Cloud native resource scheduling deployment control system and method oriented to airborne environment

Abstract

The invention discloses an airborne environment-oriented cloud native resource scheduling deployment control system and method, and belongs to the field of implementation and technical development of a cloud native architecture in an airborne environment. Comprising a heterogeneous resource definition template, a heterogeneous resource module management component, a heterogeneous resource agent component, a heterogeneous resource controller component, a heterogeneous resource scheduler component and an interface server. Performing selected type heterogeneous resource definition by adopting a heterogeneous resource definition template; creating a resource instance in the interface server; the heterogeneous resource controller component performs resource control tuning; creating a function application instance in the interface server; the heterogeneous resource scheduler component makes a scheduling decision; the heterogeneous resource agent component reads the function deployment information and issues the function deployment information to the heterogeneous resource module management component; and the heterogeneous resource module management component performs program loading and unloading operation and information updating and reporting. According to the invention, the adaptability and operation efficiency of the container scheduling engine in the airborne heterogeneous computing resource environment can be improved.

Description

Technical Field

[0001] This invention relates to the field of cloud-native architecture implementation and technology development in airborne environments, and more specifically, to a cloud-native resource scheduling, deployment, control system, and method for airborne environments. Background Technology

[0002] As an emerging technological concept born in the era of cloud computing, cloud-native computing exhibits significant technological advantages compared to traditional computing technologies. It not only significantly enhances the elasticity, scalability, agility, reliability, fault tolerance, observability, and continuous update capabilities of applications, but also greatly reduces the complexity of system management and maintenance while ensuring security. These advantages enable cloud-native technologies to effectively reduce the technical costs of system operation, improve the delivery efficiency of application development, and shorten application deployment cycles, thereby significantly enhancing business innovation capabilities and becoming a core driving force for industrial development and business growth. Currently, cloud-native technology has evolved from an early experiment by a few leading companies to an essential technology for mainstream enterprises, and is gradually penetrating deeply into various industries.

[0003] In the cloud-native technology ecosystem, container orchestration engines, represented by Kubernetes, are among the most representative innovative achievements. As a core technology of the cloud-native ecosystem, Kubernetes not only defines a series of industry standards but also provides an important reference framework for the development of other related technologies. Cloud-native resource management, as a key area within this ecosystem, requires the following prerequisites for implementation: resources must be uniformly described according to general-purpose computing nodes; functions deployed on general-purpose computing nodes must support uniform configuration according to expected requirements; and the cloud-native resource management system must effectively control the function deployment process based on the expected requirements of function deployment. Furthermore, each node in the system should be equipped with an independent node agent to collect resource status information and, through interaction with the cloud-native interface server, complete the reporting of resource status and the acquisition of function application scheduling and deployment results.

[0004] To achieve the above goals, the cloud-native resource scheduling and management control system needs to focus on solving three core problems: first, establishing unified general resource description templates and general function deployment description templates; second, building a function deployment controller that can accurately control function deployment based on general function deployment description instances; and third, developing a resource scheduler that can flexibly deploy functions on resources based on real-time resource status and function deployment description information. Although Kubernetes-based container orchestration engines and related components (such as node resource, deployment resource, API server, controller manager, deployment controller, scheduler, device plugin, etc.) can basically achieve the above technical goals, their functions are mainly concentrated on the management of general computing resources and GPGPU resources, and cannot yet fully meet the needs of diverse resource management.

[0005] With the migration of cloud-native technologies to embedded edge systems, the management of heterogeneous computing resources is becoming increasingly prominent. A significant characteristic of embedded systems is their inclusion of numerous heterogeneous computing resources. These resources exist to meet the diverse computing needs of various heterogeneous functions within integrated embedded systems. Currently, widely used heterogeneous computing resources mainly include PPCs, DSPs, FPGAs, and various DSAs. However, these heterogeneous computing resources differ significantly in instruction sets, computing architectures, computing modes, program loading methods, and program packaging formats. Further complicating matters, different types of embedded devices employ different heterogeneous computing resources, and even within the same type of device, different systems may be equipped with different types of heterogeneous computing resources. This diversity leads to the common practice of using multiple types of heterogeneous computing resources in combination, posing new challenges to resource management. Therefore, how to achieve efficient management and flexible scheduling of onboard heterogeneous computing resources within a cloud-native framework has become a critical technical challenge that urgently needs to be overcome.

[0006] Existing cloud-native architecture implementation technologies still have many shortcomings in terms of heterogeneous resource scheduling and deployment when applied to systems employing onboard heterogeneous computing resources, including: 1. Lack of a unified description mechanism for onboard heterogeneous resources: Currently, cloud-native scheduling and orchestration engines primarily manage general-purpose computing resources and accelerator resources, such as CPU resources and GPGPU resources. Therefore, cloud-native scheduling and orchestration engines have designed standardized description mechanisms for CPUs and GPGPUs, enabling these resources to be managed by container scheduling engines. However, these description mechanisms cannot be directly applied to onboard heterogeneous computing resources, such as DSPs and FPGAs, to achieve a standardized description of onboard heterogeneous computing resources.

[0007] 2. Lack of control mechanism for onboard heterogeneous resources: Currently, the resource control mechanism of cloud-native scheduling and orchestration engines is mainly carried out through control managers and controllers. In order to control general computing resources and accelerator resources, cloud-native scheduling and orchestration engines have designed and implemented a series of controllers. All these controllers are managed uniformly by the control manager. However, these controllers are mainly for general processor and accelerator resources, each with its own function, and cannot be directly applied to onboard heterogeneous computing resources to achieve control over onboard heterogeneous computing resources.

[0008] 3. Lack of scheduling and deployment mechanism for onboard heterogeneous resources: Currently, the scheduling and deployment mechanism of cloud-native scheduling and orchestration engines mainly monitors the status of general computing resources and accelerator resources, and then uses a plug-in scheduling algorithm framework. However, the scheduling characteristics considered by all plug-ins connected to the cloud-native scheduling and orchestration engine are for general computing and accelerator resources, which are difficult to apply directly to onboard heterogeneous resources to achieve the scheduling and deployment of onboard heterogeneous computing resources. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide a cloud-native resource scheduling and deployment control system and method for airborne environments. This system can improve the adaptability and operating efficiency of container scheduling and orchestration engines in airborne heterogeneous computing resource environments, and overcome the problems of existing cloud-native architecture implementation technologies lacking a unified description mechanism for airborne heterogeneous resources, a control mechanism for airborne heterogeneous resources, and a scheduling and deployment mechanism for airborne heterogeneous resources.

[0010] The objective of this invention is achieved through the following approach: A cloud-native resource scheduling and deployment control system for airborne environments includes: a heterogeneous resource definition template, a heterogeneous resource module management component, a heterogeneous resource proxy component, a heterogeneous resource controller component, a heterogeneous resource scheduler component, and an interface server. For each type of heterogeneous computing resource, a heterogeneous resource definition template is used to define selected types of heterogeneous resources, forming resource description files. Resource instances are created on the interface server. The heterogeneous resource controller component tunes resource control based on the desired state of the resources in the selected type of resource description file instance, ensuring the actual state of the resources reaches the desired state, and stores the target state on the interface server. Functional application instances are created on the interface server. The heterogeneous resource scheduler component makes scheduling decisions based on the desired state of each functional application on the interface server. The heterogeneous resource proxy component reads functional deployment information through interaction with the interface server and distributes it to the heterogeneous resource module management component. The heterogeneous resource module management component performs the actual operations of program loading and unloading, and updates and reports heterogeneous resource status information.

[0011] Furthermore, in an airborne heterogeneous computing resource environment, for each heterogeneous resource node, a copy of the heterogeneous resource module management component and a copy of the heterogeneous resource proxy component are deployed; for each type of heterogeneous resource node, a copy of the heterogeneous resource controller component is deployed; the entire resource scheduling deployment control system deploys a copy of the heterogeneous resource scheduler plugin, a copy of the interface service, and consistent storage.

[0012] Furthermore, the heterogeneous resource nodes include PPC nodes, DSP nodes, and FPGA nodes; the PPC nodes, DSP nodes, and FPGA nodes load various functional application components according to the functional deployment information of the scheduling system.

[0013] Furthermore, the functional application components can be loaded and unloaded onto heterogeneous computing nodes, and combined to form program units with complete onboard functions.

[0014] Furthermore, the heterogeneous resource proxy component is deployed on a general computing node or a heterogeneous resource node to enable the uploading and distribution of resource control information, resource status information, function deployment information, and function status information between the heterogeneous resource controller component, the heterogeneous resource module management component, and the interface server.

[0015] Furthermore, the heterogeneous resource controller component is deployed on the management node and is used to implement state control of heterogeneous resources based on the heterogeneous resource definition and resource tuning algorithm.

[0016] Furthermore, the heterogeneous resource scheduler component is deployed on the management node and is used to make decisions on the deployment of functions on heterogeneous resources based on the functional application definition and resource scheduling algorithm.

[0017] Furthermore, the heterogeneous resource scheduler component includes a heterogeneous resource scheduler plugin, which enables the functional application to be scheduled and deployed on heterogeneous resources.

[0018] Furthermore, the consistent storage includes three-replica consistent storage, which constitutes a high-availability system.

[0019] A cloud-native resource scheduling and deployment control method for airborne environments, based on the cloud-native resource scheduling and deployment control system for airborne environments as described above, includes the following steps: P1: Create a heterogeneous resource definition instance based on the heterogeneous resource description template; P2: Create a functional application definition instance based on the functional application definition template; P3: Deploy the interface server, consistent storage, heterogeneous resource controller component, and heterogeneous resource scheduler component on the management node; P4: Enable heterogeneous resource nodes and run the module management component: P5: Create a heterogeneous resource definition instance in the interface server; P6: The heterogeneous resource controller component deploys the heterogeneous resource proxy component on the general node based on the heterogeneous resource definition type information; P7: The heterogeneous resource controller component sends resource control information to the heterogeneous resource proxy component based on the heterogeneous resource definition expectation information and resource control tuning algorithm; P8: The heterogeneous resource proxy component sends resource control information to the module management component through the embedded service bus; P9: The module management component performs resource control and operation based on the resource control information, and sends the resource status information back to the heterogeneous resource proxy component through the embedded service bus; P10: The heterogeneous resource proxy component updates the resource status information to the interface server; P11: The interface server has entered the waiting state for the creation of functional application instances; P12: Create a functional application definition instance in the interface server; P13: The heterogeneous resource scheduler component obtains the expected information of the functional application and the status information of the function and heterogeneous resources from the interface server; P14: The scheduler component determines whether all functional application instances have been deployed. If not, it jumps to P15 for processing; if so, it jumps to P32 for processing. P15: The scheduler component selects unscheduled functions, makes function resource decisions based on resource information and heterogeneous resource scheduling algorithms, and updates the function deployment information to the interface server; P16: The heterogeneous resource proxy component obtains the function deployment information from the interface server and determines the function deployment information to be loaded on the corresponding node; P17: The heterogeneous resource proxy component sends functional deployment information to heterogeneous resource nodes PPC, DSP or FPGA via the embedded service bus; P18: The scheduling algorithm determines the type of heterogeneous resources to be deployed. If the deployment node is a PPC node, it jumps to P19 for processing; if it is a DSP node, it jumps to P22 for processing; and if it is an FPGA node, it jumps to P25 for processing. P19: Prepare for loading PPC functional application components; P20: Based on the PPC resource type and PPC resource quantity constraints in the functional application definition instance, plan the function on this heterogeneous resource and restrict resource usage; P21: The module management component loads PPC functional application components; P22: Prepare for loading DSP functional application components; P23: Based on the DSP resource type and DSP resource quantity constraints in the functional application definition instance, plan the function on this heterogeneous resource and restrict resource usage; P24: The module management component loads DSP functional application components; P25: Prepare for loading FPGA functional application components; P26: Based on the FPGA resource type and FPGA resource quantity constraints in the functional application definition instance, plan the functions on this heterogeneous resource and restrict resource usage; P27: The module management component loads FPGA functional application components; P28: Check if the functional application components are loaded successfully. If the loading is successful, proceed to P29 for processing. If the loading fails, proceed to P30 for processing. P29: The module management component updates the information on successful loading and heterogeneous resource status, and sends the update information to the heterogeneous resource proxy component; P30: The module management component updates the loading failure information and heterogeneous resource status information, and sends the update information to the heterogeneous resource proxy component; P31: The heterogeneous resource proxy component updates the function and resource status information to the interface server and jumps to P13 for processing; P32: Perform a system exit judgment. If the situation is negative, that is, no system exit signal is received, jump to P11 and enter the waiting state for the creation of functional application instances; if the situation is positive, that is, the scheduling and deployment control system receives a system exit signal and ends the operation of the overall deployment control system.

[0020] The beneficial effects of this invention include: This invention uses a heterogeneous resource definition template to achieve a unified description of airborne heterogeneous computing resources, and uses a custom extended description mechanism to describe the differentiated characteristics of heterogeneous resources. It supports airborne heterogeneous resources to be accessed by cloud-native scheduling and orchestration systems in a unified manner, thus solving the problem that cloud-native implementation technologies are difficult to adapt to diverse computing resources in the airborne field.

[0021] This invention employs a heterogeneous resource controller to control the status of onboard heterogeneous resources. By constructing a corresponding controller for each type of heterogeneous computing resource, and using a standardized tuning mechanism of a cloud-native scheduling system, it achieves standardized control of heterogeneous onboard computing resources, thus solving the problem that existing cloud-native controllers are difficult to tune and control onboard heterogeneous computing resources.

[0022] This invention employs a heterogeneous resource scheduler plugin to manage the scheduling of onboard heterogeneous resources. By constructing corresponding scheduling algorithms for onboard heterogeneous resources and implementing these algorithms as plugins that conform to the cloud-native scheduling framework, it solves the problem that existing cloud-native scheduling, orchestration, and collaboration methods are insufficient for scheduling and deploying onboard heterogeneous computing resources.

[0023] This invention uses a heterogeneous resource proxy to interact with an interface controller, enabling the reporting and dissemination of heterogeneous resource status and control information, thereby improving the loose coupling and modularity of the cloud-native resource scheduling and deployment control system.

[0024] This invention is applicable to resource scheduling, deployment, and control system products that involve multiple heterogeneous computing resources in an airborne environment. It can be widely used in the design and development of cloud-native architectures for airborne edge platforms and airborne unmanned platforms. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram illustrating the principle of a cloud-native heterogeneous resource scheduling and deployment control system for embedded environments in an embodiment of the present invention. Figure 2 for Figure 1 The operation flowchart; Figure 3 for Figure 1 Deployment diagram; Figure 4 for Figure 1 A diagram illustrating instances of heterogeneous resource definitions; Figure 5 for Figure 1 Example diagram of functional application definition. Detailed Implementation

[0027] All features disclosed in all embodiments of this specification, or steps in all methods or processes implied in the disclosure, may be combined and / or extended or replaced in any way, except for mutually exclusive features and / or steps.

[0028] As a first aspect of the present invention, in a preferred embodiment, a cloud-native resource scheduling and deployment control system for airborne environments is specifically provided, such as... Figure 1As shown, it includes: a heterogeneous resource definition template, a heterogeneous resource module management component, a heterogeneous resource proxy component, a heterogeneous resource controller component, a heterogeneous resource scheduler plugin, and an interface server. In an onboard heterogeneous computing resource environment, for each heterogeneous resource node, one heterogeneous resource module management component and one heterogeneous resource proxy component are deployed; for each type of heterogeneous resource node, one heterogeneous resource controller component is deployed; the entire resource scheduling and deployment control system deploys one heterogeneous resource scheduler plugin, one interface server, and three-replica consistent storage, which constitutes a high-availability system.

[0029] First, a specific type of heterogeneous resource is defined using a heterogeneous resource definition template for each type of heterogeneous computing resource, forming a resource description file. Second, resource instances are created on the interface server. Third, the heterogeneous resource controller component automatically tunes resource control based on the desired state of the resources in the specific type of resource description file instance, ensuring that the actual state of the resources reaches the desired state, and stores the target state on the interface server. Fourth, functional application instances are created on the interface server. Fifth, the heterogeneous resource scheduler plugin makes scheduling decisions based on the desired state of each functional application on the interface server. Sixth, the heterogeneous resource agent component reads functional deployment information through interaction with the interface server and distributes it to the heterogeneous resource module management component. Finally, the heterogeneous resource module management component performs the actual operations of adding and unloading programs and updating and reporting heterogeneous resource status information.

[0030] The system described in the above embodiments will be further explained from the perspective of heterogeneous resources as follows: Heterogeneous resource layer: This is the foundation of the airborne cloud-native resource scheduling, deployment, and control system. The heterogeneous resource layer mainly consists of the following three parts: There are three types of heterogeneous nodes: PPC nodes, DSP nodes, and FPGA nodes. Various functional application components can be loaded on these three types of heterogeneous nodes according to the functional deployment information of the scheduling system.

[0031] Module Management Component: This is software on the airborne heterogeneous computing node used for various direct controls and operations on modules, as well as for status acquisition. The acquired statuses include the module's power-on / off status, temperature status, electronic tags, program loading status, etc. The control and operations that can be performed include adding / unloading programs, module self-tests, power-on / off settings, etc.

[0032] Functional application components: These are program units that can be added to or unloaded onto heterogeneous computing nodes and combined to form complete onboard functions.

[0033] Heterogeneous resource proxy layer: It is an essential component of the airborne cloud-native resource scheduling and deployment control system. The heterogeneous resource layer is mainly composed of heterogeneous resource proxy components.

[0034] Heterogeneous resource proxy component: This is an important functional component deployed on general computing nodes or heterogeneous resource nodes to enable the uploading and distribution of resource control information, resource status information, function deployment information, and function status information between heterogeneous resource controller components, module management components, and interface servers.

[0035] The scheduling and control layer is a core component of the airborne cloud-native resource scheduling and deployment control system. It mainly comprises two parts: Heterogeneous Resource Controller Component: This is a functional component deployed on the management node of the cloud-native resource scheduling and deployment control system. It is used to implement state control of heterogeneous resources based on the definition of heterogeneous resources and resource tuning algorithms.

[0036] The heterogeneous resource scheduler component is deployed on the management node of the cloud-native resource scheduling and deployment control system. It is used to make decisions on deploying functions on heterogeneous resources based on application definitions and resource scheduling algorithms. This component mainly includes a resource scheduler framework, a general resource scheduler plugin, and a heterogeneous resource scheduler plugin. The resource scheduler framework and general resource scheduler plugin are built-in functions of the general cloud-native resource scheduling and orchestration engine. The heterogeneous resource scheduler plugin is an extension design of this invention specifically for the heterogeneous resource characteristics of airborne environments, used to enable the scheduling and deployment of applications on heterogeneous resources.

[0037] Interface Layer: This is the entry point for the airborne cloud-native resource scheduling, deployment, and control system. The interface layer mainly consists of two parts: Interface server: It serves as the information exchange hub for the operation of the airborne cloud-native resource scheduling, deployment, and control system. The heterogeneous resource controller component, heterogeneous resource scheduler component, and heterogeneous resource agent component all acquire and update resource and functional information through interaction with the interface server.

[0038] Consistent storage is an information storage center that enables the airborne cloud-native resource scheduling, deployment, and control system. This consistent storage is formed through three replicas to achieve high availability.

[0039] As a second aspect of the present invention, the operating flow of the system described in the above embodiments is provided, namely, a cloud-native resource scheduling and deployment control method for airborne environments is provided, such as... Figure 2 As shown, it includes the following steps: P1: Create a heterogeneous resource definition instance based on the heterogeneous resource description template; P2: Create a functional application definition instance based on the functional application definition template; P3: Deploy the interface server, consistent storage, heterogeneous resource controller component, and heterogeneous resource scheduler component on the management node; P4: Enable heterogeneous resource nodes and run the module management component: P5: Create a heterogeneous resource definition instance in the interface server; P6: The heterogeneous resource controller component deploys the heterogeneous resource proxy component on the general node based on the heterogeneous resource definition type information; P7: The heterogeneous resource controller component sends resource control information to the heterogeneous resource proxy component based on the heterogeneous resource definition expectation information and resource control tuning algorithm; P8: The heterogeneous resource proxy component sends resource control information to the module management component through the embedded service bus; P9: The module management component performs resource control and operation based on the resource control information, and sends the resource status information back to the heterogeneous resource proxy component through the embedded service bus; P10: The heterogeneous resource proxy component updates the resource status information to the interface server; P11: The interface server has entered the waiting state for the creation of functional application instances; P12: Create a functional application definition instance in the interface server; P13: The heterogeneous resource scheduler component obtains the expected information of the functional application and the status information of the function and heterogeneous resources from the interface server; P14: The scheduler component determines whether all functional application instances have been deployed. If not, it jumps to P15 for processing; if so, it jumps to P32 for processing. P15: The scheduler component selects unscheduled functions, makes function-resource decisions based on resource information and heterogeneous resource scheduling algorithms, and updates the function deployment information to the interface server; P16: The heterogeneous resource proxy component obtains the function deployment information from the interface server and determines the function deployment information to be loaded on the corresponding node; P17: The heterogeneous resource proxy component sends functional deployment information to heterogeneous resource nodes PPC, DSP or FPGA via the embedded service bus; P18: The scheduling algorithm determines the type of heterogeneous resources to be deployed. If the deployment node is a PPC node, it jumps to P19 for processing; if it is a DSP node, it jumps to P22 for processing; and if it is an FPGA node, it jumps to P25 for processing. P19: Prepare for loading PPC functional application components; P20: Based on the PPC resource type and PPC resource quantity constraints in the functional application definition instance, plan the function on this heterogeneous resource and restrict resource usage; P21: The module management component loads PPC functional application components; P22: Prepare for loading DSP functional application components; P23: Based on the DSP resource type and DSP resource quantity constraints in the functional application definition instance, plan the function on this heterogeneous resource and restrict resource usage; P24: The module management component loads DSP functional application components; P25: Prepare for loading FPGA functional application components; P26: Based on the FPGA resource type and FPGA resource quantity constraints in the functional application definition instance, plan the functions on this heterogeneous resource and restrict resource usage; P27: The module management component loads FPGA functional application components; P28: Check if the functional application components are loaded successfully. If the loading is successful, proceed to P29 for processing. If the loading fails, proceed to P30 for processing. P29: The module management component updates the information on successful loading and heterogeneous resource status, and sends the update information to the heterogeneous resource proxy component; P30: The module management component updates the loading failure information and heterogeneous resource status information, and sends the update information to the heterogeneous resource proxy component; P31: The heterogeneous resource proxy component updates the function and resource status information to the interface server and jumps to P13 for processing; P32: Perform a system exit judgment. If the situation is negative, that is, no system exit signal is received, jump to P11 and enter the waiting state for the creation of functional application instances; if the situation is positive, that is, the scheduling and deployment control system receives a system exit signal and ends the operation of the overall deployment control system.

[0040] As a third aspect of the present invention, a deployment method for the system described in the above embodiments is provided, such as... Figure 3 As shown, it includes: a management node, a general computing resource pool, and a heterogeneous computing resource pool, with the following specific deployment components: 1) The management node deploys the interface server, consistent storage, heterogeneous resource controller component, and heterogeneous resource scheduler component; 2) The general computing resource pool consists of general computing resource nodes. The general computing resource nodes deploy two types of containers, mainly including heterogeneous resource proxy component containers and functional application component containers. These two types of containers are heterogeneous resource proxy components and functional application components that have been containerized and packaged to run in a containerized manner. 3) The heterogeneous computing resource pool consists of heterogeneous computing resource nodes. The heterogeneous computing resource nodes deploy two types of components, mainly including module management components and functional application components deployed on heterogeneous computing resources. These two types of components are deployed in the form of executable programs.

[0041] like Figure 4 As shown, heterogeneous resource definition is mainly implemented through the heterogeneous resource definition module. The heterogeneous resource definition template is defined in YAML (Yet Another Markup Language) format, using key-value pairs to describe the resource. A typical heterogeneous resource definition includes interface version (apiVersion), type (kind), metadata (metadata), and specification (spec). The interface version is used to manage the version of the resource interface, the type is used to distinguish the types of heterogeneous resources, the metadata is used to define the basic information of the heterogeneous resource, such as name, namespace, and tags, and the specification is used to constrain the expected state of the resource. This constraint is the basis for the resource controller to dynamically adjust the resource.

[0042] like Figure 5 As shown, the functional application definition is mainly implemented through the functional application definition template, which is defined in YAML format and consists of key-value pairs describing the functional application. A typical functional application definition includes the interface version (apiVersion), type (kind), metadata (metadata), and specification (spec). The interface version is used for version management of the functional interface, the type is used to distinguish functional types, the metadata is used to define basic functional information such as name and namespace, and the specification is used to constrain the expected state of the functional. This constraint is the basis for the heterogeneous resource scheduler plugin to deploy and schedule the functional application.

[0043] The units described in the embodiments of the present invention can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0044] According to one aspect of the present invention, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the methods provided in the various optional implementations described above.

[0045] In another aspect, embodiments of the present invention also provide a computer-readable medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the methods described in the above embodiments.

Claims

1. A cloud-native resource scheduling and deployment control system for airborne environments, characterized in that, include: Heterogeneous resource definition template, heterogeneous resource module management component, heterogeneous resource proxy component, heterogeneous resource controller component, heterogeneous resource scheduler component, and interface server; For each type of heterogeneous computing resource, a heterogeneous resource definition template is used to define the selected type of heterogeneous resource, forming a resource description file; Resource instances are created on the interface server; The heterogeneous resource controller component tunes resource control for the desired state of resources in a selected type of resource description file instance, so that the actual state of the resource reaches the desired state, and stores the target state in the interface server. Create functional application instances in the interface server; The heterogeneous resource scheduler component makes scheduling decisions based on the desired state of each functional application in the interface server. The heterogeneous resource proxy component reads the function deployment information and sends it to the heterogeneous resource module management component through interaction with the interface server; The heterogeneous resource module management component performs the actual operations of adding and unloading programs, as well as updating and reporting heterogeneous resource status information.

2. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 1, characterized in that, In an airborne heterogeneous computing resource environment, for each heterogeneous resource node, a copy of the heterogeneous resource module management component and a copy of the heterogeneous resource proxy component are deployed; for each type of heterogeneous resource node, a copy of the heterogeneous resource controller component is deployed; the entire resource scheduling and deployment control system deploys a copy of the heterogeneous resource scheduler plugin, a copy of the interface service, and consistent storage.

3. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 2, characterized in that, The heterogeneous resource nodes include PPC nodes, DSP nodes, and FPGA nodes; the PPC nodes, DSP nodes, and FPGA nodes load various functional application components according to the functional deployment information of the scheduling system.

4. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 3, characterized in that, The functional application components can be loaded and unloaded onto heterogeneous computing nodes, and combined to form program units with complete airborne functions.

5. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 2, characterized in that, The heterogeneous resource proxy component is deployed on general computing nodes or heterogeneous resource nodes to enable the uploading and distribution of resource control information, resource status information, function deployment information, and function status information between the heterogeneous resource controller component, the heterogeneous resource module management component, and the interface server.

6. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 2, characterized in that, The heterogeneous resource controller component is deployed on the management node and is used to control the state of heterogeneous resources according to the definition of heterogeneous resources and the resource tuning algorithm.

7. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 2, characterized in that, The heterogeneous resource scheduler component is deployed on the management node and is used to make decisions on the deployment of functions on heterogeneous resources based on the functional application definition and resource scheduling algorithm.

8. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 7, characterized in that, The heterogeneous resource scheduler component includes a heterogeneous resource scheduler plugin, which is used to enable the functional application to schedule and deploy on heterogeneous resources.

9. The cloud-native resource scheduling and deployment control system for airborne environments according to claim 2, characterized in that, The consistent storage includes three-replica consistent storage, which constitutes a highly available system.

10. A cloud-native resource scheduling and deployment control method for airborne environments, characterized in that, The cloud-native resource scheduling and deployment control system for airborne environments as described in claim 1 includes the following steps: P1: Create a heterogeneous resource definition instance based on the heterogeneous resource description template; P2: Create a functional application definition instance based on the functional application definition template; P3: Deploy the interface server, consistent storage, heterogeneous resource controller component, and heterogeneous resource scheduler component on the management node; P4: Enable heterogeneous resource nodes and run the module management component: P5: Create a heterogeneous resource definition instance in the interface server; P6: The heterogeneous resource controller component deploys the heterogeneous resource proxy component on the general node based on the heterogeneous resource definition type information; P7: The heterogeneous resource controller component sends resource control information to the heterogeneous resource proxy component based on the heterogeneous resource definition expectation information and resource control tuning algorithm; P8: The heterogeneous resource proxy component sends resource control information to the module management component through the embedded service bus; P9: The module management component performs resource control and operation based on the resource control information, and sends the resource status information back to the heterogeneous resource proxy component through the embedded service bus; P10: The heterogeneous resource proxy component updates the resource status information to the interface server; P11: The interface server has entered the waiting state for the creation of functional application instances; P12: Create a functional application definition instance in the interface server; P13: The heterogeneous resource scheduler component obtains the expected information of the functional application and the status information of the function and heterogeneous resources from the interface server; P14: The scheduler component determines whether all functional application instances have been deployed. If not, it jumps to P15 for processing; if so, it jumps to P32 for processing. P15: The scheduler component selects unscheduled functions, makes function resource decisions based on resource information and heterogeneous resource scheduling algorithms, and updates the function deployment information to the interface server; P16: The heterogeneous resource proxy component obtains the function deployment information from the interface server and determines the function deployment information to be loaded on the corresponding node; P17: The heterogeneous resource proxy component sends functional deployment information to heterogeneous resource nodes PPC, DSP or FPGA via the embedded service bus; P18: The scheduling algorithm determines the type of heterogeneous resources to be deployed. If the deployment node is a PPC node, it jumps to P19 for processing; if it is a DSP node, it jumps to P22 for processing; and if it is an FPGA node, it jumps to P25 for processing. P19: Prepare for loading PPC functional application components; P20: Based on the PPC resource type and PPC resource quantity constraints in the functional application definition instance, plan the function on this heterogeneous resource and restrict resource usage; P21: The module management component loads PPC functional application components; P22: Prepare for loading DSP functional application components; P23: Based on the DSP resource type and DSP resource quantity constraints in the functional application definition instance, plan the function on this heterogeneous resource and restrict resource usage; P24: The module management component loads DSP functional application components; P25: Prepare for loading FPGA functional application components; P26: Based on the FPGA resource type and FPGA resource quantity constraints in the functional application definition instance, plan the functions on this heterogeneous resource and restrict resource usage; P27: The module management component loads FPGA functional application components; P28: Check if the functional application components are loaded successfully. If the loading is successful, proceed to P29 for processing. If the loading fails, proceed to P30 for processing. P29: The module management component updates the information on successful loading and heterogeneous resource status, and sends the update information to the heterogeneous resource proxy component; P30: The module management component updates the loading failure information and heterogeneous resource status information, and sends the update information to the heterogeneous resource proxy component; P31: The heterogeneous resource proxy component updates the function and resource status information to the interface server and jumps to P13 for processing; P32: Perform a system exit judgment. If the situation is negative, that is, no system exit signal is received, jump to P11 and enter the waiting state for the creation of functional application instances; if the situation is positive, that is, the scheduling and deployment control system receives a system exit signal and ends the operation of the overall deployment control system.

Images