Calculation power scheduling method and system

Abstract

The invention discloses a computing power scheduling method and system. The method comprises the following steps: under a computing power scheduling task, aiming at configuration information, task resources and workflow definition selected by a user, determining a computing power business demand; target scheduling paths of the computing power service requirements under different preset dimensions are determined according to a preset scheduling mechanism on the basis of the whole network computing power resource information, the computing power scheduling task is completed on the basis of the target scheduling paths, and the preset dimensions comprise one of the cost dimension and the distance dimension. According to the scheme, the scheduling of the computing power resources can be realized.

Description

Technical Field

[0001] The present invention belongs to the field of computing power scheduling, and in particular relates to a computing power scheduling method and system. Background Art

[0002] Artificial intelligence platforms need to handle a large number of computing tasks, including model training, reasoning, etc. However, different computing tasks have different requirements for computing power. If inappropriate computing resources are allocated to computing tasks, the computing tasks may not be executed due to insufficient resources.

[0003] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0004] In order to solve the deficiencies in the prior art, the present invention provides a computing power scheduling method and system to solve the technical problem that inappropriate computing resources are allocated to computing tasks, resulting in the inability to execute computing tasks due to insufficient resources.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions.

[0006] The present invention first discloses a computing power scheduling method, which includes the following steps:

[0007] Under the computing power scheduling task, determine the computing power business requirements based on the configuration information, task resources, and workflow definitions selected by the user;

[0008] Based on the computing power resource information of the entire network, the target scheduling path of the computing power business demand under different preset dimensions is determined according to the preset scheduling mechanism, and the computing power scheduling task is completed based on the target scheduling path, wherein the preset dimension includes one of the cost dimension and the distance dimension.

[0009] The present invention further includes the following preferred embodiments:

[0010] Before the step of determining the target scheduling path of the computing power service demand under different preset dimensions according to the preset scheduling mechanism based on the computing power resource information of the whole network, the method further includes:

[0011] Receive computing power feature information, computing power service information, and network indicator information fed back by computing power nodes;

[0012] Determine the latency and path information between different computing nodes;

[0013] Inputting the computing power feature information, the computing power service information, the network indicator information, the latency information and the path information into a target prediction model to obtain resource information of a computing power node;

[0014] Determine the computing power resource information of the entire network based on the resource information of the computing power node.

[0015] The determining the computing power resource information of the entire network according to the resource information of the computing power node further includes:

[0016] The CFN protocol based on the distributed architecture receives the resource information of the computing nodes sent by the CFN router and determines the computing resource information of the entire network.

[0017] The distributed architecture is an intelligent distributed deployment synthesis component, which integrates PXE service, DHCP service, TFTP service and MySql service in the central deployment node; wherein,

[0018] The PXE service is responsible for network booting in the initial stage of node startup;

[0019] The DHCP service matches the MAC addresses of deployed nodes and issues IP addresses;

[0020] The TFTP service is responsible for transmitting the boot file, wherein the boot file includes the kernel file and the initialization root file system;

[0021] The MySql service is responsible for recording node deployment status to facilitate system management and maintenance.

[0022] The distributed architecture is used to automatically generate the network configuration and storage configuration of the node according to the hardware information of the cluster, and generate the configuration of Kolla, the deployment parameter of OpenStack, according to the optimized placement of the registry, and deploy the installation package after obtaining the components from the official GitHub of OpenStack. Finally, the node preparation and configuration are completed, and Kolla is used to deploy it into an OpenStack cluster with one click.

[0023] The CFN protocol based on the distributed architecture receives the resource information of the computing nodes sent by the CFN router and determines the computing resource information of the whole network, further comprising:

[0024] Based on a preset collection method, the resource information of the computing power node is collected through the CFN router, and then the resource information of the computing power node sent by the CFN router is received, wherein the preset collection method includes a first collection method and a second collection method, the first collection method is a method in which the computing power node registers the resource information to the CFN router, and the second collection method is a method in which the CFN router periodically collects information.

[0025] The preset scheduling mechanism includes a first scheduling mechanism and a second scheduling mechanism, wherein the first scheduling mechanism is a scheduling mechanism based on a virtual machine, and the second scheduling mechanism is a scheduling mechanism based on a container; wherein,

[0026] The method further includes: determining a target scheduling path for the computing power service demand under different preset dimensions based on the computing power resource information of the entire network according to a preset scheduling mechanism, and completing the computing power scheduling task based on the target scheduling path.

[0027] Based on the computing power resource information of the entire network, determine the target scheduling path of the computing power business demand under different preset dimensions according to the first scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path; or

[0028] Based on the computing power resource information of the entire network, the target scheduling path of the computing power business demand under different preset dimensions is determined according to the second scheduling mechanism, and the computing power scheduling task is completed based on the target scheduling path.

[0029] The present invention also discloses a computing power scheduling system using the aforementioned computing power scheduling method, comprising:

[0030] The determination module is used to determine the computing power business requirements based on the configuration information, task resources and workflow definitions selected by the user under the computing power scheduling task;

[0031] The completion module is used to determine the target scheduling path of the computing power business demand under different preset dimensions based on the computing power resource information of the entire network and according to the preset scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path, wherein the preset dimension includes one of the cost dimension and the distance dimension.

[0032] Accordingly, the present application also discloses a device, including a processor and a storage medium;

[0033] The storage medium is used to store instructions;

[0034] The processor is used to operate according to the instructions to execute the steps of the aforementioned computing power scheduling method.

[0035] Correspondingly, the present application also discloses a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the aforementioned computing power scheduling method are implemented.

[0036] The beneficial effect of the present invention lies in that, compared with the prior art, the present invention provides a computing power scheduling method and system, which determines the computing power business demand based on the configuration information, task resources and workflow definition selected by the user; based on the computing power resource information of the entire network, determines the target scheduling path of the computing power business demand under different preset dimensions according to the preset scheduling mechanism, and completes the computing power scheduling task based on the target scheduling path, thereby solving the technical problem that the computing task is unable to be executed due to insufficient resources due to the allocation of inappropriate computing resources to the computing task, making the execution of computing power scheduling more efficient and reliable, thereby avoiding the inability to execute certain computing tasks due to insufficient resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of the first embodiment of the computing power scheduling method of the present application is provided;

[0038] Figure 2 A flowchart of the second embodiment of the computing power scheduling method of the present application is provided;

[0039] Figure 3 A schematic diagram of the computing environment migration of virtual machines and containers provided in Example 2 of the present application;

[0040] Figure 4 This is a schematic diagram of the module structure of the computing power scheduling system according to an embodiment of the present application;

[0041] Figure 5 Schematic diagram of the device structure of the hardware operating environment involved in the computing power scheduling method in the embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention more clear, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0043] The embodiments described in this application are only some embodiments of the present invention, not all embodiments. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the protection scope of the present invention.

[0044] In view of the deficiencies of the prior art, the present invention proposes a computing power scheduling method and system, which determines the computing power business demand according to the configuration information, task resources and workflow definition selected by the user under the computing power scheduling task; based on the computing power resource information of the whole network, determines the target scheduling path of the computing power business demand under different preset dimensions according to the preset scheduling mechanism, and completes the computing power scheduling task based on the target scheduling path. The preset dimension includes one of the cost dimension and the distance dimension.

[0045] See also Figure 1As shown, the computing power scheduling method disclosed in the present invention includes the following steps:

[0046] Step S10, under the computing power scheduling task, determine the computing power business requirements based on the configuration information, task resources and workflow definition selected by the user.

[0047] Specifically, the scheduling center classifies and displays the computing power of the entire platform, and counts the number of data centers currently connected, the number of successfully running scheduling tasks, the number of scheduling requests, the number of resources, the number of workflow models, etc., to provide users with a clear resource overview, allowing users to judge and choose the computing power required for their business based on their actual needs.

[0048] Step S20, based on the computing power resource information of the entire network, determine the target scheduling path of the computing power business demand under different preset dimensions according to the preset scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path, wherein the preset dimension includes one of the cost dimension and the distance dimension.

[0049] Among them, the computing power resource information of the entire network refers to the resource information of all computing power nodes in the entire network, and the preset scheduling mechanisms include virtual machine-based scheduling mechanisms and container-based scheduling mechanisms.

[0050] In the specific implementation, the steps for determining the target scheduling path are as follows:

[0051] Step S21, enter the scheduling task creation process by selecting the corresponding data center, create a scheduling task, and display the connectivity between data centers in the form of visual animation.

[0052] Step S22: After entering the task process, select the task scenario and computing power configuration according to the actual needs of the user and select or create a project. The scheduled task will run in the project created or selected by the user. If no resource configuration is specified, the platform will flexibly schedule according to the idle status of all resources.

[0053] Step S23: Upload the task resources that need to be run according to the prompts. The platform provides two resource modes: file resources and data source. After successful upload, the resources will be managed uniformly in the resource center.

[0054] Step S24: The platform supports a flexible drag-and-drop creation method, selects a workflow model, and provides model creation for various big data workflows such as SHELL, SQL, FLINK, and application scripts such as python and java.

[0055] Step S25: Based on the configuration information, resource size, and workflow definition selected by the user, a target scheduling path is recommended by various dimensions such as cost, distance, and comprehensive conditions.

[0056] In the specific implementation, the unified interface definition and implementation of data scheduling and task scheduling can be completed based on general computing frameworks such as general AI training / reasoning (Tensorflow / Pytorch), big data analysis (Spark / Flink), and scientific computing (Vasp). Specifically, general AI training (Tensorflow / Pytorch) has two implementation forms in the computing power operation platform: encapsulate the parameters and data needed for the model and pass them to the model, select the container image and the area where the computing power resources are located, copy the model file to the container, and then execute, obtain the execution results and monitor; in addition, the model training can be run in the form of container images, provided that the user has encapsulated the model into the container image in advance in the established form. General AI reasoning (Tensorflow / Pytorch) deploys the trained and parameterized model to the computing power operation platform in the form of k8s service, exposing the corresponding port to accept external service requests. Big data analysis (Spark / Flink) and scientific computing (Vasp) are similar to general AI training. All of the above are completed by writing programs in Java computer language to define and implement the unified interface for data scheduling and task scheduling.

[0057] In the specific implementation, the cross-domain resource management (monitoring and management of storage resources) of the Paas layer realizes the unified monitoring and management of functions such as alarms, permissions, reports, and linkage control by deploying distributed collection units and monitoring systems. Specifically, for example, the monitoring data collection uses the open source solution categralf, and the monitoring system uses the cloud monitoring module integrated in the computing power operation platform, which can realize the linkage functions such as viewing the monitoring data in chart form, alarms and notifications, reports, and simple automatic fault handling.

[0058] The computing power scheduling method proposed in this application solves the technical problem that inappropriate computing resources are allocated to computing tasks, resulting in the inability to execute computing tasks due to insufficient resources. The computing power business requirements are determined based on the configuration information, task resources and workflow definitions selected by the user, and then a more appropriate target scheduling path is designed for the user based on the computing power business requirements, so that the execution of computing power scheduling is more efficient and reliable, thereby avoiding the inability to execute certain computing tasks due to insufficient resources.

[0059] In the second embodiment of the present application, refer to Figure 2 , before step S20, further comprising steps S201 to S204:

[0060] Step S201, receiving computing power feature information, computing power service information, and network indicator information fed back by computing power nodes;

[0061] Among them, computing power feature information may include the processing power, load information, deployment location, etc. of the Central Processing Unit (CPU) / Graphics Processing Unit (GPU); computing power service information may include service type and service duration, etc.; network indicator information may include latency, traffic, packet loss, etc.

[0062] Step S202, determining the delay information and path information between different computing power nodes;

[0063] Specifically, since computing power resources change dynamically, the computing power perception center also needs to measure the delay information and path information between different computing power nodes.

[0064] Step S203, inputting the computing power feature information, the computing power service information, the network indicator information, the delay information and the path information into a target prediction model to obtain resource information of the computing power node;

[0065] Preferably, the target prediction model is a prediction model that utilizes an AI traffic prediction model combined with an AI deep neural algorithm learning model. The target prediction model can timely predict the resource information of the computing power nodes, thereby improving the resource allocation speed and utilization efficiency of the entire network.

[0066] Step S204, determining the computing power resource information of the entire network based on the resource information of the computing power node.

[0067] In a preferred embodiment, determining the computing power resource information of the entire network based on the resource information of the computing power nodes further includes a CFN protocol based on a distributed architecture, receiving the resource information of the computing power nodes sent by the CFN router, and determining the computing power resource information of the entire network.

[0068] Specifically, the ultimate goal of the computing network is to provide users with computing resources in the form of application services. The computing resources are located at the infrastructure layer, and their physical location is generally different from that of the users. This requires the use of network functions to dispatch the tasks that users need to process to the computing resources. In the initial network, the location of the computing resources and the amount of resources they have are unknown to the users and the entire computing network. Communication messages are needed as carriers to interact in the network according to specific protocols to complete the sharing of computing resource information.

[0069] Specifically, in the TCP / IP architecture, as long as the authorized computing resource IP is reachable, these computing resources can be considered available. The communication protocol that carries the computing resource information can be located at any layer above the network layer (including the network layer), and the computing resource information is forwarded based on the IP message based on the network layer protocol.

[0070] Specifically, the computing power network controls and distributes computing resource information based on the CFN (Computing First Network) protocol, which can solve problems such as complex MEC deployment, low efficiency, and low resource reuse rate, enabling the network to have the ability to dynamically route built-in computing services.

[0071] The CFN protocol publishes the computing resource status and network status as routing information to the network, and routes the computing task message to the most suitable computing node based on the virtual service ID, so as to achieve the goal of optimal user experience, optimal computing resource utilization, and optimal network efficiency. The CFN protocol inherits the design ideas of the traditional label forwarding protocol, is carried on the IP network, establishes sessions between adjacent routers that support the CFN protocol, and publishes the acquired computing resource information to adjacent CFN routers with the help of the routing protocol, so as to realize the full network diffusion of computing resource information. At the same time, the CFN router builds a service routing information table according to different services, guides the business message to be forwarded with the service ID as the destination address, so as to realize the utilization of decentralized computing resources in the form of services.

[0072] Specifically, in response to the current difficulties in quickly replicating and deploying IAAS virtualized resource pools, PAAS cloud service platforms, and upper-layer applications in multiple data centers, a large-scale OpenStack deployment architecture with automatic placement of distributed warehouses is adopted, and containerized deployment of multiple warehouse registries is used to make deployment lighter and easier to operate. In a large-scale multi-data center heterogeneous environment, an OpenStack deployment architecture (i.e., distributed architecture) with intelligent distributed registries placement is adopted.

[0073] In a preferred embodiment, the distributed architecture is an intelligent distributed deployment synthesis component, such as an intelligent distributed registry deployment (IDRD) synthesis component. The intelligent distributed deployment synthesis component integrates Pre-boot Execution Environment (PXE) service, Dynamic Host Configuration Protocol (DHCP) service, Trivial File Transfer Protocol (TFTP) service and MySql service in the central deployment node; wherein the PXE service is responsible for network booting in the initial stage of node startup; the DHCP service matches the MAC address of the deployed node and issues the IP address; the TFTP service is responsible for transmitting the boot file, wherein the boot file includes the kernel file (vmlinuz) and the initialized root file system (initrd); the MySql service is responsible for recording information such as node deployment status to facilitate system management and maintenance.

[0074] Furthermore, multiple registries are set up in the distributed architecture and placed on different nodes in the multi-data center cluster. The registries are copies of the registry in the central server and process requests from cluster nodes to pull images through scheduling. The requests are distributed to different registries to achieve load balancing and relieve the pressure on the central server.

[0075] All nodes in the multi-data center cluster are linked together through a high-speed deployment network and a Baseboard Management Controller (BMC) control network. IDRD agent components are placed on each node to intelligently generate node configuration files and deployment files.

[0076] In a preferred embodiment, the distributed architecture is used to automatically generate the network configuration and storage configuration of the node according to the hardware information of the cluster, and generate the configuration of OpenStack deployment parameter Kolla according to the optimized placement of the registry, and deploy the installation package after obtaining the components from the official GitHub of OpenStack. Finally, the node preparation and configuration are completed, and Kolla is used to deploy it into an OpenStack cluster with one click.

[0077] Specifically, the distributed architecture includes three basic modules: the node preparation module, the automatic configuration file generation module, and the cluster deployment module. According to the hardware information of the cluster, the network configuration and storage configuration of the node are automatically generated, and according to the optimized placement of the registry, the configuration of Kolla, the deployment parameter of OpenStack, is generated. Bootstrap is a live system and will not have any impact on the server data when it starts. Therefore, in the node preparation node module, components are obtained from the official GitHub of OpenStack and packaged into a one-click deployment installation package to simplify operations in the later deployment process and install the OS for the bare metal server. Finally, the node preparation and configuration are completed, and Kolla is used to deploy it into an OpenStack cluster with one click.

[0078] In a preferred embodiment, the CFN protocol based on the distributed architecture receives resource information of computing nodes sent by the CFN router and determines the computing resource information of the entire network, further comprising:

[0079] Based on a preset collection method, the resource information of the computing power node is collected through the CFN router, and then the resource information of the computing power node sent by the CFN router is received, wherein the preset collection method includes a first collection method and a second collection method, the first collection method is a method in which the computing power node registers the resource information to the CFN router, and the second collection method is a method in which the CFN router periodically collects information.

[0080] In the computing network, to achieve the integration of computing resource information and the use of computing resource information anytime and anywhere, it is necessary to complete the synchronization of information across the entire network. The CFN router is responsible for collecting local computing resource information and spreading the information across the entire network through IP packets or routing protocol packets. All CFN routers generate service routing information tables locally based on the complete computing resource information obtained and combined with the network topology information to guide the forwarding of business packets. The specific implementation process is as follows:

[0081] ①CFN routers A and D complete the collection of local computing resource information. The collection process can be done by the local computing resource node registering the computing resource information to the CFN router, or by the CFN router periodically collecting information.

[0082] ②CFN routers A and D carry computing resource information in IP protocol or routing protocol and publish it to other CFN routers in the network to achieve network-wide information sharing;

[0083] ③CFN routers generate service routing information tables locally based on the network-wide information obtained and the network topology learned through routing protocols to guide the forwarding of service messages.

[0084] Routers B and C, as transit routers, do not need to support the CFN protocol because the computing resource information is carried in the IP protocol or routing protocol. B and C only need to forward the IP message or routing protocol message carrying the computing resource information without parsing the CFN related information in the message.

[0085] Different applications in the computing network have different requirements for computing resources. For example, the processing of two-dimensional images requires higher CPU, the processing of videos and AI requires higher GPU, and the processing of network messages requires higher NPU. Different service routing information entries will be generated on the computing network routers according to different application services and required computing resources. Each service routing information entry on each computing network router will be forwarded according to different guidance on computing resource requirements.

[0086] When the number of application services is huge and the network scale is huge, each router needs to obtain the entire network information for each application service and then independently calculate the path. The workload of maintaining the entire network is very huge. In addition, the research on the details of the CFN protocol on aggregation, interaction between IGP and BGP, and interaction between ASs is not yet mature. Therefore, in order to ensure the feasibility of the operation of the computing network, it is necessary to uniformly manage the computing network, centralize the synchronization of information and the calculation of paths, and send the service routing information table items to the router after calculation. The router is only responsible for forwarding business messages at the data level, which is consistent with the idea of ​​SDN.

[0087] In specific implementations, the CFN protocol can also be based on a centralized architecture. The difference from a distributed architecture is that routers do not need to communicate directly with each other, nor do they need to generate service routing information tables through local calculations. Instead, they only need to generate table entries locally to guide forwarding based on the entries sent down by the computing power network controller.

[0088] In the design of the centralized architecture, in order to determine whether to send the computing resource information directly to the computing network controller, which will perform the calculations uniformly, or to follow the idea in the distributed architecture and send the computing resource information to the router, which will then send it to the computing network controller. Compared with routers, considering the large number of computing resource nodes, if each computing resource node needs to communicate with the computing network controller, the pressure on the computing network controller will be too great. The router can continue to bear the responsibility of collecting computing resource information. The specific implementation process of the detailed centralized control architecture is as follows:

[0089] ① Routers A and D complete the collection of local computing resource information. The collection process can be done by the local computing resource node registering the computing resource information to the router, or by the router periodically collecting information.

[0090] ② Routers A and D carry computing resource information in IP protocol or routing protocol and publish it to the computing network controller;

[0091] ③ The computing network controller generates a service information flow table based on the complete computing resource information and the completed network topology calculation;

[0092] ④ The computing network controller sends the service information flow table to routers A and D;

[0093] ⑤ Routers A and D generate service information flow tables locally based on the computing power network controller information received to guide the forwarding of service messages.

[0094] In the specific implementation, since it is difficult to describe the internal skew of data, a data skew theory and a quantitative evaluation model were proposed. It was found that the non-uniformity of the key values ​​within the data and the long-tail effect are the main reasons for the low efficiency of data-intensive computing in supercomputing systems. Based on the Pareto / Zipf and other power-law distributions, a quantitative evaluation model, benchmark test set and test method were proposed, which made it possible to accurately calculate the impact of data skew on node load and improve scheduling efficiency.

[0095] In a preferred embodiment, the preset scheduling mechanism includes a first scheduling mechanism and a second scheduling mechanism, the first scheduling mechanism is a scheduling mechanism based on a virtual machine, and the second scheduling mechanism is a scheduling mechanism based on a container; wherein, based on the computing power resource information of the entire network, the target scheduling path of the computing power business demand under different preset dimensions is determined according to the preset scheduling mechanism, and the computing power scheduling task is completed based on the target scheduling path, further comprising:

[0096] Based on the computing power resource information of the entire network, determine the target scheduling path of the computing power business demand in different preset dimensions according to the first scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path; or, based on the computing power resource information of the entire network, determine the target scheduling path of the computing power business demand in different preset dimensions according to the second scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path.

[0097] Specifically, Figure 3 As shown in the figure, in the case of a surge in the number of concurrent user accesses, emergency response, etc., which leads to a surge in data computing volume, the standardized computing environment can be quickly replicated and deployed through virtual machine and container standardized environment migration technology and distributed architecture, so as to provide unified and elastic computing, storage, and network resource services to the outside world.

[0098] Specifically, virtual machine-based scheduling is implemented through OpenStack-Train and cloud management, and container-based scheduling is implemented based on the k8s computing power operation platform.

[0099] This application also provides a computing power scheduling system, please refer to Figure 4 , the computing power scheduling system includes:

[0100] The determination module 10 is used to determine the computing power business requirements according to the configuration information, task resources and workflow definition selected by the user under the computing power scheduling task;

[0101] Completion module 20 is used to determine the target scheduling path of the computing power business demand under different preset dimensions based on the computing power resource information of the entire network and in accordance with the preset scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path, wherein the preset dimension includes one of the cost dimension and the distance dimension.

[0102] The beneficial effect of the present invention lies in that, compared with the prior art, the present invention provides a computing power scheduling method and system, which determines the computing power business demand based on the configuration information, task resources and workflow definition selected by the user; based on the computing power resource information of the entire network, determines the target scheduling path of the computing power business demand under different preset dimensions according to the preset scheduling mechanism, and completes the computing power scheduling task based on the target scheduling path, thereby solving the technical problem that the computing task is unable to be executed due to insufficient resources due to the allocation of inappropriate computing resources to the computing task, making the execution of computing power scheduling more efficient and reliable, thereby avoiding the inability to execute certain computing tasks due to insufficient resources.

[0103] refer to Figure 5 , which shows a schematic diagram of the structure of a computing power scheduling device suitable for implementing the embodiment of the present application. The computing power scheduling device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The computing power scheduling device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0104] like Figure 5As shown, the computing power scheduling device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the xxx device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the computing power scheduling device to communicate with other devices wirelessly or by wire to exchange data.

[0105] Based on the spirit of the present invention, those skilled in the art can easily think that a computer program product can be obtained based on the aforementioned computing power scheduling method. The computer program product may include a computer-readable storage medium on which computer-readable program instructions are loaded for enabling a processor to implement various aspects of the present disclosure. That is, the present application also includes a device, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to perform the steps according to the aforementioned computing power scheduling method.

[0106] Computer readable storage medium can be a tangible device that can keep and store the instructions used by the instruction execution device. Computer readable storage medium can be, for example, - but not limited to - electrical storage device, magnetic storage device, optical storage device, electromagnetic storage device, semiconductor storage device or any suitable combination of the above. More specific examples (non-exhaustive list) of computer readable storage medium include: portable computer disk, hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, mechanical encoding device, for example, punch card or groove protrusion structure with instructions stored thereon and any suitable combination of the above. Computer readable storage medium used here is not interpreted as instantaneous signal itself, such as radio wave or other free propagating electromagnetic wave, electromagnetic wave propagated by waveguide or other transmission medium (for example, light pulse by optical fiber cable) or electrical signal transmitted by wire.

[0107] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0108] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages-such as Smalltalk, C++, etc., and conventional procedural programming languages-such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network-including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA) may be personalized by utilizing the state information of a computer-readable program instruction, and the electronic circuit may execute a computer-readable program instruction, thereby realizing various aspects of the present disclosure.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A computing power scheduling method, It is characterized in that The following steps are involved: Under the computing power scheduling task, determine the computing power business requirements based on the configuration information, task resources, and workflow definitions selected by the user; Based on the computing power resource information of the entire network, the target scheduling path of the computing power business demand under different preset dimensions is determined according to the preset scheduling mechanism, and the computing power scheduling task is completed based on the target scheduling path, wherein the preset dimension includes one of the cost dimension and the distance dimension.

2. The method according to claim 1, It is characterized in that Before the step of determining the target scheduling path of the computing power service demand under different preset dimensions according to the preset scheduling mechanism based on the computing power resource information of the whole network, the method further includes: Receive computing power feature information, computing power service information, and network indicator information fed back by computing power nodes; Determine the latency and path information between different computing nodes; Inputting the computing power feature information, the computing power service information, the network indicator information, the latency information and the path information into a target prediction model to obtain resource information of a computing power node; Determine the computing power resource information of the entire network based on the resource information of the computing power node.

3. The method according to claim 2, It is characterized in that The determining the computing power resource information of the entire network according to the resource information of the computing power node further includes: The CFN protocol based on the distributed architecture receives the resource information of the computing nodes sent by the CFN router and determines the computing resource information of the entire network.

4. The method according to claim 3, It is characterized in that The distributed architecture is an intelligent distributed deployment synthesis component, which integrates PXE service, DHCP service, TFTP service and MySql service in the central deployment node; wherein, The PXE service is responsible for network booting in the initial stage of node startup; The DHCP service matches the MAC addresses of deployed nodes and issues IP addresses; The TFTP service is responsible for transmitting the boot file, wherein the boot file includes the kernel file and the initialization root file system; The MySql service is responsible for recording node deployment status to facilitate system management and maintenance.

5. The method according to claim 3, It is characterized in that The distributed architecture is used to automatically generate the network configuration and storage configuration of the node according to the hardware information of the cluster, and generate the configuration of Kolla, the deployment parameter of OpenStack, according to the optimized placement of the registry, and deploy the installation package after obtaining the components from the official GitHub of OpenStack. Finally, the node preparation and configuration are completed, and Kolla is used to deploy it into an OpenStack cluster with one click.

6. The method according to claim 3, It is characterized in that The CFN protocol based on the distributed architecture receives the resource information of the computing nodes sent by the CFN router and determines the computing resource information of the whole network, further comprising: Based on a preset collection method, the resource information of the computing power node is collected through the CFN router, and then the resource information of the computing power node sent by the CFN router is received, wherein the preset collection method includes a first collection method and a second collection method, the first collection method is a method in which the computing power node registers the resource information to the CFN router, and the second collection method is a method in which the CFN router periodically collects information.

7. The method according to claim 1, It is characterized in that The preset scheduling mechanism includes a first scheduling mechanism and a second scheduling mechanism, wherein the first scheduling mechanism is a scheduling mechanism based on a virtual machine, and the second scheduling mechanism is a scheduling mechanism based on a container; wherein, The method further includes: determining a target scheduling path for the computing power service demand under different preset dimensions based on the computing power resource information of the entire network according to a preset scheduling mechanism, and completing the computing power scheduling task based on the target scheduling path. Based on the computing power resource information of the entire network, determine the target scheduling path of the computing power business demand under different preset dimensions according to the first scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path; or Based on the computing power resource information of the entire network, the target scheduling path of the computing power business demand under different preset dimensions is determined according to the second scheduling mechanism, and the computing power scheduling task is completed based on the target scheduling path.

8. A computing power scheduling system, It is characterized in that The system comprises: The determination module is used to determine the computing power business requirements based on the configuration information, task resources and workflow definitions selected by the user under the computing power scheduling task; The completion module is used to determine the target scheduling path of the computing power business demand under different preset dimensions based on the computing power resource information of the entire network and according to the preset scheduling mechanism, and complete the computing power scheduling task based on the target scheduling path, wherein the preset dimension includes one of the cost dimension and the distance dimension.

9. A device comprising a processor and a storage medium; Features: The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the computing power scheduling method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the steps of the computing power scheduling method described in any one of claims 1-7 are implemented.

Images