System starting method, task execution method, and cloud computing system

Abstract

Embodiments of the present disclosure provide a system starting method, a task execution method, and a cloud computing system. The system starting method comprises: acquiring a virtual machine starting request; querying a target node corresponding to a first resource specification parameter from a node resource configuration table of a cloud computing system, wherein the target node comprises a resource having undergone pre-initialization processing on the basis of the first resource specification parameter; and scheduling the virtual machine starting request to the target node, so that the target node starts a target virtual machine on the basis of the resource having undergone initialization processing. Resource pre-initialization processing is carried out at a plurality of nodes. When a virtual machine starting request is obtained, a scheduling end determines a corresponding target node from the node resource configuration table of the cloud computing system on the basis of the first resource specification parameter and schedules the virtual machine starting request to the target node, so that the target node does not need to perform initialization processing on a resource, whereby the target virtual machine is quickly started, the virtual machine starting efficiency is improved, and the flexibility of the cloud computing system is improved.

Description

System startup method, task execution method and cloud computing system

[0001] This disclosure claims priority to the Chinese patent application filed with the China Patent Office on September 26, 2023, with application number 202311257088.0 and application name “System startup method, task execution method and cloud computing system”, the entire content of which is incorporated by reference into this disclosure. Technical Field

[0002] The present disclosure relates to the field of cloud computing, and more particularly to a system startup method, a task execution method, and a cloud computing system. Background Art

[0003] With the development of cloud computing technology, cloud computing systems will distribute the software and hardware resources used for computing on multiple nodes, facilitating dynamic expansion and contraction and providing elastic capabilities.

[0004] Currently, cloud computing systems execute computing tasks through virtual machines running on each node. The startup time of the virtual machine is an important indicator of the elasticity of the cloud computing system. The faster the system startup time, the faster the cloud computing system responds to elastic expansion needs and the better the user experience.

[0005] However, a significant portion of the virtual machine startup process is consumed by initializing node resources. For example, for a given resource specification, the time it takes from the user requesting instance creation on the frontend to the instance successfully being created is 77 seconds. Of this, initializing the graphics processing unit (GPU) on the node takes 43 seconds, accounting for over 50% of the total time. This significant drag on virtual machine startup efficiency is severely hampered by the node resource initialization process, resulting in inefficient system startup and insufficient cloud computing system elasticity. Therefore, a highly efficient system startup method is urgently needed to improve the elasticity of cloud computing systems.

[0006] Summary of the Invention

[0007] In view of this, embodiments of the present disclosure provide a system startup method. One or more embodiments of the present disclosure also relate to another system startup method, a task execution method, a system startup device, another system startup device, a task execution device, a cloud computing system, a computing device, a computer-readable storage medium, and a computer program to address technical deficiencies in the prior art.

[0008] The present disclosure provides a system startup method, which is applied to a scheduling end of a cloud computing system, including:

[0009] Obtaining a virtual machine startup request, wherein the virtual machine startup request carries a first resource specification parameter required to start a target virtual machine;

[0010] Querying a target node corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, wherein the target node includes a resource that has been pre-initialized according to the first resource specification parameter;

[0011] The virtual machine startup request is dispatched to the target node, so that the target node starts the target virtual machine based on the initialized resources.

[0012] In an embodiment of the present disclosure, a virtual machine startup request is obtained, wherein the virtual machine startup request carries the first resource specification parameter required to start the target virtual machine; the target node corresponding to the first resource specification parameter is queried from the node resource configuration table of the cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter; and the virtual machine startup request is dispatched to the target node so that the target node starts the target virtual machine based on the initialized resources. Resources are pre-initialized on multiple nodes. When the scheduling end obtains the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and dispatches the virtual machine startup request to the target node so that the target node does not need to initialize the resources again and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup and further improving the elasticity of the cloud computing system. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIG1 is a flow chart of a system startup method provided by one embodiment of the present disclosure;

[0014] FIG2 is a structural diagram of a cloud computing system in a system startup method provided by one embodiment of the present disclosure;

[0015] FIG3 is a schematic diagram of a process of resource initialization in a system startup method provided by an embodiment of the present disclosure;

[0016] FIG4 is a schematic diagram of a process of resource initialization in another system startup method provided by an embodiment of the present disclosure;

[0017] FIG5 is a schematic diagram of a process of resource initialization in another system startup method provided by an embodiment of the present disclosure;

[0018] FIG6 is a flow chart of a system startup method provided by an embodiment of the present disclosure;

[0019] FIG7 is a flowchart of another system startup method provided by an embodiment of the present disclosure;

[0020] FIG8 is a flowchart of a task execution method provided by one embodiment of the present disclosure;

[0021] FIG9 is a flowchart of a processing process of a system startup method applied to an e-commerce live broadcast data stream provided by one embodiment of the present disclosure;

[0022] FIG10 is a schematic structural diagram of a system startup device provided by an embodiment of the present disclosure;

[0023] FIG11 is a schematic structural diagram of another system startup device provided by an embodiment of the present disclosure;

[0024] FIG12 is a schematic structural diagram of a task execution device provided by one embodiment of the present disclosure;

[0025] FIG13 is a schematic diagram of the structure of a cloud computing system provided by one embodiment of the present disclosure;

[0026] FIG14 is a structural block diagram of a computing device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] The following description sets forth many specific details to facilitate a full understanding of the present disclosure. However, the present disclosure can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of the present disclosure. Therefore, the present disclosure is not limited to the specific implementations disclosed below.

[0028] The terms used in one or more embodiments of the present disclosure are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of the present disclosure. The singular forms "a", "the", and "the" used in one or more embodiments of the present disclosure and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of the present disclosure refers to and includes any or all possible combinations of one or more associated listed items.

[0029] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present disclosure, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of one or more embodiments of the present disclosure, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0030] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of the present disclosure are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0031] In one or more embodiments of the present disclosure, a large model refers to a deep learning model with large-scale model parameters, which typically contains hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than ten trillion model parameters. A large model can also be called a cornerstone model / foundation model (Foundation Model), which is pre-trained by using large-scale unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks, and the model has good generalization capabilities, such as a large-scale language model (LLM), a multi-modal pre-training model, etc.

[0032] In actual applications, large models only require a small number of samples to fine-tune the pre-trained model and can be applied to different tasks. Large models can be widely used in natural language processing (NLP), computer vision and other fields. Specifically, they can be applied to computer vision tasks such as visual question answering (VQA), image caption (IC), and image generation, as well as natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. The main application scenarios of large models include digital assistants, intelligent robots, search, online education, office software, e-commerce, and intelligent design.

[0033] First, the terms involved in one or more embodiments of the present disclosure are explained.

[0034] Cloud computing system: A computing system based on a distributed node cluster. Computing resources (such as hardware, software, and data storage) are distributed across multiple nodes, and computing services are provided through multiple nodes.

[0035] Node: A distributed node in a cloud computing system. Each node contains computing resources. For example, a 1TB disk can be split into 512GB segments during virtual machine startup to create a virtual machine instance. A single node corresponds to a cloud server, which independently executes tasks by invoking resources through virtual machines.

[0036] Virtual Machine (VM): A software program that emulates the functionality of one or more physical computers. Running VM instances on a node and configuring each VM instance with appropriate resources to perform tasks can more efficiently utilize resources and provide improved security and reliability. A VM instance is a virtual computer environment running on a node, simulating a complete computer system. VM instances can provide independent operating environments for multiple computing tasks, thereby more efficiently utilizing node resources and improving the availability and resiliency of cloud computing systems.

[0037] Multilayer Perceptron (MLP) model: is a multi-layer feedforward neural network used for regression and classification tasks.

[0038] Support Vector Regression (SVR) model: a support vector machine model used for regression tasks.

[0039] XGBoost (Extreme Gradient Boosting): An artificial intelligence algorithm based on the gradient boosting algorithm. For classification and regression problems, it uses the gradient boosting algorithm to build decision trees and uses specific optimization algorithms to improve the accuracy and efficiency of the model.

[0040] In the present disclosure, a system startup method is provided. The present disclosure also relates to another system startup method, a task execution method, a system startup device, another system startup device, a task execution device, a cloud computing system, a computing device, a computer-readable storage medium and a computer program, which are described in detail one by one in the following embodiments.

[0041] Referring to FIG. 1 , FIG. 1 shows a flow chart of a system startup method provided by an embodiment of the present disclosure. The method is applied to a scheduling end of a cloud computing system and includes the following specific steps:

[0042] Step 102: Obtain a virtual machine startup request, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine.

[0043] The embodiments of the present disclosure are applied to the scheduling end of a cloud computing system. The scheduling end is a functional module for scheduling instruction requests. The scheduling end records the node resource configuration table of the cloud computing system. After obtaining a virtual machine startup request, it is reasonably scheduled to a distributed node, and the virtual machine startup is completed on the node, thereby realizing system startup. Figure 2 shows a structural diagram of a cloud computing system in a system startup method provided by an embodiment of the present disclosure. As shown in Figure 2: The cloud computing system includes a virtual machine scheduling end (i.e., the scheduling end), an artificial intelligence-based prediction system, a database, and multiple distributed nodes: node 1, node 2, node 3, and node 4.

[0044] The virtual machine startup request is an instruction request for the control node to start the virtual machine instance of the virtual machine. The target virtual machine is the virtual machine to be started for the virtual machine startup request, specifically the virtual machine instance. It does not actually exist before step 106, but actually exists after the corresponding startup on the target node in the subsequent step 106. The target virtual machine needs to be started and run based on the resources of the first resource specification parameter. The virtual machine startup request can be generated and sent to the scheduling end after the user determines the target virtual machine to be started at the front end of the cloud computing system. For example, the user selects the target virtual machine Unnamed.VM to be started at the front end, selects the first resource specification parameter: CPU number (1); GPU number (4) model (A100); memory capacity (8GB); disk capacity (512GB); disk type (SCSI); I / O controller type (LSI Logic); operating system version (Centos 7 64-bit), and generates a virtual machine startup request. The virtual machine startup request can also be generated by the scheduling end when the virtual machine startup condition is triggered. For example, when the computing task volume of the cloud computing system exceeds the operating upper limit of the current node, a virtual machine startup request is generated to perform elastic expansion.

[0045] Resources are the hardware and software resources deployed on a node. Hardware resources include, but are not limited to, central processing units (CPUs), graphics processing units, memory, disks, I / O controllers, and network hardware. Software resources include, but are not limited to, operating systems, databases, execution methods, execution classes, and interfaces.

[0046] Resource specification parameters are resource specification parameters, including specification parameters of software and hardware resources. Among them, hardware resource specification parameters include but are not limited to: central processing unit specification parameters (model and number of units, etc.), graphics processing unit specification parameters (model, number of units and video memory, etc.), memory specification parameters (memory medium model, memory medium type, number of memory media and memory capacity, etc.), disk specification parameters (disk medium model, disk medium type, number of disk media and disk capacity, etc.), I / O controller specification parameters (I / O controller type, etc.), network hardware specification parameters (network type, network address, gateway, subnet, server address, network bandwidth and network connection method, etc.) and hardware compatibility. Software resource specification parameters include but are not limited to: operating system specification parameters (type, version and running location, etc.), database specification parameters (database source, database capacity and database location, etc.), running method and running class specification parameters (version, throughput and running time, etc.), interface (version and interface type, etc.). The first resource specification parameters are the specification parameters of the resources required to start the target virtual machine, including specification parameters of software and hardware resources.

[0047] Obtaining the virtual machine startup request may be receiving the virtual machine startup request sent by the front end, or generating the virtual machine startup request when a preset virtual machine startup condition is triggered.

[0048] For example, the cloud computing system is a public cloud computing system provided by a cloud service provider. A game company chooses the public cloud computing system and needs to rent a node cluster consisting of 10 distributed nodes on the public cloud computing system and run the target virtual machine game.VM_i (i∈[0,9]) on the 10 distributed nodes to realize the cloud server construction of the game. The developer of the game company selects 10 target virtual machines on the front-end page of the public cloud computing system and selects the first resource specification parameters required to start the 10 target virtual machines: number of CPUs (1); number of GPUs (4) model (A100); memory capacity (8GB); disk capacity (512GB) disk type (SCSI); I / O controller type (LSI Logic); operating system version (Centos 7 64-bit). The front-end generates a corresponding virtual machine startup request and sends the virtual machine startup request to the scheduling end of the public cloud computing system. The scheduling end receives the virtual machine startup request sent by the front-end.

[0049] A virtual machine startup request is obtained, wherein the virtual machine startup request carries the first resource specification parameter required to start the target virtual machine, providing a query basis for subsequent querying the corresponding target node.

[0050] Step 104: querying a target node corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter.

[0051] The node resource configuration table of a cloud computing system records the resource configuration specifications of nodes in the cloud computing system. The node resource configuration table is dynamically updated based on the node resource configuration specifications. When a node's resource configuration specifications are updated, the node's resource specification parameters recorded in the node resource configuration table are also updated accordingly. For example, the node configuration table records the resource configuration specifications of 100 nodes in the cloud computing system: Node 1: Current CPU specifications; Current GPU specifications; Current memory specifications; Current disk specifications; Current operating system specifications... Node 100: Current CPU specifications; Current GPU specifications; Current memory specifications; Current disk specifications; Current operating system specifications.

[0052] The target node is a node that has completed pre-initialization processing of resources in accordance with the first resource specification parameters. Pre-initialization processing is the process of setting up a virtual machine operating environment for the resources on the node. The resources that have undergone pre-initialization processing can be directly used to create a virtual machine instance to achieve rapid startup of the virtual machine, for example, loading the running method, running class, setting the data read and write path, establishing an NVLink high-speed channel between the two graphics processing units, pulling the operating system image file, and memory or disk partition processing. Optionally, the selection of the target node can be determined based on the time distribution and spatial distribution on the basis of matching the first resource specification parameters. For example, the spatial locations of the two nodes are A and B respectively, and the front end is in A. The target node located in A is preferred. For another example, the status of the two nodes in the next 10 minutes is running and idle respectively. The latter is preferred as the target node.

[0053] For example, from the node resource configuration table of the public cloud computing system, the first resource specification parameter is queried to correspond to 10 target nodes: node 2, node 6, node 13...node 77, wherein the 10 distributed nodes of node 2, node 6, node 13...node 77 include pre-initialized hardware resources: number of CPUs (1); number of GPUs (4) model (A100); memory capacity (8GB); disk capacity (512GB) disk type (SCSI); I / O controller type (LSI Logic), and software resources: operating system version (Centos 7 64-bit).

[0054] A target node corresponding to the first resource specification parameter is queried from a node resource configuration table of the cloud computing system, where the target node includes resources that have been pre-initialized according to the first resource specification parameter. Based on the first resource specification parameter required to start the target virtual machine, the corresponding target node is accurately determined from the node resource configuration table of the cloud computing system, facilitating the subsequent dispatch of the virtual machine startup request to the target node and laying the foundation for normal and rapid startup of the target virtual machine.

[0055] Step 106: dispatching the virtual machine startup request to the target node, so that the target node starts the target virtual machine based on the initialized resources.

[0056] The target virtual machine is started based on the initialized resources, specifically by creating a virtual machine instance of the target virtual machine based on the initialized resources, running the virtual machine instance, and starting the target virtual machine.

[0057] Exemplarily, the virtual machine startup request is scheduled to node 2, node 6, node 13...node 77, a total of 10 target nodes, so that the 10 target nodes create virtual machine instances Instance.game.VM_i (i∈[0,9]) of the target virtual machine game.VM_i (i∈[0,9]) based on the initialized hardware resources (CPU, GPU, memory, disk and I / O controller) and software resources (operating system), run the virtual machine instances respectively, and start the target virtual machine.

[0058] In an embodiment of the present disclosure, a virtual machine startup request is obtained, wherein the virtual machine startup request carries the first resource specification parameter required to start the target virtual machine; the target node corresponding to the first resource specification parameter is queried from the node resource configuration table of the cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter; and the virtual machine startup request is dispatched to the target node so that the target node starts the target virtual machine based on the initialized resources. Resources are pre-initialized on multiple nodes. When the scheduling end obtains the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and dispatches the virtual machine startup request to the target node so that the target node does not need to initialize the resources again and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup and further improving the elasticity of the cloud computing system.

[0059] In an optional embodiment of the present disclosure, after step 104, the following specific steps are further included:

[0060] In the case that the target node corresponding to the first resource specification parameter is not found, identifying whether there is an idle node, wherein the idle node is a node that has not been initialized;

[0061] If so, the virtual machine startup request is dispatched to an idle node, so that resources of the idle node are initialized based on the first resource specification parameter, and the target virtual machine is started based on the initialized resources.

[0062] In step 104, the target node corresponding to the first resource specification parameter is queried from the node resource configuration table. However, if the corresponding target node is not found, the virtual machine startup is terminated. The virtual machine startup does not have high availability and the elasticity of the cloud computing system cannot be guaranteed.

[0063] An idle node is a node whose resources have not been initialized in advance. An idle node includes idle resources that have not been initialized, including idle software and hardware resources. The idle resources on the idle node have not completed the process of setting up the virtual machine operating environment. If you want to create a virtual machine instance and start the virtual machine, you need to initialize the resources and then start the target virtual machine based on the initialized resources. Optionally, an idle node is a node that includes resources required by the first resource specification parameters and has not been initialized. An idle node includes idle resources required by the first resource specification parameters that have not been initialized. Optionally, the selection of idle nodes can also be determined based on time distribution and spatial distribution.

[0064] Identify whether there are idle nodes. The identification methods include but are not limited to: from the node resource configuration table, determine that the nodes without recorded resource configuration specifications are idle nodes; determine whether each node is an idle node by sending heartbeat detection to each node; determine whether there are idle nodes based on the resource usage reported by each node; query whether there are reserved idle resources, and if so, determine that the nodes with reserved idle resources are idle nodes. There is no limitation here.

[0065] It should be noted that the embodiment of the present disclosure is equivalent to a normal virtual machine startup process when steps 102 to 106 cannot be executed.

[0066] Exemplarily, when the target node corresponding to the first resource specification parameter is not found, 10 distributed nodes without recorded resource configuration specifications are determined as idle nodes from the node resource configuration table: node 1, node 4, node 18...node 84, and the virtual machine startup request is scheduled to the 10 idle nodes so that the resources of the idle nodes are based on the first resource specification parameter: number of CPUs (1); number of GPUs (4) model (A100); memory capacity (8GB); disk capacity (512GB) disk type (SCSI); I / O controller type (LSI Logic), and software resources: operating system version (Centos7 64-bit). The hardware resources and software resources are initialized, and based on the initialized hardware resources and software resources, a virtual machine instance Instance.game.VM_i (i∈[0,9]) of the target virtual machine game.VM_i (i∈[0,9]) is created, and the virtual machine instances are run respectively to start the target virtual machine.

[0067] If no target node corresponding to the first resource specification parameter is found, the system determines whether there is an idle node, where an idle node is a node that has not been initialized. If so, the system dispatches the virtual machine startup request to the idle node, initializes the idle node's resources based on the first resource specification parameter, and then starts the target virtual machine based on the initialized resources. This ensures high availability of virtual machine startup and further enhances the elasticity of the cloud computing system.

[0068] In an optional embodiment of the present disclosure, after identifying whether there is an idle node, the following specific steps are further included:

[0069] If it does not exist, determining a reference node, wherein the reference node includes resources that have been initialized in advance and the virtual machine has not been started;

[0070] The virtual machine startup request is dispatched to the reference node, so that the reference node resets the resources of the reference node, initializes the processing resources based on the first resource specification parameter, and starts the target virtual machine.

[0071] In the above embodiment, whether there is an idle node is identified for backup node scheduling to complete the virtual machine startup. However, if no idle node is identified, the virtual machine startup is terminated. The virtual machine startup still does not have high availability and the elasticity of the cloud computing system cannot be guaranteed.

[0072] The reference node is a node that includes resources that have been pre-initialized and have not started a virtual machine. Optionally, the reference node includes resources that have been pre-initialized according to other specification parameters different from the first resource specification parameters and have not started a virtual machine. For example, the first resource specification parameters are: number of CPUs (1); number of GPUs (4) model (A100); memory capacity (8GB); disk capacity (512GB) disk type (SCSI); I / O controller type (LSI Logic); operating system version (Centos 7 64-bit). The other resource specification parameters are: number of CPUs (1); number of GPUs (6) model (A100); memory capacity (16GB); disk capacity (1TB) disk type (SATA); I / O controller type (LSI Logic); operating system version (Centos 7 32-bit). The reference node pre-initializes resources according to the other resource specification parameters different from the first resource specification parameters, and the resources do not start a virtual machine.

[0073] The reference node is determined by querying the reference node corresponding to other resource specification parameters from the node resource configuration table, wherein the other resource specification parameters are different from the first resource specification parameters.

[0074] Optionally, after querying the reference nodes corresponding to other resource specification parameters, the following specific steps are also included:

[0075] If no reference node corresponding to other resource specification parameters is found, the virtual machine startup is terminated.

[0076] The resources of the reference node are reset based on the first resource specification parameters, and the target virtual machine is started. Specifically, the resources are reset, the resources are initialized based on the first resource specification parameters, and the target virtual machine is started based on the initialized resources.

[0077] For example, if it does not exist, query other resource specification parameters from the node resource configuration table: number of CPUs (1); number of GPUs (6) model (A100); memory capacity (16GB); disk capacity (1TB) disk type (SATA); I / O controller type (LSI Logic); operating system version (Centos 7 32-bit), and the corresponding 10 distributed nodes that have not started the virtual machine are reference nodes: node 3, node 5, node 16...node 80, and schedule the virtual machine startup request to the 10 reference nodes so that the 10 reference nodes reset the hardware resources and software resources, initialize the hardware resources and software resources based on the first resource specification parameters, and create a virtual machine instance Instance.game.VM_i (i∈[0,9]) of the target virtual machine game.VM_i (i∈[0,9]) based on the initialized hardware and software resources, run the virtual machine instances respectively, and start the target virtual machine.

[0078] If no such node exists, a reference node is determined, where the reference node includes pre-initialized resources that have not yet started a virtual machine. The virtual machine startup request is dispatched to the reference node, causing the reference node to reset its resources, initialize the processing resources based on the first resource specification parameters, and start the target virtual machine. This further ensures high availability of virtual machine startup and further enhances the elasticity of the cloud computing system.

[0079] In an optional embodiment of the present disclosure, before step 104, the following specific steps are further included:

[0080] Initialize the resources of nodes in the cloud computing system based on sample resource specification parameters;

[0081] The resource specification parameters of the initialized node are recorded in the node resource configuration table of the cloud computing system.

[0082] Sample resource specification parameters are the resource specification parameters required to start a sample virtual machine, including software and hardware resource specification parameters. The sample virtual machine is a virtual machine that has been started before the initialization process. The sample virtual machine runs on the node according to the sample resources with sample resource specification parameters.

[0083] In the embodiment of the present disclosure, the node in the cloud computing system is a node that includes resources that have not been initialized and the virtual machine has not been started. It should be noted that the fact that the virtual machine has not been started does not mean that the virtual machine has never been started on the idle node, but that the resources on the node that meet the first resource specification parameters have not currently started the virtual machine.

[0084] An initialized node is a node that has completed resource initialization according to specific resource specification parameters. The specific resource specification parameters can be the first resource specification parameters or resource specification parameters different from the first resource specification parameters. There is at least one initialized node, which can be understood as at least one set of initialized node resource plans. When a virtual machine needs to be started, the node resource plan with the most matching specifications is selected to enable rapid startup of the virtual machine.

[0085] Based on the sample resource specification parameters, the resources of the nodes in the cloud computing system are initialized. The specific method is: according to a predetermined initialization strategy, based on the sample resource specification parameters, the resources of the nodes in the cloud computing system are initialized, wherein the initialization strategy includes at least one of a node state retention strategy, a node state redundancy strategy and an artificial intelligence prediction strategy. The node state retention strategy is a strategy for retaining the sample resource specification parameters of the node where the historical virtual machine is located. The node state redundancy strategy is a strategy for initializing the resources on the backup node based on the sample resource specification parameters of the sample virtual machine. The artificial intelligence prediction strategy is a strategy for outputting predicted resource specification parameters based on the sample resource specification parameters using a prediction model and initializing the resources on the idle nodes.

[0086] Exemplarily, according to a predetermined node status retention strategy, node status redundancy strategy and artificial intelligence prediction strategy, based on sample resource specification parameters, the resources of 60 nodes in the cloud computing system are initialized, and the resource specification parameters of the 60 nodes after initialization are recorded in the node resource configuration table of the cloud computing system.

[0087] Based on the sample resource specifications, the resources of the cloud computing system nodes are initialized and recorded in the node resource configuration table of the cloud computing system. This provides a basis for subsequent queries of the corresponding target node and provides initialized resources for subsequent virtual machine startup, improving the stability and efficiency of virtual machine startup and the stability and efficiency of system startup.

[0088] In an optional embodiment of the present disclosure, initializing resources of nodes in a cloud computing system based on sample resource specification parameters includes the following specific steps:

[0089] For any historical virtual machine, when a virtual machine stop request for any historical virtual machine is obtained, resetting the resource state of the first node that started any historical virtual machine;

[0090] Based on the sample resource specification parameters required to start any historical virtual machine, the resources of the first node are initialized.

[0091] When the historical virtual machine stops, the resources of the first node where the historical virtual machine is located are reset, and reinitialized according to the sample resource specification parameters of the historical virtual machine, and the sample resource specification parameters of the resources on the first node are recorded in the node resource configuration table. In step 104, when the sample resource specification parameters of the first node match the first resource specification parameters, the first node is determined to be the corresponding target node, and the resource initialization process can be skipped directly, which saves the startup time of the virtual machine and improves the efficiency of system startup.

[0092] A historical virtual machine is a virtual machine (specifically, a virtual machine instance) running on a node before its resources are initialized. A virtual machine stop request is an instruction request for controlling a node to stop a virtual machine instance. These include virtual machine stop requests and virtual machine deletion requests. For example, if virtual machine VM_1.1 is running on Node 1, upon receiving a virtual machine stop request, VM_1.1 is stopped and resources on Node 1 are initialized. At this point, VM_1.1 becomes a historical virtual machine.

[0093] The first node is the node where the historical virtual machine is located. The first node needs to initialize the resources according to the sample resource specification parameters of the historical virtual machine. The resource status of the first node is the running status of the resources on the first node, including the reset state (the state where the resources are not initialized, which can be understood as the state where the resources are formatted), the waiting state and the running state.

[0094] FIG3 shows a schematic diagram of a process of resource initialization in a system startup method provided by an embodiment of the present disclosure, as shown in FIG3 :

[0095] Start; when a virtual machine stop request for a historical virtual machine is obtained, stop the historical virtual machine; reset the resource status of the resources on the node of the historical virtual machine; initialize the resources of the node based on the sample resource specification parameters required by the historical virtual machine; record the resource specification parameters of the initialized node into the node resource configuration table of the cloud computing system; end.

[0096] When a virtual machine stop request is obtained for any historical virtual machine, the resource status of the first node that starts any historical virtual machine is reset. The specific method is: when a virtual machine stop request is obtained for any historical virtual machine, the virtual machine stop request is scheduled to the first node that starts any historical virtual machine, so that the first node stops running any historical virtual machine and changes the resource status of the first node from the running state to the reset state.

[0097] For example, for any historical virtual machine: the virtual machine test.VM on node 32, when a virtual machine stop request for the virtual machine test.VM is obtained, the virtual machine stop request is scheduled to node 32, so that node 32 stops running the virtual machine instance of the virtual machine test.VM, and changes the resource state of node 32 from the running state to the reset state, completing the formatting processing of the resources on node 32.

[0098] For any historical virtual machine, if a stop request is received, the resource state of the first node that started the historical virtual machine is reset; based on the sample resource specification parameters required to start the historical virtual machine, the resources of the first node are initialized. This provides initialized resources for subsequent virtual machine startups, improving the stability and efficiency of virtual machine startup, and improving the stability and efficiency of system startup.

[0099] In an optional embodiment of the present disclosure, initializing resources of nodes in a cloud computing system based on sample resource specification parameters includes the following specific steps:

[0100] For any sample virtual machine, upon receiving a virtual machine creation request for any sample virtual machine, create any sample virtual machine;

[0101] Determine sample resource specification parameters and idle second nodes required to start any sample virtual machine;

[0102] Based on the sample resource specification parameters, the resources of the second node are initialized.

[0103] When a virtual machine creation request is obtained for any sample virtual machine, if there is a second node with idle resources, a backup node with the same sample resource specification parameters, i.e., the second node, is initialized according to the sample virtual machine just created, and the sample resource specification parameters of the resources on the second node are recorded in the node resource configuration table. In step 104, when the sample resource specification parameters of the resources on the second node match the first resource specification parameters, the second node is determined to be the corresponding target node, and the resource initialization process can be skipped directly, which saves the startup time of the virtual machine and improves the efficiency of system startup.

[0104] A sample virtual machine is a virtual machine, specifically a virtual machine instance, running on another node before the resources of the current node are initialized. A virtual machine creation request is an instruction request used to control other nodes to create a sample virtual machine instance. For example, when a virtual machine creation request is obtained, the virtual machine creation request is dispatched to node 2, virtual machine VM_1.2 is created on node 2, and resources on node 3 are initialized based on the resource specification parameters for creating virtual machine VM_1.2. From the perspective of node 3, the current node, node 2 is another node, and virtual machine VM_1.2 is a sample virtual machine running on another node.

[0105] The second node is the node where resources are to be initialized. The second node needs to initialize the resources according to the sample resource specification parameters of the sample virtual machine. The resources on the second node are idle before initialization, that is, idle resources. The second node that completes the resource initialization can be understood as a backup node for other nodes where the sample virtual machine is located.

[0106] FIG4 shows a schematic diagram of a process of resource initialization in another system startup method provided by an embodiment of the present disclosure, as shown in FIG4 :

[0107] Start; create a sample virtual machine; creation is complete; determine whether there are still nodes with idle resources; if so, initialize the node's resources based on the sample resource specification parameters of the sample virtual machine, record the resource specification parameters of the initialized node in the node resource configuration table of the cloud computing system, and end; if not, end directly.

[0108] When a virtual machine creation request for any sample virtual machine is obtained, any sample virtual machine is created. The specific method is: when a virtual machine creation request for any sample virtual machine is obtained, the virtual machine creation request is dispatched to other nodes so that other nodes create any sample virtual machine.

[0109] Exemplarily, for any sample virtual machine: virtual machine test.VM, when a virtual machine creation request for the sample virtual machine test.VM is obtained, the virtual machine creation request is scheduled to other nodes: node 20, so that node 20 creates a virtual machine instance of the sample virtual machine test.VM, determines the sample resource specification parameters required to start the virtual machine instance of the sample virtual machine test.VM and the idle second node: node 30, and initializes the resources of node 30 based on the sample resource specification parameters.

[0110] For any sample virtual machine, upon receiving a virtual machine creation request for any sample virtual machine, the system creates the sample virtual machine; determines the sample resource specification parameters and an idle second node required to start the sample virtual machine; and initializes the resources of the second node based on the sample resource specification parameters. This provides initialized resources for subsequent virtual machine startup, improving the stability and efficiency of virtual machine startup, and enhancing the stability and efficiency of system startup.

[0111] In an optional embodiment of the present disclosure, initializing resources of nodes in a cloud computing system based on sample resource specification parameters includes the following specific steps:

[0112] For at least one sample virtual machine, obtaining sample resource specification parameters and distribution parameters of the at least one sample virtual machine;

[0113] Based on the sample resource specification parameters and distribution parameters, a prediction model is used to predict and obtain the predicted resource specification parameters required to start any target virtual machine;

[0114] Based on the predicted resource specification parameters, the resources of the idle third node are initialized.

[0115] Using the prediction model, prediction is performed based on the sample resource specification parameters and distribution parameters to obtain the predicted resource specification parameters required to start any target virtual machine. Based on the predicted resource specification parameters, the resources of the idle third node are initialized, and the sample resource specification parameters of the resources on the third node are recorded in the node resource configuration table. In step 104, when the sample resource specification parameters of the third node match the first resource specification parameters, the third node is determined to be the corresponding target node, and the resource initialization process can be directly skipped, which saves the startup time of the virtual machine and improves the efficiency of system startup.

[0116] A sample virtual machine is a virtual machine running on other nodes before the resources of the current node are initialized, specifically a virtual machine instance. A virtual machine creation request is an instruction request for controlling other nodes to create sample virtual machine instances. For example, virtual machine VM_1.4 is running on node 4, virtual machine VM_1.5 is running on node 5, and virtual machine VM_1.6 is running on node 6. On node 7, the resources of node 7 are initialized based on the resource specification parameters of virtual machines VM_1.4, VM_1.5, and VM_1.6. From the perspective of node 7, the current node, nodes 4, 5, and 6 are other nodes, and virtual machines VM_1.4, VM_1.5, and VM_1.6 are sample virtual machines running on other nodes.

[0117] The distribution parameters are the distribution attribute parameters of the sample virtual machine, including but not limited to: time distribution parameters, space distribution parameters and instance quantity distribution parameters. The time distribution parameters are the distribution parameters of the startup time of the sample virtual machine. For example, the startup time of the sample virtual machine is 12:00:00 on September 15, 2023, and the time distribution parameters are: (2023-9-15; 12:00:00). The space distribution parameters are the spatial distribution parameters of the node where the sample virtual machine is located. For example, the node where the sample virtual machine is located is in location A, and the space distribution parameters are: location A. The instance quantity distribution parameters are the distribution parameters of the virtual machine instances on the nodes. For example, the nodes where the sample virtual machine is located include: node 9, node 10 and node 11, and the number of instances on each node is: 12, 4 and 9 respectively. The instance quantity distribution parameters are: (12; 4; 9).

[0118] The prediction model is a neural network model used to predict resource specifications, including but not limited to multi-layer perceptrons and support vector regression models. Based on sample resource specifications and distribution parameters, the prediction model determines the relationship between the resource specifications and distribution parameters. Based on the input distribution parameters, the prediction model outputs predicted resource specifications.

[0119] The predicted resource specification parameters are the predicted resource specification parameters, including the specification parameters of software and hardware resources.

[0120] FIG5 shows a schematic diagram of a process of resource initialization in another system startup method provided by an embodiment of the present disclosure, as shown in FIG5 :

[0121] Start; determine whether there are any nodes with idle resources; if not, end directly; if so, analyze the sample resource specification parameters and distribution parameters of the sample virtual machine, use the prediction model to predict, and obtain the predicted resource specification parameters required to start any target virtual machine; determine whether there is a node with predicted resource specification parameters; if not, end directly; if so, send the predicted resource specification parameters to the node, initialize the node's resources based on the predicted resource specification parameters, record the resource specification parameters of the initialized node in the node resource configuration table of the cloud computing system, and end.

[0122] Based on the sample resource specification parameters and distribution parameters, the prediction model is used to perform prediction to obtain the predicted resource specification parameters required to start any target virtual machine. The specific method is to perform regression prediction based on the sample resource specification parameters and distribution parameters, and obtain the predicted resource specification parameters required to start any target virtual machine.

[0123] Exemplarily, for 1,000 sample virtual machines, the sample resource specification parameters, time distribution parameters, spatial distribution parameters and instance quantity distribution parameters of the 1,000 sample virtual machines are obtained. Based on the sample resource specification parameters, time distribution parameters, spatial distribution parameters and instance quantity distribution parameters, a prediction model is used to perform predictions to obtain the predicted resource specification parameters required to start any target virtual machine. Based on the predicted resource specification parameters, the resources of the idle third node: node 40 are initialized.

[0124] For at least one sample virtual machine, sample resource specification parameters and distribution parameters are obtained for the sample virtual machine. Based on the sample resource specification parameters and distribution parameters, a prediction model is used to predict the predicted resource specification parameters required to start any target virtual machine. Based on the predicted resource specification parameters, resources of an idle third node are initialized. Using a neural network model for prediction, initialized resources are provided for subsequent virtual machine startup, improving the stability and efficiency of virtual machine startup, and ultimately, the stability and efficiency of system startup.

[0125] In an optional embodiment of the present disclosure, the distribution parameter includes at least one of a time distribution parameter and a space distribution parameter;

[0126] Correspondingly, based on the sample resource specification parameters and distribution parameters, a prediction model is used to perform prediction to obtain the predicted resource specification parameters required to start any target virtual machine, including the following specific steps:

[0127] Based on the sample resource specification parameters and time distribution parameters, a prediction model is used to perform time regression prediction on the target virtual machine startup requests obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup requests;

[0128] or,

[0129] Based on the sample resource specification parameters and spatial distribution parameters, a prediction model is used to perform spatial regression prediction on the target virtual machine startup requests that have obtained the prediction space, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup requests;

[0130] or,

[0131] Based on the sample resource specification parameters, time distribution parameters and spatial distribution parameters, the prediction model is used to perform spatiotemporal regression prediction on the target virtual machine startup requests that obtain the prediction space within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup requests.

[0132] The time distribution parameter is the distribution parameter of the sample virtual machine's startup time. For example, if the sample virtual machine is started at 12:00:00 on September 15, 2023, the time distribution parameter is: (2023-9-15; 12:00:00). The spatial distribution parameter is the spatial distribution parameter of the node where the sample virtual machine is located. For example, if the node where the sample virtual machine is located is in location A, the spatial distribution parameter is: location A.

[0133] In the disclosed embodiments, the prediction model is a neural network model that supports regression. The prediction model is constructed based on sample resource specification parameters and distribution parameters to form a regression model, which models the regression relationship between resource specification parameters and distribution parameters. The prediction model outputs predicted resource specification parameters based on the input distribution parameters. The prediction model is optimized using an extreme gradient algorithm, using a gradient boosting algorithm to construct a decision tree and employing a specific optimization algorithm to improve the accuracy and efficiency of the prediction model.

[0134] Based on the sample resource specification parameters and time distribution parameters, the prediction model is used to perform time regression prediction on the target virtual machine startup requests obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. The specific method is: based on the sample resource specification parameters and time distribution parameters, the prediction model is used to construct a regression model, and the trained regression model is used to perform time regression prediction on the target virtual machine startup requests obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request.

[0135] Exemplarily, based on the sample resource specification parameters and time distribution parameters of 1,000 sample virtual machines and the XGBoost algorithm, a regression model is constructed using a prediction model. The trained regression model is used to perform time regression prediction on the target virtual machine startup request obtained within the prediction time (20:00:00 on September 16, 2023). The predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. Based on the predicted resource specification parameters, the resources of the idle third node: node 40 are initialized.

[0136] Based on the sample resource specification parameters and spatial distribution parameters, the prediction model is used to perform spatial regression prediction on the target virtual machine startup request that obtains the prediction space, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. The specific method is: based on the sample resource specification parameters and spatial distribution parameters, the prediction model is used to construct a regression model, and the trained regression model is used to perform spatial regression prediction on the target virtual machine startup request in the prediction space, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request.

[0137] Exemplarily, based on the sample resource specification parameters and spatial distribution parameters, based on the XGBoost algorithm, a regression model is constructed using a prediction model, and the trained regression model is used to perform spatial regression prediction on the target virtual machine startup request in the prediction space (A), and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. Based on the predicted resource specification parameters, the resources of the idle third node: node 40 are initialized.

[0138] Based on the sample resource specification parameters, time distribution parameters and spatial distribution parameters, the prediction model is used to perform spatiotemporal regression prediction on the target virtual machine startup request that obtains the prediction space within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. The specific method is: based on the sample resource specification parameters, time distribution parameters and spatial distribution parameters, the prediction model is used to construct a regression model, and the trained regression model is used to perform spatiotemporal regression prediction on the target virtual machine startup request that obtains the prediction space within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request.

[0139] Exemplarily, based on the sample resource specification parameters, time distribution parameters and spatial distribution parameters, and based on the XGBoost algorithm, a regression model is constructed using a prediction model, and the trained regression model is used to perform spatiotemporal regression prediction on the target virtual machine startup request obtained in the prediction space (A) within the prediction time (20:00:00 on September 16, 2023). The predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request, and based on the predicted resource specification parameters, the resources of the idle third node: node 40 are initialized.

[0140] Based on sample resource specification parameters and time distribution parameters, a prediction model is used to perform time regression prediction on target virtual machine startup requests obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. Alternatively, based on sample resource specification parameters and space distribution parameters, a prediction model is used to perform space regression prediction on target virtual machine startup requests obtained within the prediction space, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. Alternatively, based on sample resource specification parameters, time distribution parameters, and space distribution parameters, a prediction model is used to perform spatiotemporal regression prediction on target virtual machine startup requests obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request. The use of a neural network model for spatiotemporal regression prediction accurately provides initialized resources for subsequent virtual machine startup in both time and space, further ensuring the stability of virtual machine startup, further improving the efficiency of virtual machine startup, and further improving the efficiency of system startup.

[0141] FIG6 shows a flow chart of a system startup method provided by an embodiment of the present disclosure, as shown in FIG6 :

[0142] Start; query whether there is a target node corresponding to the first resource specification parameter from the node resource configuration table of the cloud computing system.

[0143] If yes, the virtual machine startup request is dispatched to the target node. The target node quickly starts the target virtual machine based on the initialized resources, and the process ends. This method significantly shortens the virtual machine startup time and improves the efficiency of system startup.

[0144] If not, determine whether there is an idle node. If so, schedule the virtual machine startup request to the idle node, initialize the idle node's resources based on the first resource specification parameter, and start the target virtual machine based on the initialized resources, and then terminate. This method performs a normal virtual machine startup, and the startup time is consistent with the current system startup method, without increasing the virtual machine startup time.

[0145] If not, identify whether there are idle nodes; if not, determine whether there is a reference node that has been pre-initialized and has not yet started a virtual machine; if not, terminate immediately; if so, dispatch the virtual machine startup request to the reference node, reset the reference node's resources based on the first resource specification parameters, start the target virtual machine based on the reset resources, and terminate. This method adds the operation of resetting resources, but because the resource reset time is relatively short, it does not significantly increase the virtual machine startup time.

[0146] Referring to FIG. 7 , FIG. 7 shows a flowchart of another system startup method provided by an embodiment of the present disclosure. The method is applied to a target node of a cloud computing system and includes the following specific steps:

[0147] Step 702: Receive a virtual machine startup request scheduled by the scheduling end, wherein the virtual machine startup request carries first resource specification parameters required to start the target virtual machine, and the first resource specification parameters are resource specification parameters of resources on the target node.

[0148] Step 704: Based on the initialized resources, the target virtual machine is started, thereby achieving system startup.

[0149] The embodiment of the present disclosure is applied to the target node of the cloud computing system. The target node is a node containing resources pre-initialized according to the first resource specification parameters. The scheduling end of the cloud computing system is a functional module for scheduling instruction requests. The scheduling end records the node resource configuration table of the cloud computing system. After obtaining the virtual machine startup request, it is reasonably scheduled to the distributed deployed node to complete the virtual machine startup on the node.

[0150] The target virtual machine is started based on the initialized resources, specifically by creating a virtual machine instance of the target virtual machine based on the initialized resources, running the virtual machine instance, and starting the target virtual machine.

[0151] For example, Node 2, Node 6, Node 13...Node 77, a total of 10 target nodes, receive a virtual machine startup request scheduled by the scheduling end of the public cloud computing system, and based on the initialized hardware resources (CPU, GPU, memory, disk and I / O controller) and software resources (operating system), create a virtual machine instance Instance.game.VM_i (i∈[0,9]) of the target virtual machine game.VM_i (i∈[0,9]), run the virtual machine instance respectively, and start the target virtual machine.

[0152] In an embodiment of the present disclosure, a virtual machine startup request scheduled by a scheduling end is received, wherein the virtual machine startup request carries a first resource specification parameter required to start a target virtual machine, and the first resource specification parameter is a resource specification parameter of a resource on a target node; based on the initialized resources, the target virtual machine is started. Resources are pre-initialized on the target node. When the scheduling end receives the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and schedules the virtual machine startup request to the target node, so that the target node does not need to initialize the resources again, and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup and further improving the elasticity of the cloud computing system.

[0153] 8 , which shows a flowchart of a task execution method provided by an embodiment of the present disclosure. The method is applied to a scheduling end of a cloud computing system and includes the following specific steps:

[0154] Step 802: Receive a task processing request for a target task sent by the front end.

[0155] Step 804: In response to the task processing request, a virtual machine startup request is generated, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine.

[0156] Step 806: Query the target node corresponding to the first resource specification parameter from the node resource configuration table of the cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter.

[0157] Step 808: dispatching the virtual machine startup request to the target node, so that the target node starts the target virtual machine based on the initialized resources and executes the target task on the target virtual machine.

[0158] Step 810: Receive the task execution result of the target task fed back by the target node, and feed back the task execution result to the front end.

[0159] The embodiment of the present disclosure is applied to the scheduling end of the cloud computing system. The scheduling end is a functional module for scheduling instruction requests. The scheduling end records the node resource configuration table of the cloud computing system. After obtaining the virtual machine startup request, it is reasonably scheduled to the distributed deployed node, and the virtual machine startup and computing task are completed on the node.

[0160] Target tasks are computing tasks that require distributed execution, including but not limited to: neural network model training tasks, neural network model training tasks (such as question-answering tasks, text processing tasks, and image processing tasks), graph computing tasks, and data stream processing tasks. Task processing requests are instruction requests for executing target tasks based on the cloud computing system. Task execution results are the results obtained by the cloud computing system executing the target tasks.

[0161] In the embodiment of the present disclosure, the specific methods of steps 804 to 808 refer to steps 102 to 106 and are not repeated here.

[0162] Exemplarily, a task processing request for a game data stream task sent by a game operator from a front end of a cloud computing system is received, and in response to the task processing request, a virtual machine startup request is generated. Twelve target nodes including pre-initialized resources corresponding to first resource specification parameters (CPU quantity (1); GPU quantity (4); model (A100); memory capacity (8GB); disk capacity (512GB); disk type (SCSI); I / O controller type (LSI Logic); operating system version (Centos 7 64-bit)) are queried from a node resource configuration table of the cloud computing system, and the virtual machine startup request is dispatched to the 12 target nodes so that the 12 target nodes start the target virtual machines based on the initialized resources and execute the game data stream task on the target virtual machines. The task execution results of the game data stream task fed back by the 12 target nodes are received, and the task execution results are fed back to the front end of the cloud computing system.

[0163] In an embodiment of the present disclosure, a task processing request for a target task is received from a front end; in response to the task processing request, a virtual machine startup request is generated, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine; a target node corresponding to the first resource specification parameter is queried from a node resource configuration table of a cloud computing system, wherein the target node includes resources pre-initialized according to the first resource specification parameter; the virtual machine startup request is dispatched to the target node so that the target node starts the target virtual machine based on the initialized resources and executes the target task on the target virtual machine; the task execution result of the target task fed back by the target node is fed back to the front end. Resources are pre-initialized on multiple nodes. When the scheduling end obtains the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and dispatches the virtual machine startup request to the target node so that the target node does not need to initialize the resources again and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup, thereby improving the elasticity of the cloud computing system, realizing highly elastic task processing, and improving the efficiency, stability and availability of task processing.

[0164] The following further illustrates the system startup method provided by the present disclosure, taking its application in e-commerce live streaming as an example, in conjunction with FIG9 . FIG9 shows a flowchart of a process for a system startup method applied to an e-commerce live streaming data stream, provided by one embodiment of the present disclosure. The method is applied to the scheduling end of a cloud computing system and includes the following specific steps:

[0165] Step 902: For any historical virtual machine, when a virtual machine stop request is obtained for any historical virtual machine, the node state of the first node that starts any historical virtual machine is reset, and the resources of the first node are initialized based on the sample resource specification parameters required to start any historical virtual machine. The resource specification parameters of the node after initialization are recorded in the node resource configuration table of the cloud computing system.

[0166] Step 904: For any sample virtual machine, when a virtual machine creation request for any sample virtual machine is obtained, any sample virtual machine is created, the sample resource specification parameters required to start any sample virtual machine and the idle second node are determined, and the resources of the second node are initialized based on the sample resource specification parameters, and the resource specification parameters of the initialized node are recorded in the node resource configuration table of the cloud computing system.

[0167] Step 906: For at least one sample virtual machine, obtain sample resource specification parameters and distribution parameters of at least one sample virtual machine, and use the prediction model to perform prediction based on the sample resource specification parameters and distribution parameters to obtain the predicted resource specification parameters required to start any target virtual machine. Based on the predicted resource specification parameters, initialize the resources of the idle third node, and record the resource specification parameters of the initialized node into the node resource configuration table of the cloud computing system.

[0168] Step 908: For the target virtual machine that processes the e-commerce live broadcast data stream, receive a virtual machine startup request carrying the first resource specification parameters required to start the target virtual machine.

[0169] Step 910: Query the target node corresponding to the first resource specification parameter from the node resource configuration table of the cloud computing system, and schedule the virtual machine startup request to the target node, so that the target node starts the target virtual machine based on the initialized resources to process the e-commerce live broadcast data stream.

[0170] Step 912: If the target node corresponding to the first resource specification parameter is not found, an idle node is identified and the virtual machine startup request is scheduled to the idle node so that the idle node initializes the resources based on the first resource specification parameter, and based on the initialized resources, the target virtual machine is started to process the e-commerce live broadcast data stream.

[0171] Step 914: If there is no idle node, determine a reference node that includes resources that have been pre-initialized and the virtual machine has not been started, schedule the virtual machine startup request to the reference node, so that the reference node resets the resources of the reference node, initializes the processing resources based on the first resource specification parameters, and starts the target virtual machine to process the e-commerce live broadcast data stream.

[0172] In the embodiment of the present disclosure, different initialization strategies are used to complete the initialization processing of resources on the node in advance. When a virtual machine startup request is received for a target virtual machine that processes an e-commerce live broadcast data stream, the target node corresponding to the first resource specification parameter carried in the virtual machine startup request is queried from the node resource configuration table of the cloud computing system, and the virtual machine startup request is directly scheduled to the target node. Based on the resources that have been initialized, the startup of the target virtual machine is realized to process the e-commerce live broadcast data stream. There is no need to initialize the resources again. The target virtual machine can be quickly started based on the initialized resources, which improves the efficiency of system startup, thereby improving the elasticity of the cloud computing system, and at the same time improving the data processing efficiency and stability of the e-commerce live broadcast data stream.

[0173] Corresponding to the above method embodiment, the present disclosure also provides a system startup device embodiment. Figure 10 shows a schematic diagram of the structure of a system startup device provided by one embodiment of the present disclosure. As shown in Figure 10, the device is applied to the scheduling end of the cloud computing system, and the device includes:

[0174] The acquisition module 1002 is configured to acquire a virtual machine startup request, wherein the virtual machine startup request carries a first resource specification parameter required to start a target virtual machine;

[0175] A query module 1004 is configured to query a target node corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, wherein the target node includes a resource that has been pre-initialized according to the first resource specification parameter;

[0176] The first scheduling module 1006 is configured to schedule the virtual machine startup request to the target node, so that the target node starts the target virtual machine based on the initialized resources.

[0177] Optionally, the device further comprises:

[0178] The second scheduling module is configured to identify whether there is an idle node when the target node corresponding to the first resource specification parameter is not queried, where the idle node is a node that has not been initialized; if so, schedule the virtual machine startup request to the idle node so that the resources of the idle node are initialized based on the first resource specification parameter, and start the target virtual machine based on the initialized resources.

[0179] Optionally, the device further comprises:

[0180] The third scheduling module is configured to determine a reference node if it does not exist, wherein the reference node includes resources that have been initialized in advance and the virtual machine has not been started; schedule the virtual machine startup request to the reference node so that the reference node resets the resources of the reference node, initializes the processing resources based on the first resource specification parameters, and starts the target virtual machine.

[0181] Optionally, the device further comprises:

[0182] The pre-initialization module is configured to initialize the resources of the nodes in the cloud computing system based on the sample resource specification parameters; and record the resource specification parameters of the nodes after the initialization processing into the node resource configuration table of the cloud computing system.

[0183] Optionally, the pre-initialization module is further configured to:

[0184] For any historical virtual machine, when a virtual machine stop request for any historical virtual machine is obtained, the resource status of the first node that starts any historical virtual machine is reset; based on the sample resource specification parameters required to start any historical virtual machine, the resources of the first node are initialized.

[0185] Optionally, the pre-initialization module is further configured to:

[0186] For any sample virtual machine, upon receiving a virtual machine creation request for any sample virtual machine, create any sample virtual machine; determine the sample resource specification parameters and idle second nodes required to start any sample virtual machine; and initialize the resources of the second node based on the sample resource specification parameters.

[0187] Optionally, the pre-initialization module is further configured to:

[0188] For at least one sample virtual machine, obtain sample resource specification parameters and distribution parameters of at least one sample virtual machine; based on the sample resource specification parameters and distribution parameters, use the prediction model to perform prediction to obtain the predicted resource specification parameters required to start any target virtual machine; based on the predicted resource specification parameters, initialize the resources of the idle third node.

[0189] Optionally, the distribution parameter includes at least one of a time distribution parameter and a space distribution parameter;

[0190] Correspondingly, the pre-initialization module is further configured as follows:

[0191] Based on the sample resource specification parameters and time distribution parameters, the prediction model is used to perform time regression prediction on the target virtual machine startup request obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request; or, based on the sample resource specification parameters and spatial distribution parameters, the prediction model is used to perform spatial regression prediction on the target virtual machine startup request obtained within the prediction space, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request; or, based on the sample resource specification parameters, time distribution parameters and spatial distribution parameters, the prediction model is used to perform spatiotemporal regression prediction on the target virtual machine startup request obtained within the prediction space within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request.

[0192] In an embodiment of the present disclosure, a virtual machine startup request is obtained, wherein the virtual machine startup request carries the first resource specification parameter required to start the target virtual machine; the target node corresponding to the first resource specification parameter is queried from the node resource configuration table of the cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter; and the virtual machine startup request is dispatched to the target node so that the target node starts the target virtual machine based on the initialized resources. Resources are pre-initialized on multiple nodes. When the scheduling end obtains the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and dispatches the virtual machine startup request to the target node so that the target node does not need to initialize the resources again and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup and further improving the elasticity of the cloud computing system.

[0193] The above is a schematic scheme of a system startup device of this embodiment. It should be noted that the technical solution of the system startup device and the technical solution of the system startup method described above are based on the same concept. For details not described in detail in the technical solution of the system startup device, please refer to the description of the technical solution of the system startup method described above.

[0194] Corresponding to the above method embodiment, the present disclosure also provides a system startup device embodiment. Figure 11 shows a schematic diagram of the structure of another system startup device provided by one embodiment of the present disclosure. As shown in Figure 11, the device is applied to the target node of the cloud computing system, and the device includes:

[0195] The receiving module 1102 is configured to receive a virtual machine startup request scheduled by the scheduling end, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine, and the first resource specification parameter is a resource specification parameter of a resource on the target node;

[0196] The startup module 1104 is configured to start the target virtual machine based on the initialized resources.

[0197] In an embodiment of the present disclosure, a virtual machine startup request scheduled by a scheduling end is received, wherein the virtual machine startup request carries a first resource specification parameter required to start a target virtual machine, and the first resource specification parameter is a resource specification parameter of a resource on a target node; based on the initialized resources, the target virtual machine is started. Resources are pre-initialized on the target node. When the scheduling end receives the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and schedules the virtual machine startup request to the target node, so that the target node does not need to initialize the resources again, and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup and further improving the elasticity of the cloud computing system.

[0198] The above is a schematic scheme of a system startup device of this embodiment. It should be noted that the technical solution of the system startup device and the technical solution of the system startup method described above are based on the same concept. For details not described in detail in the technical solution of the system startup device, please refer to the description of the technical solution of the system startup method described above.

[0199] Corresponding to the above method embodiments, the present disclosure also provides a task execution device embodiment. FIG12 shows a schematic diagram of the structure of a task execution device provided by an embodiment of the present disclosure. As shown in FIG12, the device is applied to the scheduling end of the cloud computing system, and the device includes:

[0200] The request receiving module 1202 is configured to receive a task processing request for a target task sent by the front end;

[0201] The request generating module 1204 is configured to generate a virtual machine startup request in response to the task processing request, wherein the virtual machine startup request carries a first resource specification parameter required for starting the target virtual machine;

[0202] The node query module 1206 is configured to query a target node corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, wherein the target node includes a resource that has been pre-initialized according to the first resource specification parameter;

[0203] The request scheduling module 1208 is configured to schedule the virtual machine startup request to the target node, so that the target node starts the target virtual machine based on the initialized resources and executes the target task on the target virtual machine;

[0204] The result forwarding module 1210 is configured to receive the task execution result of the target task fed back by the target node, and feed back the task execution result to the front end.

[0205] In an embodiment of the present disclosure, a task processing request for a target task is received from a front end; in response to the task processing request, a virtual machine startup request is generated, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine; a target node corresponding to the first resource specification parameter is queried from a node resource configuration table of a cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter; the virtual machine startup request is dispatched to the target node so that the target node starts the target virtual machine based on the initialized resources and executes the target task on the target virtual machine; the task execution result of the target task fed back by the target node is fed back to the front end. Resources are pre-initialized on multiple nodes. When the scheduling end obtains the virtual machine startup request, the scheduling end determines the corresponding target node from the node resource configuration table of the cloud computing system based on the first resource specification parameter required to start the target virtual machine, and dispatches the virtual machine startup request to the target node so that the target node does not need to initialize the resources again and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup, thereby improving the elasticity of the cloud computing system, achieving highly elastic task processing, and improving the efficiency, stability and availability of task processing.

[0206] The above is a schematic scheme of a task execution device of this embodiment. It should be noted that the technical scheme of the task execution device and the technical scheme of the task execution method described above are of the same concept. For details not described in detail in the technical scheme of the task execution device, please refer to the description of the technical scheme of the task execution method described above.

[0207] Corresponding to the above-mentioned method embodiment, the present disclosure also provides a cloud computing system embodiment. FIG13 shows a schematic diagram of the structure of a cloud computing system provided by one embodiment of the present disclosure. As shown in FIG13 , the cloud computing system includes a scheduling terminal 1302 and a target node 1304. The target node 1304 includes resources that have been pre-initialized according to first resource specification parameters.

[0208] The scheduling end 1302 is configured to obtain a virtual machine startup request, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine, query a target node 1304 corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, and schedule the virtual machine startup request to the target node 1304;

[0209] The target node 1304 is configured to receive a virtual machine startup request scheduled by the scheduling end 1302 and to start the target virtual machine based on the initialized resources.

[0210] Optionally, the scheduling end 1302 is further configured to initialize nodes in the cloud computing system based on the sample resource specification parameters, and record the resource specification parameters of the initialized nodes into a node resource configuration table of the cloud computing system.

[0211] In an embodiment of the present disclosure, a cloud computing system includes a scheduling end and a target node, wherein the target node includes resources that have been pre-initialized according to first resource specification parameters; the scheduling end is used to obtain a virtual machine startup request, wherein the virtual machine startup request carries the first resource specification parameters required to start the target virtual machine, queries the target node corresponding to the first resource specification parameters from the node resource configuration table of the cloud computing system, and schedules the virtual machine startup request to the target node; the target node is used to receive the virtual machine startup request scheduled by the scheduling end, and starts the target virtual machine based on the initialized resources. The resources of the node have been pre-initialized, and when the scheduling end obtains the virtual machine startup request, based on the first resource specification parameters required to start the target virtual machine, it determines the corresponding target node from the node resource configuration table of the cloud computing system, and schedules the virtual machine startup request to the target node. The target node does not need to initialize the resources again, and can quickly start the target virtual machine based on the initialized resources, thereby improving the efficiency of system startup and thereby improving the elasticity of the cloud computing system.

[0212] The above is a schematic diagram of a cloud computing system according to this embodiment. It should be noted that the technical solution of the cloud computing system and the technical solution of the system startup method described above are based on the same concept. For details not described in detail in the technical solution of the cloud computing system, please refer to the description of the technical solution of the system startup method described above.

[0213] Figure 14 shows a block diagram of a computing device according to an embodiment of the present disclosure. Components of the computing device 1400 include, but are not limited to, a memory 1410 and a processor 1420. The processor 1420 is connected to the memory 1410 via a bus 1430, and a database 1450 is used to store data.

[0214] Computing device 1400 also includes an access device 1440 that enables computing device 1400 to communicate via one or more networks 1460. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. Access device 1440 may include one or more of any type of network interface (e.g., a network interface controller (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, or a near field communication (NFC) interface.

[0215] In one embodiment of the present disclosure, the aforementioned components of computing device 1400 and other components not shown in FIG14 may also be connected to each other, for example, via a bus. It should be understood that the block diagram of the computing device structure shown in FIG14 is for illustrative purposes only and does not limit the scope of the present disclosure. Those skilled in the art may add or replace other components as needed.

[0216] Computing device 1400 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, personal digital assistant, laptop computer, notebook computer, netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or personal computer (PC). Computing device 1400 may also be a mobile or stationary server.

[0217] The processor 1420 is configured to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the above-mentioned system startup method or task execution method.

[0218] The above is a schematic diagram of a computing device according to this embodiment. It should be noted that the technical solution of this computing device is based on the same concept as the technical solutions of the aforementioned system startup method and task execution method. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solutions of the aforementioned system startup method or task execution method.

[0219] An embodiment of the present disclosure further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned system startup method or task execution method.

[0220] The above is a schematic diagram of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium is based on the same concept as the technical solutions of the aforementioned system startup method and task execution method. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solutions of the aforementioned system startup method or task execution method.

[0221] An embodiment of the present disclosure further provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned system startup method or task execution method.

[0222] The above is a schematic diagram of a computer program according to this embodiment. It should be noted that the technical solution of this computer program is based on the same concept as the technical solutions of the aforementioned system startup method and task execution method. For details not described in detail in the technical solution of the computer program, please refer to the description of the technical solutions of the aforementioned system startup method or task execution method.

[0223] The foregoing description describes specific embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0224] The computer instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0225] It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present disclosure are not limited by the order of the actions described, because according to the embodiments of the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of the present disclosure.

[0226] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0227] The preferred embodiments of the present disclosure disclosed above are only used to help illustrate the present disclosure. The optional embodiments do not describe all details in detail, nor do they limit the invention to only the specific embodiments described. Obviously, many modifications and variations can be made based on the content of the embodiments of the present disclosure. The present disclosure selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of the present disclosure, so that those skilled in the art can better understand and utilize the present disclosure. The present disclosure is limited only by the claims and their full scope and equivalents.

Claims

1. A system startup method, applied to a scheduling end of a cloud computing system, the method comprising: Obtaining a virtual machine startup request, wherein the virtual machine startup request carries a first resource specification parameter required to start a target virtual machine; Querying a target node corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, wherein the target node includes resources that have been pre-initialized according to the first resource specification parameter; The virtual machine startup request is dispatched to the target node, so that the target node starts the target virtual machine based on the initialized resources.

2. The method according to claim 1, after querying the target node corresponding to the first resource specification parameter from the node resource configuration table of the cloud computing system, further comprises: In the case that the target node corresponding to the first resource specification parameter is not found, identifying whether there is an idle node, wherein the idle node is a node that has not been initialized; If so, the virtual machine startup request is scheduled to the idle node, so that the resources of the idle node are initialized based on the first resource specification parameter, and the target virtual machine is started based on the initialized resources.

3. The method according to claim 2, after identifying whether there is an idle node, further comprising: If not, determining a reference node, wherein the reference node includes resources that have been pre-initialized and the virtual machine has not been started; The virtual machine startup request is dispatched to the reference node, so that the reference node resets the resources of the reference node, initializes and processes the resources based on the first resource specification parameters, and starts the target virtual machine.

4. The method according to any one of claims 1 to 3, before querying the target node corresponding to the first resource specification parameter from the node resource configuration table of the cloud computing system, further comprising: Initializing resources of nodes in the cloud computing system based on sample resource specification parameters; The resource specification parameters of the node after the initialization process are recorded in the node resource configuration table of the cloud computing system.

5. The method according to claim 4, wherein the initializing processing of the resources of the nodes in the cloud computing system based on the sample resource specification parameters comprises: For any historical virtual machine, when a virtual machine stop request for the any historical virtual machine is obtained, resetting the resource state of the first node that starts the any historical virtual machine; Based on the sample resource specification parameters required to start any of the historical virtual machines, the resources of the first node are initialized.

6. The method according to claim 4, wherein the initializing processing of the resources of the nodes in the cloud computing system based on the sample resource specification parameters comprises: For any sample virtual machine, when a virtual machine creation request for the any sample virtual machine is obtained, create the any sample virtual machine; Determine sample resource specification parameters and idle second nodes required to start any of the sample virtual machines; Based on the sample resource specification parameters, the resources of the second node are initialized.

7. The method according to claim 4, wherein the initializing processing of the resources of the nodes in the cloud computing system based on the sample resource specification parameters comprises: For at least one sample virtual machine, obtaining sample resource specification parameters and distribution parameters of the at least one sample virtual machine; Based on the sample resource specification parameters and the distribution parameters, a prediction model is used to perform prediction to obtain the predicted resource specification parameters required to start any target virtual machine; Based on the predicted resource specification parameters, the idle resources of the third node are initialized.

8. The method according to claim 7, wherein the distribution parameter comprises at least one of a time distribution parameter and a space distribution parameter; The step of using a prediction model to predict based on the sample resource specification parameters and the distribution parameters to obtain the predicted resource specification parameters required to start any target virtual machine includes: Based on the sample resource specification parameters and the time distribution parameters, a prediction model is used to perform time regression prediction on the target virtual machine startup request obtained within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request; or, Based on the sample resource specification parameters and the spatial distribution parameters, a prediction model is used to perform spatial regression prediction on the target virtual machine startup request that obtains the prediction space, and the prediction resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request; or, Based on the sample resource specification parameters, the time distribution parameters and the space distribution parameters, a prediction model is used to perform spatiotemporal regression prediction on the target virtual machine startup request that obtains the prediction space within the prediction time, and the predicted resource specification parameters required to start any target virtual machine are obtained from the target virtual machine startup request.

9. A system startup method, applied to a target node of a cloud computing system, wherein the target node includes a resource that has been pre-initialized according to a first resource specification parameter, comprising: Receive a virtual machine startup request scheduled by a scheduling end, wherein the virtual machine startup request carries a first resource specification parameter required to start a target virtual machine, and the first resource specification parameter is a resource specification parameter of a resource on the target node; Based on the pre-initialized resources, the target virtual machine is started.

10. A task execution method, applied to a scheduling end of a cloud computing system, comprising: Receive task processing requests for target tasks sent by the front end; In response to the task processing request, generating a virtual machine startup request, wherein the virtual machine startup request carries a first resource specification parameter required to start the target virtual machine; Querying a target node corresponding to the first resource specification parameter from a node resource configuration table of the cloud computing system, wherein the target node includes a resource that has been pre-initialized according to the first resource specification parameter; Dispatching the virtual machine startup request to the target node, so that the target node starts the target virtual machine based on the initialized resources and executes the target task on the target virtual machine; Receive the task execution result of the target task fed back by the target node, and feed back the task execution result to the front end.

11. A cloud computing system, the system comprising a scheduling end and a target node, the target node comprising a resource that has been pre-initialized according to a first resource specification parameter; The scheduling end is used to obtain a virtual machine startup request, wherein: The virtual machine startup request carries a first resource specification parameter required to start the target virtual machine, queries the target node corresponding to the first resource specification parameter from the node resource configuration table of the cloud computing system, and schedules the virtual machine startup request to the target node; The target node is used to receive a virtual machine startup request scheduled by a scheduling end, and to start the target virtual machine based on the initialized resources.

12. According to the cloud computing system of claim 11, the scheduling end is also used to initialize the nodes in the cloud computing system based on sample resource specification parameters, and record the resource specification parameters of the initialized nodes into the node resource configuration table of the cloud computing system.

13. A computing device comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, implement the steps of the method according to any one of claims 1 to 10.

15. A computer program, when executed in a computer, causes the computer to execute the steps of the method according to any one of claims 1 to 10.

Images