Regional computing power management platform system and regional computing power resource scheduling method

Abstract

This invention relates to a regional computing power management platform system and a regional computing power resource scheduling method. The system includes a user interface, a front-end module, a back-end module, a control node, a group of worker nodes, an application instance unit, and external resources, including an image repository and a storage server. A mechanism is introduced where the platform agent collaborates with the control node to achieve automatic registration and performance evaluation of new nodes. Compared to the traditional process of adding new servers that requires manual configuration, the platform can intelligently and quickly incorporate heterogeneous computing power into a unified schedule. The control node acts as a global resource scheduling engine, dynamically allocating workloads based on task requirements and the status of each worker node to maximize resource utilization and optimize performance.

Claims

1. A regional computing power management platform system, comprising: User interface; A front-end module that communicates with the user interface; A backend module that communicates with the frontend module; The system comprises: a control node, communicatively connected to the backend module; a group of worker nodes, communicatively connected to the control node; an application instance unit, communicatively connected to the group of worker nodes; and an external resource, communicatively connected to the application instance unit, the external resource including an image repository and a storage server, the image repository communicatively connected to the user interface and the backend module, and the storage server communicatively connected to the application instance unit; wherein the group of worker nodes has multiple worker nodes, each worker node being a physical server with computing power; and wherein the group of worker nodes can accommodate at least one new worker node, the at least one new worker node having a platform agent installed, the platform agent being configured to perform a performance benchmark, install a cluster agent, and register hardware information and a capability metric with the control node and the backend module.

2. The regional computing power management platform system as described in claim 1, wherein, This control node is a control module.

3. The regional computing power management platform system as described in claim 1, wherein, These worker nodes are permanently hosted by a cluster agent.

4. A method for scheduling regional computing resources, comprising: Request receiving steps: Receive a request through a front-end module; A request type determination step: receiving the request from the front-end module through a back-end module to determine the request type; and a scenario execution step: executing a usage scenario according to the request classification through an application instance unit; wherein, the usage scenario corresponding to the request type is new worker node registration and management, including: installing a platform agent on a new worker node, performing hardware performance benchmark tests and environmental detection on the new worker node to evaluate the computing power of the new worker node, installing a cluster agent on the new worker node, connecting the new worker node to a control node, an administrator reviewing whether the hardware configuration and performance indicators of the new worker node meet a standard, and if the review is passed because it meets the standard, approving the new worker node to join a cluster; if the review is rejected because it does not meet the standard, the registration fails; wherein, the standard includes at least an enterprise policy; wherein, the usage scenario corresponding to the request type includes one of the following: The use case corresponding to this request type is for a user to deploy an application instance, including: allowing the user to select the application to be deployed and one of the target worker nodes through the front-end module; generating a deployment script through the back-end module and sending it to the control node; instructing one of the worker nodes through the control node to automatically pull an application image and complete the deployment; and generating a dynamic URL through the back-end module after deployment for the user to access an application instance unit. The use case corresponding to this request type is user operation of the storage server, including: the user connects to a storage server through the front-end module and browses a file list. After selecting the desired data, the front-end module sends a download request to the back-end module, which reads a file from the storage server and directly transmits it to an application instance unit. The use case corresponding to this request type is user operation of the storage server, and also includes: during reverse upload, the front-end module first queries the back-end module for the file list in the application instance unit. After the user selects a file to upload, the back-end module is responsible for reading data from the application instance unit and uploading the read data to the storage server. The use case corresponding to this request type is administrator monitoring and authorization, including: an administrator logs in through the front-end module to query the status of each worker node in the cluster and the multiple application instances deployed on it within the application instance unit; collects the real-time resource usage of each worker node through the control node and presents it through the front-end module; and the use case corresponding to this request type is pushing and scanning images, including: the user pushes an application image to an image repository; the image repository automatically performs a security vulnerability scan and reports a scan result along with image information to the back-end module; the back-end module notifies the administrator to review it, and the administrator reviews whether to publicly disclose the image information for use or deployment by other users according to the standard.

Images