CN109688222B - Shared computing resource scheduling method, shared computing system, server and storage medium - Google Patents

Abstract

The invention discloses a scheduling method for shared computing resources. The method includes: acquiring a shared computing task to be executed; acquiring a list of all candidate shared computing nodes; selecting a shared computing task from the shared computing node list matching shared computing nodes; and delivering the shared computing tasks to the shared computing nodes matching the shared computing tasks. The present invention also provides a shared computing system, server and storage medium. The present invention can select an appropriate shared computing node according to the resource requirements of the user, and make corresponding scheduling in real time in response to the fluctuation of the node.

Description

Scheduling method for shared computing resources, shared computing system, server and storage medium

technical field

The present invention relates to the technical field of shared computing, and in particular, to a scheduling method for shared computing resources, a shared computing system, a server and a storage medium.

Background technique

At present, many enterprises need to use a large amount of bandwidth, disk, and CPU resources to provide stable and high-speed services for users distributed in different regions and different network environments. At the same time, the bandwidth and storage resources of the home environment are largely idle. The intelligent hardware in the home acts as a home node, and building a shared computing system can make full use of these resources and greatly reduce the service cost of enterprises. Home nodes have the following characteristics: 1. There are a large number of them, possibly as many as 100,000, millions or even higher orders of magnitude; 2. The stability of home nodes is lower than that of server nodes; 3. The nodes are interconnected by the public network, and the IP addresses of the nodes It is dynamic; 4. The physical resources owned by a single node are few and fluctuate in real time.

In the above mode, flexible and efficient management of the resources collected by intelligent hardware is the core point. It requires the rapid deployment of different business programs, resource management and security control for business programs, and the resource management of each node according to the business. The usage is scheduled in real time to maximize the use of the physical resources of the node. There is no mature solution in the industry to abstract virtual computing, storage, and network resources for more than one million nodes deployed in the home network environment.

SUMMARY OF THE INVENTION

In view of this, the present invention proposes a scheduling method for sharing computing resources, a sharing computing system, a server and a storage medium to solve at least one of the above technical problems.

First, in order to achieve the above purpose, the present invention provides a scheduling method for sharing computing resources, characterized in that the method includes:

Obtain shared computing tasks to be executed;

Get a list of all alternative shared computing nodes;

Selecting a shared computing node matching the shared computing task from the shared computing node list;

Delivering the shared computing task to the shared computing node matching the shared computing task.

Optionally, the shared computing node list includes IDs and available resource data of each shared computing node;

The shared computing tasks include requirements for shared computing resources that need to be configured;

The selecting a shared computing node matching the shared computing task from the shared computing node list includes:

According to the requirements of the shared computing resources to be configured and the available resource data of each shared computing node, a shared computing node matching the shared computing task is selected from the shared computing node list.

Optionally, the requirements for sharing computing resources include at least one of bandwidth requirements, storage space requirements, and computing resource requirements.

Optionally, the available resource data in the shared computing node list is calculated according to the real-time status of the node, the task status uploaded by each shared computing node, and the data generated when the task is executed on the node.

Optionally, selecting a shared computing node matching the shared computing task from the shared computing node list according to the requirements of the shared computing resources to be configured and available resource data of each shared computing node includes:

obtaining available resource data of each shared computing node in the shared computing node list;

Selecting a shared computing node whose available resource data reaches a preset value from the shared computing node list to generate a list of available nodes;

Score each shared computing node in the available node list according to the preset index, and use the binning algorithm to divide the requirements of the shared computing resources that need to be configured to the shared computing nodes whose scoring value exceeds the preset threshold to obtain the final A list of matching nodes.

Optionally, selecting a shared computing node matching the shared computing task from the shared computing node list according to the requirements of the shared computing resources to be configured and available resource data of each shared computing node further includes:

Obtaining the current available resource data of the selected shared computing node at regular intervals;

Whether it is necessary to add or delete nodes is determined according to the requirements of the shared computing resources to be configured and the current available resource data of the shared computing nodes.

Optionally, the preset indicators include regional resource surplus and historical stability.

Optionally, the obtaining the shared computing task to be executed includes: obtaining a docker image generated according to the shared computing task to be executed.

Optionally, the delivering the shared computing task to the shared computing node matching the shared computing task includes: delivering a docker image corresponding to the shared computing task to the shared computing node matching the shared computing task. Shared computing nodes matched by shared computing tasks.

In addition, in order to achieve the above object, the present invention also provides a server, the server includes a memory and a processor, the memory stores a scheduler for sharing computing resources that can run on the processor, and the shared computing When the resource scheduler is executed by the processor, the above-mentioned scheduling method for shared computing resources is implemented.

Further, in order to achieve the above object, the present invention also provides a shared computing system, the system includes:

a task management unit, configured to receive the shared computing task to be executed from the client, and dispatch the shared computing task to the scheduling service unit;

The scheduling service unit is configured to obtain the shared computing task from the task management unit, obtain a list of all alternative shared computing nodes according to the status and historical data of each shared computing node provided by the node management unit and the data warehouse, and Selecting a shared computing node matching the shared computing task from the shared computing node list;

A deployment service unit, configured to deliver the shared computing task to a shared computing node selected by the scheduling service unit that matches the shared computing task.

Further, in order to achieve the above object, the present invention also provides a storage medium, where the storage medium stores a scheduler for sharing computing resources, and the scheduler for sharing computing resources can be executed by at least one processor, so that the At least one processor executes the scheduling method for sharing computing resources as described above.

The scheduling method for shared computing resources, the shared computing system, the server and the storage medium proposed by the present invention can uniformly manage a Docker cluster composed of millions of shared computing nodes, and allocate resources required for a shared computing task to match the task. shared computing nodes, and schedule nodes at any time according to changes in node status to maintain the stability of the total amount of resources.

Description of drawings

1 is a schematic diagram of the architecture of a shared computing system proposed by a first embodiment of the present invention;

2 is a schematic diagram of the architecture of a scheduling server proposed by a second embodiment of the present invention;

3 is a schematic flowchart of a scheduling method for sharing computing resources proposed by a third embodiment of the present invention;

FIG. 4 is a schematic diagram of the refinement flow of S24 in FIG. 3 .

The realization, functional characteristics and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

It should be noted that the descriptions involving "first", "second", etc. in the present invention are only for the purpose of description, and should not be construed as indicating or implying their relative importance or implying the number of indicated technical features . Thus, a feature delimited with "first", "second" may expressly or implicitly include at least one of that feature. In addition, the technical solutions between the various embodiments can be combined with each other, but must be based on the realization by those of ordinary skill in the art. When the combination of technical solutions is contradictory or cannot be realized, it should be considered that the combination of such technical solutions does not exist. , is not within the scope of protection required by the present invention.

first embodiment

Referring to FIG. 1 , a first embodiment of the present invention provides a shared computing system. The above-mentioned shared computing system is a set of IaaS (Infrastructure as a Service) system constructed by using distributed node resources. It carries the program logic of the user, and makes corresponding scheduling and adjustment in real time in response to fluctuations in the node's network location, bandwidth, and storage.

In this embodiment, the shared computing system 1 includes a server 10 and a shared computing node 19 . The above server 10 includes a task management unit 11 , a scheduling service unit 12 , a node management unit 13 , a data warehouse 14 , a deployment service unit 15 and an image warehouse 17 . The above-mentioned shared computing system 1 and the client 2 perform data communication through the network, so as to allocate the corresponding shared computing node 19 according to the shared computing task initiated by the client 2 to execute the shared computing task.

The client 2 is used to select the specification and capacity of the required resources and the program logic to be executed, automatically generate a Docker (application container engine) image according to the above program logic, and encapsulate the selected required resources into standardized shared computing tasks. In this embodiment, the user can select the desired value through a management console, a CLI (Command-line Interface, command line interface) tool, an API (Application Programming Interface, application programming interface) interface invocation and other methods on the client 2. The specification and capacity of the required resources (such as bandwidth, storage, etc.), select the program logic to be executed (which can be implemented in multiple languages), and after processing through the debugging platform and cross-compilation platform, the program logic will automatically generate a Docker image. For example, the resource requirement is 100Gbps bandwidth, 10PB storage, and the logic code to execute is hello.py. At the same time, the user of the client 2 can also control the above program logic, such as starting, stopping, adding, deleting, and so on. After the client 2 encapsulates the standardized task for the selected required resource, it submits the task to the task management unit 11 . The program logic selected by the user on the client 2 will be encapsulated into a standardized Docker image, shielding differences in programming languages and execution environments, and then submitted to the image repository 17 .

The task management unit 11 is configured to dispatch the task to the scheduling service unit 12 after receiving the above task from the client 2 . In this embodiment, the task management unit 11 will arrange the received tasks into a plurality of parallelized pipelines according to the priority and the degree of relevance, and the scheduling service unit 12 obtains tasks from the pipelines in sequence.

The scheduling service unit 12 is configured to obtain tasks from the task management unit 11 , and select a shared computing node 19 matching the shared computing task according to the status and historical data of each shared computing node 19 provided by the node management unit 13 and the data warehouse 14 . The selection of nodes by the scheduling service unit 12 needs to rely on the real-time status of all nodes obtained from the node management unit 13 and the historical data of nodes and tasks obtained from the data warehouse 14 (eg, historical stability of nodes, etc.). For example, the scheduling service unit 12 first obtains a list of all current candidates for shared computing nodes, and the above-mentioned shared computing node list includes the ID and available resource data of each shared computing node 19. The above-mentioned available resource data can be uploaded according to each shared computing node 19. The real-time state of the node, the task state and the data generated when the task is executed on the node are calculated. Then, the scheduling service unit 12 divides the resource requirements of the task, and selects them according to the region, ISP (Internet Service Provider, Internet Service Provider), NAT (Network Address Translation, network address translation) type, bandwidth, storage space, computing resources, etc. List of available nodes up to the preset value. Finally, each shared computing node in the available node list is scored according to preset indicators such as regional resource reserve and historical stability. According to the resource cost, the binning algorithm is used to split the configuration required for the task according to the principle of maximizing resource utilization. The demand for shared computing resources reaches the shared computing node 19 whose scoring value exceeds the preset threshold, and the final matching node list is selected. In addition, after the selected shared computing node 19 uploads the node real-time state and task state (thus obtaining the currently available resource data), the scheduling service unit 12 is further configured to further determine whether to add or delete nodes.

The node management unit 13 is configured to receive the node real-time status and task status uploaded by each shared computing node 19 and provide them to the scheduling service unit 12 for scheduling.

The data warehouse 14 is configured to receive the data generated when executing tasks uploaded by each shared computing node 19 and provide the data to the scheduling service unit 12 for scheduling.

The deployment service unit 15 is configured to deliver the deployment task to the shared computing node 19 selected by the scheduling service unit 12 .

The image repository 17 is used to receive the Docker image generated by the client 2 and provide the Docker image to the shared computing node 19 .

The shared computing node 19 is used to receive and execute the task deployed by the deployment service unit 15, download the corresponding Docker image from the image repository 17, start the image instance, and upload the node real-time status, task status and data generated on the node. In this embodiment, the shared computing node 19 downloads the Docker image from the mirror repository 17 . In other embodiments, the Docker image downloaded by other shared computing nodes 19 may be obtained through P2P transmission between the shared computing nodes 19 . After the above Docker image has been downloaded, the Docker image can also be transferred to other shared computing nodes through P2P.

Further, the above-mentioned shared computing system 1 also includes:

The signaling gateway 16 is configured to deliver the task deployed by the deployment service unit 15 to the corresponding shared computing node 19 , and receive the node real-time status and task status uploaded by the shared computing node 19 , and send them to the node management unit 13 .

The data gateway 18 is configured to transmit the Docker image to the shared computing node 19 , and receive the data generated during the execution of the Docker instance uploaded by the shared computing node 19 , and upload the data to the data warehouse 14 .

The transmission of the above signaling and data is dynamically accelerated by a content delivery network (Content Delivery Network, CDN).

Further, the shared computing node 19 includes a local signaling broker 190 , a local data broker 192 and a Docker manager 194 . Through the local signaling agent 190, the local data agent 192 and the Docker manager 194 deployed on each shared computing node 19, the node resources are virtualized and managed, and the node and task status are collected in real time, and The data.

Local signaling agent 190 is used to receive signaling (eg, deployed tasks) from signaling gateway 16 , parse the signaling, pass to Docker manager 194 , and upload node real-time status and task status to signaling gateway 16 . The Docker manager 194 is used to download the Docker image according to the task received by the local signaling agent 190, load and start the image instance. The local data agent 192 is used to receive the Docker image downloaded from the image warehouse 17 from the data gateway 18 or obtain the Docker image from other shared computing nodes 19 through P2P transmission, and upload the data generated during the execution of the Docker instance, for example, during the execution of the Docker instance. The generated results, logs, core dumps (Coredump), etc., the above-mentioned data can be subsequently used as the historical data of the node as a reference when the scheduling service unit 12 performs scheduling. After some of the shared computing nodes 19 have downloaded the Docker image, P2P diffusion can be performed through the local data proxy 192 to reduce the download bandwidth pressure of the data gateway 18 .

The shared computing system 1 provided by this embodiment can use Docker to perform lightweight virtualization on resource-constrained home intelligent hardware, and uniformly manage a Docker cluster composed of millions of public network nodes. cluster management and fault tolerance. The transmission of signaling and data is dynamically accelerated by the CDN network, and the Docker image is diffused and distributed through the P2P method, which improves the distribution efficiency and saves the bandwidth of the server. The Docker image instance carried by the shared computing node 19 is in the public network environment, and the NAT type, operator, and region of the node will change dynamically. The scheduling service unit 12 constantly increases and decreases the node through the packing algorithm, which can maintain the stability of the total amount of resources.

Second Embodiment

Referring to FIG. 2 , a second embodiment of the present invention provides a server 10 .

The above server 10 includes: a memory 21 , a processor 23 , a network interface 25 and a communication bus 27 . Wherein, the network interface 25 may optionally include a standard wired interface and a wireless interface (eg, a WI-FI interface). The communication bus 27 is used to realize the connection communication between these components.

The memory 21 includes at least one type of readable storage medium. The above-mentioned at least one type of readable storage medium may be a non-volatile storage medium such as a flash memory, a hard disk, a multimedia card, a card-type memory, or the like. In some embodiments, the above-mentioned memory 21 may be an internal storage unit of the server 10 , such as a hard disk of the server 10 . In other embodiments, the above-mentioned memory 21 may also be an external storage unit of the server 10, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) equipped on the server 10 card, flash card (Flash Card) and so on.

The above-mentioned memory 21 may be used to store application software installed on the server 10 and various types of data, such as the program code of the scheduler 20 sharing computing resources and related data generated during the running process thereof.

The processor 23 may be a central processing unit, a microprocessor or other data processing chip in some embodiments, and is used to execute program codes or process data stored in the memory 21 .

Figure 2 only shows server 10 with components 21-27 and scheduler 20 sharing computing resources, but it should be understood that Figure 2 does not show all components of server 10, and more or less may be implemented instead. components.

In the embodiment of the server 10 shown in FIG. 2 , the program code of the scheduler 20 for sharing computing resources is stored in the memory 21 as a computer storage medium. When the processor 23 executes the program code of the scheduler 20 for sharing computing resources, , implement the following methods:

(1) Obtain the shared computing task to be executed.

(2) Obtain a list of all candidate shared computing nodes.

(3) Select a shared computing node 19 matching the shared computing task from the shared computing node list.

(4) Delivering the shared computing task to the shared computing node 19 matching the shared computing task.

For the detailed description of the above method, please refer to the following third embodiment, which will not be repeated here.

Third Embodiment

Referring to FIG. 3 , a third embodiment of the present invention proposes a scheduling method for sharing computing resources, which is applied to the above server 10 . In this embodiment, according to different requirements, the execution order of the steps in the flowchart shown in FIG. 3 can be changed, and some steps can be omitted. The method includes:

S20: Acquire a shared computing task to be executed.

In this embodiment, the above-mentioned shared computing tasks include requirements for shared computing resources that need to be configured. The above requirements for shared computing resources include at least one of bandwidth requirements, storage space requirements, and computing resource requirements. After the user selects the specification and capacity of the required resources and the program logic to be executed on the client 2, the client 2 automatically generates a Docker image according to the above program logic, and encapsulates the selected required resources into standardized tasks. Then, the client 2 submits the task to the task management unit 11 , and submits the Docker image to the image repository 17 . The task management unit 11 will arrange the received tasks into a plurality of parallelized pipelines according to the priority and the degree of association, and the scheduling service unit 12 obtains tasks from the pipelines in sequence.

S22: Obtain a list of all candidate shared computing nodes.

In this embodiment, the above-mentioned shared computing node list includes the IDs and available resource data of each shared computing node 19 , and the above-mentioned available resource data can be generated according to the real-time status and task status of the nodes uploaded by each shared computing node 19 and when tasks are executed on the nodes. data is calculated. The node management unit 13 receives the node real-time status and task status uploaded by each shared computing node 19 and provides them to the scheduling service unit 12 for scheduling. The data warehouse 14 receives the generated data uploaded by each shared computing node 19 and provides it to the scheduling service unit 12 for scheduling. The selection of nodes by the scheduling service unit 12 needs to rely on the real-time status of all nodes obtained from the node management unit 13 and the historical data of nodes and tasks obtained from the data warehouse 14 (eg, historical stability of nodes, etc.).

S24: Select a shared computing node 19 matching the shared computing task from the shared computing node list.

The scheduling service unit 12 selects a shared computing node 19 matching the shared computing task from the shared computing node list according to the shared computing resource requirements to be configured and the available resource data of each shared computing node 19 . For example, the scheduling service unit 12 first obtains a list of all current candidate shared computing nodes, then divides the resource requirements of the task, and selects a preset value according to region, ISP, NAT type, bandwidth, storage space, computing resources, etc. Finally, according to preset indicators such as regional resource surplus and historical stability, each shared computing node in the available node list is scored 19. According to the resource cost, the binning algorithm is used to split according to the principle of maximizing resource utilization. The shared computing resources that need to be configured for the task are required to reach the shared computing nodes 19 whose scoring values exceed the preset threshold, and the final matching node list is selected. In addition, after the selected shared computing node 19 uploads the node real-time state and task state (thus obtaining the currently available resource data), the scheduling service unit 12 is further configured to further determine whether to add or delete nodes.

Referring to FIG. 3 , it is a schematic diagram of the refinement flow of the above S24. The refinement process includes:

S240: Obtain available resource data of each shared computing node 19 in the shared computing node list.

S242, select a shared computing node 19 whose available resource data reaches a preset value from the shared computing node list, and generate an available node list.

S244, score each shared computing node 19 in the available node list according to the preset index, and use a binning algorithm to divide the shared computing resource requirements that need to be configured for the task to the shared computing node 19 whose scoring value exceeds a preset threshold, Get the final list of matching nodes.

S246, obtain the current available resource data of the selected shared computing node 19 at regular intervals.

S248, according to the requirement of the shared computing resource and the current available resource data of the shared computing node 19, determine whether it is necessary to add or delete nodes. For example, nodes may need to be added or deleted when the online and offline status of nodes changes, NAT type or operator changes, disk storage changes, and task load changes.

S26: Deliver the shared computing task to the shared computing node 19 matching the shared computing task.

After the scheduling service unit 12 selects the shared computing nodes 19 , the above tasks obtained from the task management unit 11 can be allocated to each of the selected shared computing nodes 19 , and then the deployment service unit 15 assigns each selected shared computing node 19 The tasks assigned to the nodes 19 are delivered to the corresponding shared computing nodes 19 .

The shared computing node 19 receives and executes the assigned task, downloads the corresponding Docker image from the image repository 17, starts the image instance, and uploads the real-time status of the node, the status of the task and the data generated on the node.

The scheduling method for shared computing resources provided in this embodiment can use Docker to perform lightweight virtualization on resource-constrained home intelligent hardware, and uniformly manage a Docker cluster composed of millions of public network nodes. Operator cluster management and fault tolerance. The Docker image instance carried by the shared computing node 19 is in the public network environment, and the NAT type, operator, and region of the node will change dynamically. The scheduling service unit 12 constantly increases and decreases the node through the packing algorithm, which can maintain the stability of the total amount of resources.

Fourth Embodiment

Another embodiment of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a scheduler 20 for sharing computing resources, and the scheduler 20 for sharing computing resources can be processed by at least one The above-mentioned at least one processor executes the above-mentioned scheduling method for sharing computing resources.

From the description of the above embodiments, those skilled in the art can clearly understand that the method of the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is better implementation. Based on this understanding, the technical solutions of the present invention can be embodied in the form of software products in essence or the parts that make contributions to the prior art, and the computer software products are stored in a storage medium (such as ROM/RAM, magnetic disk, CD-ROM), including several instructions to make a client (which may be a mobile phone, a computer, an electronic device, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of the present invention.

The above are only preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields , are similarly included in the scope of patent protection of the present invention.

Claims (8)

1. A scheduling method for shared computing resources, wherein the method comprises: Obtain the Docker image generated according to the shared computing task to be executed from the image repository; Get a list of all alternative shared computing nodes; Selecting a shared computing node matching the shared computing task from the shared computing node list includes: selecting a shared computing node whose available resource data reaches a preset value from the shared computing node list to generate a list of available nodes, and generating a list of available nodes according to the preset value. Set the index to score each shared computing node in the available node list, and use the binning algorithm to split the shared computing resource requirements that need to be configured for the shared computing task to the shared computing node whose scoring value exceeds the preset threshold to obtain a matching node list. , periodically obtain the current available resource data of the shared computing nodes in the matching node list, and determine whether it is necessary to perform node additions or deletions on the shared computing nodes in the matching node list according to the requirements of the shared computing resources and the available resource data operation, when it is determined that an addition or deletion operation needs to be performed, an addition or deletion operation is performed on the shared computing nodes in the matching node list; The Docker image is transmitted to the matching shared computing node by means of CDN dynamic acceleration, and the real-time node status and task status returned by the matching shared computing node after downloading the Docker image, loading and starting the image instance are received. and data generated on the node, wherein: after any of the matching shared computing nodes completes the download of the Docker image, the downloaded Docker image is spread to other matching shared computing nodes through P2P through a local data agent. 2 . The method for scheduling shared computing resources according to claim 1 , wherein the shared computing node list includes IDs and available resource data of each shared computing node. 3 . 3. The scheduling method for shared computing resources according to claim 1 or 2, wherein the requirements for the shared computing resources include at least one of: bandwidth requirements, storage space requirements, and computing resource requirements. 4. The scheduling method for shared computing resources according to claim 1 or 2, wherein the available resource data in the shared computing node list is based on the real-time status of the node, the task status and the on-node status uploaded by each shared computing node. The data generated when the task is executed is calculated. 5 . The scheduling method for shared computing resources according to claim 1 , wherein the preset indicators include regional resource surplus and historical stability. 6 . 6. A server, characterized in that the server comprises a memory and a processor, and the memory stores a scheduler of shared computing resources that can run on the processor, and the scheduler of the shared computing resources is The processor implements the method according to any one of claims 1-5 when executed. 7. A shared computing system, wherein the system comprises: A task management unit, configured to receive the Docker image generated by the client according to the shared computing task to be executed from the image warehouse, and distribute the Docker image to the scheduling service unit; The scheduling service unit is used to obtain the Docker image from the task management unit, obtain a list of all alternative shared computing nodes according to the status and historical data of each shared computing node provided by the node management unit and the data warehouse, and from Selecting a shared computing node matching the shared computing task in the shared computing node list includes: selecting a shared computing node whose available resource data reaches a preset value from the shared computing node list to generate a list of available nodes, according to a preset The indicator scores each shared computing node in the available node list, and uses the binning algorithm to split the shared computing resource requirements that need to be configured for the shared computing task to the shared computing node whose score exceeds a preset threshold to obtain a matching node list. Periodically obtain the current available resource data of the shared computing nodes in the matching node list, and determine whether it is necessary to perform node addition or deletion operations on the shared computing nodes in the matching node list according to the requirements of the shared computing resources and the available resource data , when it is determined that an addition or deletion operation needs to be performed, an addition or deletion operation is performed on the shared computing nodes in the matching node list; A deployment service unit, configured to deliver the Docker image to the matched shared computing node by means of CDN dynamic acceleration; The node management unit is configured to receive the node real-time status, task status and data generated on the node returned after the matched shared computing node has downloaded the Docker image, loaded and started the image instance; The data warehouse is used to receive data generated on each shared computing node; Wherein: after any of the matching shared computing nodes completes the downloading of the Docker image, the downloaded Docker image is spread to other matching shared computing nodes through P2P through the local data agent. 8. A storage medium storing a scheduler for sharing computing resources, the scheduler for sharing computing resources being executable by at least one processor, so that the at least one processor executes the process according to claim 1- The scheduling method for shared computing resources according to any one of 5.

Images