Virtual workstation application method based on computing power scheduling, electronic equipment and storage medium

Abstract

The invention relates to the technical field of cloud processing, in particular to a virtual workstation application method based on computing power scheduling, electronic equipment and a storage medium. An application end applies for a virtual workstation through a computing power scheduling platform; the platform dispatches a pre-created workstation; the cloud management platform prepares an application end data disk and a game disk (based on the latest Ceph snapshot); mounting a disk; starting a game server, and pushing a game stream through WebRTC; and closing and destroying the workstation after the application end exits. Compared with a traditional physical server mode, the hardware cost is remarkably reduced, and the method is suitable for old-using scenes; a ceph hierarchical storage framework is adopted, the newest snapshot is updated when the cloud computer game disk is created, it is ensured that the newest cloud game version can be obtained every time an application end applies for, and meanwhile storage space is saved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cloud processing, and in particular to a virtual workstation application method based on computing power scheduling, an electronic device and a storage medium. BACKGROUND

[0002] With the rapid development of cloud computing technology, cloud gaming, remote desktop, high-performance computing and other application scenarios are increasingly popular. The core of these scenarios is to migrate the computing tasks (such as graphics rendering, data processing) originally running on the user's local terminal to the virtualized computing instances in the cloud for execution, and then transmit the results stream (such as video stream) to the user. This mode enables users to enjoy high-quality computing services without the need for high-performance local hardware.

[0003] Currently, the mainstream technical solution to achieve the above services is usually based on large cloud computing platforms (such as IaaS platforms based on OpenStack) or container scheduling platforms (such as Kubernetes). At the resource provision level, there are mainly two ways: Physical server-based solution: directly allocate independent physical servers or GPU cards to users. Although this solution can provide extreme performance, it has high hardware costs, low resource utilization, lacks flexibility, and cannot achieve rapid reuse and efficient scheduling of resources, making it difficult to support concurrent access requirements of a large number of users.

[0004] Virtual machine-based solution: create virtual machines on physical servers to provide services to users. Although this approach improves resource utilization to some extent, it still faces many challenges when dealing with high real-time and high graphics performance requirements such as cloud gaming: Large resource scheduling delay: the traditional cloud platform needs to complete a whole set of processes such as virtual machine creation, system initialization, software installation, etc. after receiving the user's application, which takes a long time, resulting in long user waiting time and poor experience.

[0005] Storage cost and efficiency issues: to ensure that each user can access the latest version of games or applications, a complete and up-to-date copy of the game disk needs to be allocated to each virtual machine instance. This not only causes a huge waste of storage space, but also makes the game version update process tedious and slow, requiring batch image rebuilding and high maintenance costs.

[0006] Performance and cost balance: using expensive enterprise servers and storage arrays as underlying hardware, although the performance is reliable, the overall total cost of ownership (TCO) is too high. If a PC cluster with lower cost is used, there is a lack of a supporting management solution that can achieve efficient scheduling and resource pooling.

[0007] In addition, although existing storage technologies such as the distributed storage system Ceph are widely used in cloud environments, the potential of the snapshot and cloning functions thereof in a dynamic and multi-tenant cloud game scenario has not been fully tapped, and the functions cannot be deeply integrated with the computing power scheduling process to solve the contradiction between rapid updating of game versions and saving of storage space.

[0008] Therefore, there is an urgent need in the art for a cloud computing resource providing method that can balance low latency, low cost, high performance and easy maintenance, and in particular, a virtual workstation scheduling scheme that can quickly respond to applications, ensure that users obtain the latest application version, and significantly reduce storage and hardware costs. SUMMARY

[0009] In view of the deficiencies of the prior art, the present application discloses a virtual workstation application method based on computing power scheduling, an electronic device and a storage medium, which are used to solve the above problems.

[0010] The present application is implemented by the following technical solutions: In a first aspect, the present application provides a virtual workstation application method based on computing power scheduling, comprising the following steps: (S1) A user initiates a virtual workstation application request through a computing power scheduling platform, and the computing power scheduling platform schedules a pre-created virtual workstation from a buffer running pool of a cloud management platform to the user; (S2) The cloud management platform prepares a user data disk and a game disk for the virtual workstation, wherein the game disk is created based on the latest game snapshot in a Ceph cluster; (S3) The user data disk and the game disk are mounted to the virtual workstation; (S4) A virtual workstation game server is started, and a game stream is pushed to the user end through WebRTC technology; (S5) After the user exits, the virtual workstation is remotely closed and the game disk is destroyed.

[0011] Further, the step (S1) comprises: The user initiates a virtual workstation application on a visual interface of the computing power scheduling platform according to GPU type, frame rate, resolution or bandwidth requirements; The computing power scheduling platform is constructed based on KubeSphere, and the cloud management platform is constructed based on OpenStack.

[0012] Further, the virtual workstation in the buffer running pool is a Windows virtual machine pre-installed with game supporting software, which runs on a PC with a GPU and does not mount a storage disk.

[0013] Further, the step (S2) comprises: If the user has no data disk, a data disk is created for the user in the Ceph cluster; According to the latest game snapshot of the user in the machine room, an independent game disk is created for the user.

[0014] Further, the creation of the game disk adopts a Ceph hierarchical storage architecture, including a basic layer, an intermediate layer and a writable layer, the game snapshot is a read-only layer, and the user game disk is created based on the snapshot and the writable layer is incrementally stored.

[0015] Further, the step (S3) is implemented through a disk mounting agent, including: After the virtual workstation is started, the disk mounting agent detects that the allocation state is allocated, and obtains Ceph configuration and authentication information; Mount the user data disk and the game disk, and report the mounting state.

[0016] Further, the step (S4) includes: After the disk mounting is successful, the game server is called to apply an order, and the MAC address of the virtual workstation is input; After the order application is successful, the cloud game server is started and pushed to the user end through WebRTC.

[0017] Further, the step (S5) includes: After receiving the user exit instruction, the virtual workstation is remotely closed through the BMC interface; The virtual workstation is removed from the running pool, the data disk is released, and the game disk is destroyed.

[0018] In a second aspect, the application provides an electronic device, including a memory and a processor, the memory stores a computer program, and the processor implements the virtual workstation application method based on the computing power scheduling when executing the computer program.

[0019] In a third aspect, the application provides a storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the virtual workstation application method based on the computing power scheduling.

[0020] The application has the following advantages: The application pre-builds and maintains a virtual workstation buffer pool, so that the user does not need to go through the virtual machine creation and configuration process when applying, realizes a second-level response, improves the experience of real-time scenarios such as cloud games, replaces expensive enterprise-level servers with GPU ordinary PC machines, combines OpenStack platform resource pool management, reduces CAPEX and OPEX, supports old hardware remanufacturing, and meets the green computing concept.

[0021] The application adopts Ceph layered snapshot technology to create a game disk, shares a read-only base snapshot, and only writes user differentiated data, thereby greatly saving storage space and simplifying game version updating. Through deep integration of KubeSphere and OpenStack, automatic user mapping and strict resource isolation are realized, multi-tenant security requirements are met, and automatic full life cycle management of virtual workstations, GPU and other resources is realized. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0023] Figure 1 is a flowchart of the virtual workstation application method based on the computing power scheduling of the present application; Figure 2 is an architectural schematic diagram of the virtual workstation application method based on the computing power scheduling of the present application; Figure 3 is a specific execution flowchart of the virtual workstation application in the method of the present application; Figure 4 is a schematic diagram of the creation and mounting of the data disk and the game disk of the present application. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0025] In an embodiment, referring to Figure 1 , a virtual workstation application method based on computing power scheduling is provided, specifically including the following steps: S101: The user initiates a virtual workstation application request.

[0026] The user initiates an application request for a virtual workstation on the visual interface of the computing power scheduling platform (in this embodiment, the platform is built based on KubeSphere) according to the required GPU type (such as NVIDIA GeForce RTX 4070Ti), frame rate (such as 60FPS), resolution (such as 4K) and bandwidth and other parameters. The computing power scheduling platform provides a unified computing power market, resource management and account management interface for the user.

[0027] S102: Schedule a virtual workstation from the buffer running pool.

[0028] After receiving the user request, the computing power scheduling platform does not create a virtual machine in real time, but initiates an instruction to the cloud management platform (in this embodiment, the platform is built based on OpenStack) to schedule a virtual workstation that meets the specification requirements (such as RTX4070Ti GPU) and is in an idle state from its buffer running pool.

[0029] In further implementation of this embodiment, the buffer running pool is a resource pool pre-created and maintained by the cloud management platform, and the virtual workstations in the pool are virtual machine instances pre-installed with Windows operating system and game running library and other supporting software. These instances run on the underlying PC with GPU, but their system disks are small and are not mounted with any user data disk or game disk, and are in a “light ready” state, so they can be quickly scheduled.

[0030] The scheduling process is shown in detail in Figure 3 If successful, proceed to the next step; if failed, the user enters the waiting queue or is prompted to retry.

[0031] S103: Prepare a data disk and a game disk for the virtual workstation.

[0032] After the cloud management platform allocates the virtual workstation to the user, it immediately prepares exclusive storage resources for it. The specific process is shown in Figure 4 Data disk preparation: Check whether the user has a persistent data disk. If not, create a new block storage volume in the Ceph distributed storage cluster in the same machine room area as the virtual workstation, which serves as the user's personal data disk for saving personal configurations, archives and other private data.

[0033] ​Game disk preparation: To ensure that users have the latest game content, the cloud management platform queries the Ceph cluster for the latest game snapshot (such as `snapshot_game_20250821`) generated by the game content server. Based on this read-only snapshot, a game disk is quickly created for the user using Ceph's tiered storage technology. This game disk is a "clone volume" that shares data with the underlying snapshot and only consumes additional space when the user writes new data, greatly reducing storage costs and speeding up creation.

[0034] S104: Mount the data disk and the game disk.

[0035] After the virtual workstation is started, the pre-installed disk mounting agent inside it will automatically start. This agent queries the allocation status (`allocated`) every second. When the status is confirmed as `true` (allocated), the mounting operation is performed, and the specific process is as shown in Figure 3 1. Call the cloud management platform interface to obtain the access configuration (`ceph-conf`) and authentication key (`keyring`) of the data disk and game disk in the Ceph cluster for the corresponding user.

[0036] 2. Execute the mounting command to map the remote Ceph block device as a disk drive (such as mounting the data disk as D and the game disk as G) inside the virtual workstation.

[0037] 3. After successful mounting, the agent calls the interface to report `mount_status` as successful. If mounting fails, the process terminates and the user end displays a startup failure.

[0038] S105: Start the application service and push the stream.

[0039] After successful disk mounting, the system continues to perform the following steps: 1. Call the order application interface of the game server to pass in the MAC address and other unique identifiers of the virtual workstation to apply for starting the game service.

[0040] 2. After successful order application, start the cloud game server program inside the virtual workstation.

[0041] 3. The cloud game server uses the underlying GPU to perform real-time rendering and encoding of game pictures and pushes the video stream to the user's web browser through the low-latency transmission technology of WebRTC.

[0042] 4. After the user's browser receives the streaming media signal, it automatically enters the cloud computer streaming picture, and the user can start operating the game. At the same time, the web end starts reporting real-time streaming indicators (such as delay and frame rate). ​

[0043] S106: User exit and resource release.

[0044] When the user ends the use, clicks exit on the Web interface, the system executes the resource cleaning process: 1. The computing power scheduling platform calls the BMC (baseboard management controller) interface of the underlying PC and sends a shutdown signal to the virtual machine through the IPMI protocol, realizing remote shutdown.

[0045] 2. The virtual workstation is moved back from the "running pool" to the "buffer running pool", and its state is marked as "idle", waiting for the next scheduling.

[0046] 3. Destroy the game disc used by the user in this session. The game disc is a temporary resource and is destroyed with the end of the session, and its storage space is recycled to the Ceph cluster. The user's personal data disc is retained for next use.

[0047] Thus, a complete virtual workstation application, use and release cycle ends.

[0048] In an embodiment, referring to FIG. 2, the computing power scheduling platform is constructed based on KubeSphere and cloud management platform. After the user completes the registration, the following operations can be performed through the platform: applying for a cloud computer in the computing power market, initiating a computing power resource scheduling task, managing personal computing power resources and accounts. The user initiates a cloud computer application in the visual computing power market interface of the computing power scheduling platform according to the frame rate, resolution, bandwidth and underlying GPU type of the required cloud computer. The cloud computer mainly serves the user's game terminal, realizes the cloud game scene on the computing power scheduling platform, supports massive games, large-capacity game disc mounting, user data isolation, and ensures that the user accesses the latest version of the game.

[0049] Further, the KubeSphere is an enterprise-level distributed multi-tenant open source container platform based on the mainstream container orchestration platform Kubernetes. It provides an intuitive operation interface and guided operation, which significantly simplifies the complexity of daily development, testing and operation while reducing the user's learning curve and the threshold of using the container platform, aiming to solve the inherent challenges of Kubernetes in storage, network, security and ease of use. The cloud management platform is implemented based on OpenStack, relying on its powerful open source ecosystem, it can efficiently integrate computing, storage and network resources, support multi-tenant isolation, and provide virtual machine lifecycle management.

[0050] In this embodiment, in order to shorten the application time of cloud computer and improve the user experience of cloud game, the cloud management platform has pre-created a cloud computer instance in the buffer running pool before the user initiates the application. The instance is a Windows virtual machine pre-installed with game-related supporting software, and the pre-installed software is pre-configured through a system image. The number of cloud computer instances pre-created in the buffer running pool can be configured as needed. When the user initiates a cloud computer application in the computing power scheduling platform, the request is first scheduled to the cloud computer instance pre-created in the buffer running pool of the cloud management platform. The underlying cloud computer instance corresponds to a physical host (PC) equipped with a GPU acceleration card, and the physical host is not equipped with a local storage disk and cannot directly provide large-capacity game disk storage required by cloud game. A cloud computer instance exclusively occupies a physical host resource.

[0051] In an embodiment, after receiving the cloud computer application instruction stream sent by the computing power scheduling platform, the cloud management platform allocates the cloud computer instance in the buffer pool to the user. The platform is responsible for managing the full life cycle of the virtual machine corresponding to the cloud computer, and synchronously generates the system disk when creating the virtual machine. In the OpenStack-based cloud management platform architecture, each user is associated with an independent tenant, realizing a multi-tenant isolation mechanism. Correspondingly, one computing power scheduling platform user is mapped to one cloud management platform user. Specifically, when the user completes registration in the computing power scheduling platform, the system will automatically call the user creation interface of the cloud management platform to generate the corresponding user and tenant resources, thereby realizing multi-tenant isolation of computing power resources at the computing power scheduling platform level.

[0052] Further, while allocating the buffer pool cloud computer instance to the user, the cloud management platform needs to determine whether the user has configured a data disk and a game disk. If not, the following operations are performed: a data disk is created for the user in the Ceph cluster; at the same time, the Ceph cluster corresponding to the physical machine room where the cloud computer instance is located is located, and based on the latest cloud game disk snapshot, a dedicated game disk instance is created for the user. The Ceph cluster is a distributed storage system that provides high availability, scalability, and data redundancy capabilities by distributing data storage across multiple nodes. The system is widely used in cloud platforms such as OpenStack and Kubernetes as an underlying storage solution. In this solution, Ceph is deployed on underlying storage hardware and interfaces with the OpenStack platform deployed on the underlying PC server to provide data disk and game disk storage services for cloud computers.

[0053] In an embodiment, in view of the fact that each application of the cloud computer instance by the user needs to contain the latest version of the game software, and the version on the game content server is continuously updated, the scheme adopts the tiering technology of Ceph. This technology effectively ensures that each time the cloud computer instance obtained by the user can load the latest game version, while significantly saving the storage space overhead.

[0054] The embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the virtual workstation application method based on the computing power scheduling are implemented.

[0055] The embodiment provides a computer readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps of the virtual workstation application method based on the computing power scheduling are implemented.

[0056] In summary, the application pre-creates and maintains a buffer running pool of virtual workstations containing pre-installed systems and software, so that the user does not need to go through the long process of virtual machine creation, system installation and software configuration when applying, achieving a second-level response and resource delivery. This "hot standby" mechanism shortens the user's waiting time from minutes to seconds, greatly improving the user experience of cloud gaming, remote desktop and other applications with high real-time requirements.

[0057] The application innovatively uses a general PC with a GPU as the underlying computing power resource, replacing the expensive enterprise-level servers and storage arrays in the traditional scheme. Combined with the resource pooling management based on the OpenStack cloud management platform, not only the initial hardware investment cost (CAPEX) and operation and maintenance cost (OPEX) are greatly reduced, but also the "old" transformation of the existing PC is particularly suitable, giving the old hardware new life, in line with the concept of green computing.

[0058] The application creates a game disk by using the tiered snapshot technology of Ceph distributed storage, and realizes a revolutionary storage management method. All users share a unified, read-only basic game snapshot (read-only layer), and only the user's differentiated data is written to the independent writable layer. This makes the update of the game version only need to be updated once on the server side, and all newly applied users can immediately obtain the latest version, completely eliminating the huge storage overhead of copying a complete game disk for each user, while ensuring that "users always use the latest version", and realizing a substantial saving of storage space.

[0059] The application realizes automatic mapping of platform users and cloud platform tenants through deep integration of computing power scheduling platform (KubeSphere) and cloud management platform (OpenStack), ensures strict isolation of computing power resources, storage resources and data between different users, and meets the multi-tenant security requirements of enterprise-level applications. At the same time, the whole process realizes the automatic management of the whole life cycle of resources such as virtual workstations, GPUs, storage disks, etc., and improves the operation and maintenance efficiency.

[0060] The application forms a complete high-performance graphics application cloud solution from the selection of bottom-layer GPU hardware, to virtual workstation scheduling, to the final low-delay streaming technology of WebRTC. It ensures that the game picture of cloud rendering can be transmitted to the user terminal with high quality and low delay, reduces the cost, and does not sacrifice the core experience of picture quality and smoothness that the end user cares about.

[0061] The above examples are only used to illustrate the technical solutions of the application, but not to limit it; although the application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. A method for applying a virtual work station based on computing power scheduling, characterized in that, The method comprises the following steps: (S1) The application end initiates a virtual workstation application request through a computing power scheduling platform, and the computing power scheduling platform schedules a pre-created virtual workstation from a buffer running pool of a cloud management platform to the application end; (S2) The cloud management platform prepares an application end data disk and a game disk for the virtual workstation, wherein the game disk is created based on the latest game snapshot in a Ceph cluster; (S3) The application end data disk and the game disk are mounted to the virtual workstation; (S4) A virtual workstation game server is started, and a game stream is pushed to the application end through WebRTC technology; (S5) After the application end exits, the virtual workstation is remotely closed and the game disk is destroyed. 2.The virtual work station application method based on the computing power scheduling according to claim 1, wherein, The step (S1) comprises: The application end initiates a virtual workstation application on a visual interface of the computing power scheduling platform according to GPU type, frame rate, resolution, or bandwidth requirements; The computing power scheduling platform is built based on KubeSphere, and the cloud management platform is built based on OpenStack. 3.The virtual work station application method based on the computing power scheduling according to claim 1, wherein, The virtual workstation in the buffer running pool is a Windows virtual machine pre-installed with game supporting software, running on a PC with a GPU, and without a mounted storage disk. 4.The virtual work station application method based on the computing power scheduling according to claim 1, wherein, The step (S2) comprises: If the application end has no data disk, a data disk is created for it in the Ceph cluster; An independent game disk is created for the application end according to the latest game snapshot of the machine room where the application end is located.

5. The virtual work station application method based on the computing power scheduling according to claim 4, characterized in that, The game disk is created using a Ceph layered storage architecture, including a basic layer, an intermediate layer, and a writable layer, the game snapshot is a read-only layer, and the application end game disk is created based on the snapshot and the writable layer for incremental storage. 6.The virtual work station application method based on the computing power scheduling according to claim 1, wherein, The step (S3) is implemented through a disk mounting agent, comprising: After the virtual workstation is started, the disk mounting agent detects that the allocation state is allocated, obtains Ceph configuration and authentication information; The application end data disk and the game disk are mounted, and the mounting state is reported. 7.The virtual work station application method based on the computing power scheduling according to claim 1, wherein, The step (S4) comprises: After the disk mounting is successful, a game server order is called, and the MAC address of the virtual workstation is input; After the order application is successful, the cloud game server is started and pushed to the application end through WebRTC. 8.The virtual work station application method based on the computing power scheduling according to claim 1, wherein, The step (S5) comprises: After receiving the exit instruction of the application end, the virtual workstation is remotely closed through a BMC interface; The virtual workstation is removed from the running pool, the data disk is released, and the game disk is destroyed. 9.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to implement the virtual workstation application method based on computing power scheduling according to any one of claims 1 to 8.

10. A storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to implement the virtual workstation application method based on computing power scheduling according to any one of claims 1 to 8.

Images