to the edge cloud platform user interface module 201, and the edge cloud platform user interface module 201 returns the deployment result to the end user.

The following describes the implementation process of the embodiment of the invention in an actual application scene.

The method for deploying and calling the vGPU by the edge computing platform virtual machine (the deploying method of the edge cloud platform virtual machine) provided by the embodiment of the invention is realized by an edge cloud-based capacity distribution architecture 600 (a distributed system architecture of an edge cloud platform).

Fig. 8 is a schematic diagram of a capability distribution architecture of an edge cloud according to an embodiment of the present invention. Referring to fig. 8, the capability distribution architecture 600 of an edge cloud includes: edge node 700 and edge center node 800.

Edge center node 700 (edge center node 200) includes edge cloud platform UI module 701 (edge cloud platform UI module 201), edge application management system 702 (edge application management module 202), and Redis module 703 (storage module 203), edge cloud platform UI module 701 and edge application management system 702 are connected, and edge application management system 702 and Redis 703 are connected; the edge node 800 includes a Libvirt proxy module 801 (proxy module 301) and an Andmec orchestration module 802 (orchestration module 302), the Libvirt proxy module 801 and the Andmec orchestration module 802 being connected. The edge application management system 702 in the edge center node 700 and the edge node 800 interact with data through the RESTful API interface.

It should be noted that, the edge cloud platform UI module 701 is responsible for forwarding front-end routing information (virtual machine deployment request); the edge application management system 702 is responsible for specific task scheduling, and issues tasks to the optimal edge node (target edge node) according to the resource condition (resource information) of each node in the cluster (at least one edge node); the Libvirt proxy module 801 in the edge node 800 is responsible for the creation of vGPU resources, vGPU quantity inquiry and resource allocation; the Andmec orchestration module 802 is mainly responsible for acquiring cluster information, and the virtual machine deployment is started; the Redis module 703 is mainly responsible for data persistence service, and stores vGPU resource data of each node in the cluster (stores resource information of the edge node), and simultaneously serves as a basis for synchronization (updating GPU resource information of the target edge node) and recovery of the cluster vGPU resource information.

Fig. 9 is a schematic flow chart of an edge computing platform virtual machine deployment and invocation vGPU provided in an embodiment of the present invention, referring to fig. 9, a process of the edge computing platform virtual machine deployment and invocation vGPU includes:

and S901, initializing and sending the number of the vGPU.

The edge node creates GPU resources and sends GPU resource information of the edge node to the edge center node. The Libvirt agent module of the edge node instantiates the GPU according to the GPU resources of the node where the Libvirt agent module is located to create a plurality of vGPU instances (at least one vGPU instance), and sends the edge node name and the number of vGPU instances (GPU resource information) to an edge application management system of the edge center node (sends the resource information of the Libvirt agent module to the edge center node).

After the edge application management system of the edge center node receives the GPU resource information sent by the edge node (receives the resource information of at least one edge node), the GPU resource information is sent to a redis module, the redis module stores the GPU resource information of at least one edge node (stores the GPU resource information in a storage module), the redis module can return the stored vGPU number result to the Libvirt agent module, and the Libvirt agent module determines that the redis module has completed storing the vGPU number information after receiving the result.

S902, receiving a virtual machine deployment request and inquiring the number of vGPUs.

And receiving a virtual machine deployment request sent to the UI module of the edge cloud platform, and inquiring the vGPU resource condition of each edge node according to the request (acquiring resource information of at least one edge node from a storage module).

The method comprises the steps that a terminal user sends a virtual machine deployment request to a virtual machine platform, specifically sends the request to an edge cloud platform UI module, the edge cloud platform UI module receives the request and then sends the request to an edge application management system, the edge application management system firstly sends a request to a redis module to check the cache of the redis module so as to inquire the vGPU resource condition (GPU resource information of at least one edge node) of each edge node, and the redis module inquires the number of the vGPU resources stored by the terminal user according to the request sent by the edge application management system to check the cache of the redis module, and then the inquiry result of the number of the vGPU resources is returned to the edge application management system.

S903, sending a vGPU resource allocation request.

And determining the edge nodes which can be scheduled according to the query result of the vGPU resources of each edge node. The edge application management system inquires the vGPU resource condition of each edge node, if the edge node does not have available vGPU resources (the number of vGPU instances), the edge node is removed from the virtual machine deployment schedule at this time, and the virtual machine deployment schedule is scheduled in the rest edge nodes (candidate edge nodes).

CPU and memory use conditions (CPU and memory information of candidate edge nodes) are obtained from the edge nodes with available vGPU resources, and virtual machine deployment tasks are scheduled to the optimal edge nodes (target edge nodes) according to the resource conditions of the edge nodes.

After determining the edge nodes (candidate edge nodes) with available vGPU resources, the edge application management system sends requests for acquiring CPU and memory usage (CPU and memory information of the candidate edge nodes) to the Andmec orchestration module of the edge nodes with available vGPU resources, and the edge application management system dispatches virtual machine deployment tasks to the optimal edge nodes (target edge nodes) according to the CPU and memory usage (CPU and memory information of the candidate edge nodes) in the edge nodes with available vGPU resources, and then sends a vGPU resource allocation request (GPU resource allocation request) to the optimal edge nodes (target edge nodes) by the edge application management system.

S904, obtaining UUIDs of the virtual machines and IDs of the vGPU instances and updating vGPU information.

When the Lbvirt proxy module receives a vGPU resource allocation request (GPU resource allocation request), starting the virtual machine in a Libvirt mode to acquire a UUID (target virtual machine) of the virtual machine, selecting an ID (target GPU resource information) from the previously created vGPU instance, returning the UUID (target virtual machine) of the virtual machine and the ID (target GPU resource information) of the vGPU to the edge application management system, updating the locally available vGPU number to the Redis module by the edge application management system, returning the updated vGPU number result to the Lbvirt proxy module by the Redis module, and determining that the update of the vGPU number is completed by the Redis module after the result is received by the Lbvirt proxy module.

S905, creating a virtual machine and returning a virtual machine deployment result.

And after receiving UUID information of the virtual machine, the edge application management system creates and starts the virtual machine by utilizing the Andmec orchestration module, and returns a result (virtual machine deployment result) to the edge cloud platform UI module.

According to the method, an end user sends a virtual machine deployment request to the UI module of the edge cloud platform, so that the GPU virtual machine is created in the edge application management system as required, the edge center management application management system sends a vGPU allocation request to the optimal edge node according to the resource information of each edge node, the Libvirt agent module deployed by each edge node allocates the ID of the vGPU according to the vGPU request, the UUID of the virtual machine required for deployment is obtained, and the virtual machine with the vGPU resources is finally started by the Andmec arrangement module. When the user does not need the virtual machine, the virtual machine is destroyed, the corresponding vGPU resources are released and can be used by other virtual machines, and the flexible and effective use of the GPU resources is realized.

An embodiment of the present invention provides an edge center node device, and fig. 10 is a schematic structural diagram of the edge center node device provided in the embodiment of the present invention, as shown in fig. 10, the edge center node device 1 includes:

the edge cloud platform UI module 11 is used for receiving a virtual machine deployment request and sending the virtual machine deployment request to the edge application management module;

the edge application management module 12 is configured to receive the virtual machine deployment request, and obtain resource information of at least one edge node from the storage module according to the virtual machine deployment request; determining a target edge node from the at least one edge node according to the resource information of the at least one edge node, wherein the resource information comprises Graphic Processor (GPU) resource information corresponding to each edge node; sending a GPU resource allocation request to the target edge node; receiving target virtual machine and target GPU resource information fed back by the target edge node based on the GPU resource allocation request; sending a virtual machine deployment instruction to the target edge node according to the target virtual machine and the target GPU resource information;

a storage module 13, configured to store resource information of at least one edge node.

In some embodiments, the resource information further comprises: CPU and memory information of the central processing unit corresponding to each edge node; the edge application management module 13 is further configured to determine at least one candidate edge node based on GPU resource information of the at least one edge node, including: if the at least one edge node has the edge node with the number of the corresponding vGPU instances being greater than zero, determining the edge node as the at least one candidate edge node; and determining the target edge node according to the CPU and memory information of the at least one candidate edge node. The GPU resource information includes: the number of virtualized graphics processor vGPU instances corresponding to each edge node. The target GPU resource information indicates any vGPU instance included in the GPU resource information of the target edge node.

In some embodiments, the edge application management module 13 is further configured to receive resource information of at least one edge node before receiving the virtual machine deployment request, and store the resource information of the at least one edge node in the storage module.

In some embodiments, the edge application management module 13 is further configured to update, after receiving the target virtual machine and the target GPU resource information fed back by the target edge node based on the GPU resource allocation request, the number of vGPU instances available in the GPU resource information of the target edge node in the storage module.

In some embodiments, the edge application management module 13 is further configured to receive a virtual machine deployment result after sending a virtual machine deployment instruction to the target edge node.

An embodiment of the present invention provides a target edge node device, and fig. 11 is a schematic structural diagram of the target edge node device provided in the embodiment of the present invention, as shown in fig. 11, the target edge node device 2 includes:

the proxy module 21 is configured to send own resource information to the edge center node, where the resource information includes graphics processor GPU resource information; receiving a GPU resource allocation request fed back by an edge center node based on the resource information of the target edge node; acquiring a target virtual machine according to the GPU resource allocation request, determining target GPU resource information, and sending the target virtual machine and the target GPU resource to the edge center node;

and the orchestration module 22 is configured to receive a virtual machine deployment instruction fed back by the edge center node based on the target virtual machine and the target GPU resource, and deploy the target virtual machine according to the virtual machine deployment instruction.

In some embodiments, the resource information further comprises: the proxy module 21 is further configured to send GPU resource information, CPU and memory information of the target edge node to the edge center node.

In some embodiments, the orchestration module 22 is further configured to bind the target GPU resource information with the target virtual machine according to the virtual machine deployment instruction, so as to implement deployment of the target virtual machine.

In some embodiments, the proxy module 21 is further configured to obtain a virtual machine deployment result after the target virtual machine is deployed according to the virtual machine deployment instruction; and sending the virtual machine deployment result to the edge center node.

The embodiment of the present invention further provides an edge center node device, as shown in fig. 12, which is a schematic structural diagram of the edge center node device provided by the embodiment of the present invention, where the edge center node device 3 includes: a first memory 31 and a first processor 32;

the first memory stores 31 a computer program executable on the first processor, which when executed by the first processor 32 implements a method of deploying an edge cloud platform virtual machine as shown in fig. 2-4.

The embodiment of the present invention further provides a target edge node device, as shown in fig. 13, which is a schematic structural diagram of the target edge node device provided by the embodiment of the present invention, where the target edge node device 4 includes: a second memory 41 and a second processor 42;

The second memory 41 stores a computer program executable on the second processor 42, which when executed by the second processor 42 implements the deployment method of the edge cloud platform virtual machine as shown in fig. 5 and 6.

In the embodiment of the present invention, if the above-mentioned deployment method of the edge cloud platform virtual machine is implemented in the form of a software function module, and sold or used as an independent product, the deployment method may also be stored in a computer readable storage medium. Based on such understanding, the technical solution of the embodiments of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium, including several instructions for causing an instant messaging device (which may be a terminal, a server, etc.) to perform all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read Only Memory (ROM), a magnetic disk, an optical disk, or other various media capable of storing program codes. Thus, embodiments of the invention are not limited to any specific combination of hardware and software.

The embodiment of the invention provides a readable storage medium, wherein a program of a deployment method of an edge cloud platform virtual machine is stored on the readable storage medium, and the steps of the deployment method of the edge cloud platform virtual machine are realized when the program backed up by the virtual machine is executed by a processor.

The above description of the instant messaging device and storage medium embodiments is similar to the description of the method embodiments described above, with similar advantageous effects as the method embodiments. For technical details not disclosed in the instant messaging device and the storage medium embodiments of the present invention, please refer to the description of the method embodiments of the present invention for understanding.

In some embodiments of the invention, the storage medium may be FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; but may be a variety of devices including one or any combination of the above memories.

It should be appreciated that reference throughout this specification to "some embodiments" or "other embodiments" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, the appearances of the phrase "in some embodiments" or "in other embodiments" in various places throughout this specification are not necessarily referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. It should be understood that, in various embodiments of the present invention, the sequence numbers of the foregoing processes do not mean the order of execution, and the order of execution of the processes should be determined by the functions and internal logic thereof, and should not constitute any limitation on the implementation process of the embodiments of the present invention. The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above described device embodiments are only illustrative, e.g. the division of the units is only one logical function division, and there may be other divisions in practice, such as: multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. In addition, the various components shown or discussed may be coupled or directly coupled or communicatively coupled to each other via some interface, whether indirectly coupled or communicatively coupled to devices or units, whether electrically, mechanically, or otherwise.

The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units; can be located in one place or distributed to a plurality of network units; some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated in one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated in one unit; the integrated units may be implemented in hardware or in hardware plus software functional units.

The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. The deployment method of the virtual machine of the edge cloud platform is applied to an edge center node and is characterized by comprising the following steps:

sending a GPU resource allocation request to the target edge node;

2. The method for deploying an edge cloud platform virtual machine according to claim 1, wherein the resource information further comprises: CPU and memory information of the central processing unit corresponding to each edge node;

the determining a target edge node from the at least one edge node according to the resource information of the at least one edge node comprises:

determining at least one candidate edge node based on GPU resource information of the at least one edge node;

And determining the target edge node according to the CPU and memory information of the at least one candidate edge node.

3. The method for deploying an edge cloud platform virtual machine according to claim 2, wherein the GPU resource information comprises: the number of virtualized graphics processor vGPU instances corresponding to each edge node;

the determining candidate edge nodes based on the GPU resource information of the at least one edge node includes:

and if the edge nodes with the number of the corresponding vGPU instances being greater than zero exist in the at least one edge node, determining the edge node as the at least one candidate edge node.

4. The method for deploying an edge cloud platform virtual machine of claim 3 wherein,

the target GPU resource information indicates any vGPU instance included in the GPU resource information of the target edge node.

5. The method of deploying an edge cloud platform virtual machine of claim 1, wherein prior to receiving a virtual machine deployment request, the method further comprises:

and receiving the resource information of at least one edge node, and storing the resource information of the at least one edge node into the storage module.

6. The method for deploying an edge cloud platform virtual machine according to any one of claims 1 to 4, wherein after receiving the target virtual machine and target GPU resource information fed back by the target edge node based on the GPU resource allocation request, the method further comprises:

and updating the number of available vGPU instances in the GPU resource information of the target edge node in the storage module.

7. The method of deploying an edge cloud platform virtual machine of claim 1, wherein after sending a virtual machine deployment instruction to the target edge node, the method further comprises:

and receiving a virtual machine deployment result.

8. The deployment method of the edge cloud platform virtual machine is applied to a target edge node and is characterized by comprising the following steps:

9. The method for deploying an edge cloud platform virtual machine according to claim 8, wherein the resource information further comprises: a Central Processing Unit (CPU) and memory information;

the sending the resource information of the target edge node to the edge center node includes:

and sending the GPU resource information, the CPU and the memory information of the target edge node to the edge center node.

10. The method for deploying an edge cloud platform virtual machine according to claim 9, wherein the GPU resource information of the target edge node comprises: at least one virtualized graphics processor vGPU instance created from GPU resources of the target edge node.

11. The method for deploying an edge cloud platform virtual machine according to any one of claims 8 to 10, wherein the deploying a target virtual machine according to the virtual machine deployment instruction comprises:

and binding the target GPU resource information with the target virtual machine according to the virtual machine deployment instruction so as to realize the deployment of the target virtual machine.

12. The method of claim 11, wherein the target GPU resource information indicates any vGPU instance included in GPU resource information of the target edge node.

13. The method for deploying an edge cloud platform virtual machine according to claim 8 or 11, wherein after the deploying a target virtual machine according to the virtual machine deployment instruction, the method further comprises:

obtaining a virtual machine deployment result;

and sending the virtual machine deployment result to the edge center node.

14. An edge center node apparatus, comprising:

15. A target edge node apparatus, comprising:

16. An edge center node device, comprising: a first memory and a first processor;

the first memory stores a computer program executable on the first processor, which when executed, implements the method of any one of claims 1 to 7.

17. A target edge node device, comprising: a second memory and a second processor;

The second memory stores a computer program executable on the second processor, which when executed, implements the method of any of claims 8 to 13.

18. A computer readable storage medium, characterized in that a computer program is stored, which when executed by a first processor implements the method of any of claims 1 to 7, or which when executed by a second processor implements the method of any of claims 8 to 13.