A system and method for deployment and computation coordination optimization under limited resources

Abstract

The application provides a system and method for deployment and computing coordination optimization under limited resources, relating to the field of digital technology. The system for deployment and computing coordination optimization under limited resources innovatively divides the types of server resources of various specifications into shared resources or exclusive resources, and implements pre-occupation and coordination mechanisms, realizing the deep integration and dynamic allocation of server resources of the application deployment manager and the computing cluster manager. In the past, the application deployment manager was idle during the early stage of application deployment or when the application was not fully loaded, while the computing cluster manager often had to queue computing tasks due to insufficient resources. The technical solution breaks through this resource barrier, enabling the computing cluster manager to flexibly use the idle resources of the application deployment manager, ensuring that every server resource is fully utilized at any time.

Description

Technical Field

[0001] The present application relates to the field of digital technology, and in particular to a system and method for coordinated optimization of deployment and computing under limited resources. Background Art

[0002] In today's rapidly developing digital age, cloud computing and high-performance computing play an indispensable role in numerous fields, including scientific research and enterprise. As a bridge connecting users and computing resources, the efficiency of computing power scheduling and deployment platforms directly impacts the operational efficiency and resource utilization of various applications.

[0003] Currently, many computing power scheduling and deployment platforms combine application deployment with computing task scheduling. However, this seemingly perfect combination actually exposes the problem of underutilized server resources. Therefore, with limited server resources, how to coordinate and optimize deployment and computing resources has become a pressing technical problem. Summary of the Invention

[0004] In view of the above problems, this application is proposed to provide a system and method, electronic device, and storage medium for optimizing deployment and computing coordination under limited resources that overcomes or at least partially solves the above problems. The technical solution is as follows:

[0005] In a first aspect, a system for deployment and computing coordination optimization under limited resources is provided. The system is built using a microservices architecture and includes a resource management microservice, an interface terminal, a task scheduling microservice, an integrated microservice for an application deployment manager, and an integrated microservice for a computing cluster manager.

[0006] The resource management microservice is used to collect information about server resources, create one or more specifications of server resources based on the information about server resources, and configure the type of server resources of each specification as shared resources or exclusive resources;

[0007] The task scheduling microservice is configured to receive a work order submitted by a user through the interface to pre-occupy target-specification server resources for application deployment, and to call the application deployment manager's integrated microservice based on the work order, so that the application deployment manager's integrated microservice declares the required target-specification server resources; wherein the target-specification server resources are of a shared resource type;

[0008] The task scheduling microservice is further configured to respond to one or more real-time computing tasks and call the integrated microservice of the computing cluster manager according to the one or more real-time computing tasks, so that the integrated microservice of the computing cluster manager uses the target specification server resources of the shared resource type to execute the one or more real-time computing tasks;

[0009] The task scheduling microservice is further configured to receive an instruction from a user through the interface end to use the target specification server resources for application deployment, determine the target physical server to which the application is to be deployed based on the instruction, and call the integrated microservice of the computing cluster manager, so that the integrated microservice of the computing cluster manager creates a placeholder task on the target physical server with the same required computing resources as the target specification server resources. The placeholder task only occupies the computing resource quota but does not actually consume computing resources, that is, it is used to inform the computing cluster manager that the computing resources that are the same as the target specification server resources are occupied, but in fact, this part of the computing resources are not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment; the priority of the placeholder task is set to the highest;

[0010] The task scheduling microservice is further configured to determine whether the placeholder task is created successfully. If the placeholder task is created successfully, the integration microservice of the application deployment manager is called to use the target specification server resources on the target physical server to deploy the application.

[0011] In one possible implementation, the task scheduling microservice is also used by the integrated microservice of the computing cluster manager to apply for computing resources on the target physical server that are the same as the target specification server resources. If the application is approved, it is determined that the placeholder task is created successfully; if the application is not approved, it is determined that the placeholder task is not created successfully.

[0012] In one possible implementation, the task scheduling microservice is also used to call the integrated microservice of the computing cluster manager if the placeholder task is not created successfully, and the integrated microservice of the computing cluster manager stops all current computing tasks on the target physical server and creates the placeholder task. After the placeholder task is successfully created, the integrated microservice of the application deployment manager is called to use the target specification server resources on the target physical server to deploy the application.

[0013] In one possible implementation, the integrated microservice of the computing cluster manager is also used to re-queue the placeholder task and all the current computing tasks in order of priority for execution after the placeholder task is successfully created. The placeholder task will be put at the front for execution, informing the computing cluster manager that the computing resources with the same specifications as the target server resources are occupied. However, in fact, this part of the computing resources is not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment.

[0014] In a possible implementation, the integrated microservice of the computing cluster manager is further configured to wait for other available resources to execute the re-queued computing task.

[0015] In a possible implementation, the information of the server resources includes one or more of CPU usage, memory usage, GPU status, and hard disk remaining space.

[0016] In a possible implementation, the resource management microservice is further configured to provide the information of the server resources to the interface terminal for display.

[0017] The interface terminal is configured to receive a request for creating one or more specifications of server resources submitted by a user according to the information of the server resources.

[0018] The resource management microservice is further configured to create the one or more specifications of server resources according to the request, and configure the type of each specification of server resources as shared resources or exclusive resources.

[0019] In a second aspect, a method for deployment and computing coordination optimization under limited resources is provided, and the method comprises:

[0020] A system for deployment and computing coordination optimization under limited resources is built by using a microservice architecture, and the system comprises a resource management microservice, an interface terminal, a task scheduling microservice, an integrated microservice of an application deployment manager, and an integrated microservice of a computing cluster manager.

[0021] The resource management microservice collects information of server resources, creates one or more specifications of server resources according to the information of the server resources, and configures the type of each specification of server resources as shared resources or exclusive resources.

[0022] The task scheduling microservice receives a work order for preoccupying a target specification of server resources to deploy an application submitted by a user through the interface terminal, calls the integrated microservice of the application deployment manager according to the work order, and declares the required target specification of server resources by the integrated microservice of the application deployment manager; wherein the type of the target specification of server resources is shared resources.

[0023] The task scheduling microservice responds to one or more real-time computing tasks, calls the integrated microservice of the computing cluster manager according to the one or more real-time computing tasks, and executes the one or more real-time computing tasks by the integrated microservice of the computing cluster manager using the target specification of server resources of the type of shared resources.

[0024] The task scheduling microservice receives an instruction of using target specification server resources for application deployment issued by a user through the interface end, determines a target physical server for upcoming application deployment according to the instruction, and calls an integrated microservice of the computing cluster manager. The integrated microservice of the computing cluster manager creates a placeholder task with the same required computing resources as the target specification server resources on the target physical server. The placeholder task only occupies a computing resource quota but does not actually consume computing resources, that is, the placeholder task is used to inform the computing cluster manager that the same part of computing resources as the target specification server resources is occupied, but actually the part of computing resources is not used by the placeholder task but will be used by the integrated microservice of the application deployment manager for application deployment. The priority of the placeholder task is set to the highest.

[0025] The task scheduling microservice determines whether the placeholder task is successfully created. If the placeholder task is successfully created, the integrated microservice of the application deployment manager uses the target specification server resources on the target physical server to perform application deployment.

[0026] In a third aspect, an electronic device is provided, which includes a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the method for deployment and computing coordination optimization under limited resources according to any one of the preceding aspects.

[0027] In a fourth aspect, a storage medium is provided, which stores a computer program, wherein the computer program is configured to execute the method for deployment and computing coordination optimization under limited resources when running.

[0028] By means of the above technical solution, the system and method, electronic device, and storage medium for deployment and computing coordination optimization under limited resources provided by the embodiments of the present application innovatively divides the types of server resources of various specifications into shared resources or exclusive resources, and implements a pre-occupation and coordination mechanism, thereby realizing the deep integration and dynamic allocation of server resources of the application deployment manager and the computing cluster manager. In the past, in the early stage of application deployment or when the application was not fully loaded, the server resources managed by the application deployment manager were idle, and it was common for the computing cluster manager to queue computing tasks due to insufficient resources. The present technical solution breaks this resource barrier, enabling the computing cluster manager to flexibly utilize the idle resources of the application deployment manager, ensuring that each server resource can be fully utilized at any time. For example, departments use the application deployment manager to deploy business applications separately, and use the computing cluster manager to perform computing tasks such as data analysis. After adopting this technical solution, the originally idle server resources of the application deployment manager are reasonably allocated to the computing tasks of the computing cluster manager, greatly improving the efficiency of server resource utilization and reducing server resource waste.

[0029] Furthermore, this technical solution effectively addresses the resource agnostic and scheduling issues between the application deployment manager and the compute cluster manager by introducing an innovative resource coordination mechanism. When the application deployment manager deploys an application, the compute cluster manager's integrated microservice promptly creates a placeholder task on the target physical server, requiring the same computing resources as the target physical server. This placeholder task serves to identify resource occupancy but does not actually use resources. This task only occupies a computing resource quota but does not consume any. This serves to inform the compute cluster manager that the computing resources corresponding to the target physical server are occupied. However, these resources are not actually used by the placeholder task, but are instead used by the application deployment manager's integrated microservice for application deployment. The placeholder task is given the highest priority, ensuring that the deployment task proceeds smoothly and uninterrupted. Once the application deployment manager task completes and releases resources, the compute cluster manager's compute task can quickly reallocate resources and continue execution. This tight coordination mechanism breaks the previous isolation between deployment and computation, enabling seamless integration and collaborative operation between the two. With the help of this technical solution, when application deployment and data analysis and computing tasks are carried out simultaneously, the system can dynamically adjust according to resource requirements, ensuring the efficient and stable operation of the business and greatly improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments of the present application.

[0031] Figure 1 A structural diagram of a system for deployment and computational coordination optimization under limited resources provided by an embodiment of the present application is shown;

[0032] Figure 2 A flowchart of a method for coordinated optimization of deployment and computing under limited resources provided by an embodiment of the present application is shown;

[0033] Figure 3 A structural diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0034] The following describes exemplary embodiments of the present application in more detail with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0035] It should be noted that the terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that such usage is interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "including" and its variations are to be interpreted as open-ended terms meaning "including but not limited to."

[0036] In order to solve the above technical problems, the embodiment of the present application provides a system for deployment and computing coordination optimization under limited resources, which is built using a microservice architecture, such as Figure 1 As shown, the system may include a resource management microservice 110, an interface terminal 120, a task scheduling microservice 130, an application deployment manager integration microservice 140, and a computing cluster manager integration microservice 150. Specifically:

[0037] Resource management microservice 110, which is used to collect server resource information, create one or more server resource specifications based on the server resource information, and configure the type of each specification of server resource as a shared resource or an exclusive resource;

[0038] The task scheduling microservice 130 is configured to receive a work order submitted by a user through the interface 120 for deploying an application using a pre-occupied target-specification server resource. Based on the work order, the application deployment manager integration microservice 140 is invoked, and the application deployment manager integration microservice 140 declares the required target-specification server resource (here, the application deployment manager integration microservice 140 may specifically notify the application deployment manager to declare the required target-specification server resource). The target-specification server resource is a shared resource.

[0039] The task scheduling microservice 130 is further configured to respond to one or more real-time computing tasks and, based on the one or more real-time computing tasks, call the computing cluster manager integration microservice 150, so that the computing cluster manager integration microservice 150 executes the one or more real-time computing tasks using target-specification server resources that are shared resources (here, specifically, the computing cluster manager integration microservice 150 may notify the computing cluster manager to execute the one or more real-time computing tasks using target-specification server resources that are shared resources).

[0040] The task scheduling microservice 130 is further configured to receive an instruction from the user via the interface terminal 120 to use the target specification server resources for application deployment, determine the target physical server to which the application is to be deployed based on the instruction, and call the integration microservice 150 of the computing cluster manager. The integration microservice 150 of the computing cluster manager creates a placeholder task on the target physical server, which requires the same computing resources as the target specification server resources (here, the integration microservice 150 of the computing cluster manager may specifically notify the computing cluster manager to create a placeholder task on the target physical server, which requires the same computing resources as the target specification server resources). The placeholder task only occupies a computing resource quota but does not actually consume computing resources. That is, it is used to inform the computing cluster manager that the computing resources that are the same as the target specification server resources are occupied. However, in reality, the computing resources are not used by the placeholder task, but will be used by the integration microservice of the application deployment manager for application deployment. The priority of the placeholder task is set to the highest.

[0041] The task scheduling microservice 130 is also used to determine whether the placeholder task is created successfully. If the placeholder task is created successfully, it calls the application deployment manager's integration microservice 140 to use the target specification server resources on the target physical server for application deployment (here, the application deployment manager's integration microservice 140 can specifically notify the application deployment manager to use the target specification server resources on the target physical server for application deployment).

[0042] Here, a placeholder task is created on a 1:1 basis, and the placeholder task is created and destroyed along with the lifecycle of the application deployment task.

[0043] The system for deployment and computing coordination optimization under limited resources provided by this embodiment innovatively divides the types of server resources of various specifications into shared resources or exclusive resources, and implements a pre-occupation and coordination mechanism, thereby realizing the deep integration and dynamic allocation of server resources of the application deployment manager and the computing cluster manager. In the past, in the early stage of application deployment or when the application was not fully loaded, the server resources managed by the application deployment manager were idle, and it was common for the computing cluster manager to queue computing tasks due to insufficient resources. The present technical solution breaks this resource barrier, enabling the computing cluster manager to flexibly utilize the idle resources of the application deployment manager, ensuring that every server resource can be fully utilized at any time. For example, departments use the application deployment manager to deploy business applications respectively, and use the computing cluster manager to perform computing tasks such as data analysis. After adopting this technical solution, the originally idle server resources of the application deployment manager are reasonably allocated to the computing tasks of the computing cluster manager, which greatly improves the efficiency of server resource utilization and reduces server resource waste.

[0044] In this embodiment, when a microservice architecture is used to build a system for deployment and computing coordination optimization under limited resources (which can also be called a computing power scheduling and deployment system), frameworks such as Spring Cloud can be used for development. Here, SpringCloud is a microservice framework; the resource management microservice 110, the interface end 120, the task scheduling microservice 130, the application deployment manager integration microservice 140 and the computing cluster manager integration microservice 150 can communicate with each other through a RESTful API (Representational State Transfer Application Programming Interface).

[0045] In this embodiment, the server resource information collected by the resource management microservice 110 may include one or more of the following: CPU (Central Processing Unit) usage, memory usage, GPU (Graphics Processing Unit) status, and available hard disk space. For example, the resource management microservice 110 may collect resource information for the server with the IP (Internet Protocol) address 192.168.101.47, and display the real-time usage of its 1600GB memory, 8 GPU cards, 144-core CPU, and 10TB hard disk.

[0046] For example, consider Kubernetes as an application deployment manager and Slurm as a computing cluster manager. Kubernetes, abbreviated as K8s, is a container orchestration management platform widely used for application deployment thanks to its powerful container orchestration capabilities. It automates the deployment, scaling, and management of containerized applications, providing users with a convenient application deployment experience. Slurm, a mature cluster resource manager, excels in high-performance computing task scheduling, effectively allocating cluster resources and ensuring efficient execution of computing tasks.

[0047] The resource management microservice 110 can create one or more specifications of server resources based on the information of server resources, and configure the type of server resources of each specification as shared resources or exclusive resources. For example, a specification is created with the name "47-Shared Computing Power-k8s", and then the quantity of various resources required for the specification is set, namely 200G memory, 4 GPU cards, 44-core CPU, and 1TB hard disk; at the same time, the type of server resources of this specification is configured as a shared resource, indicating that the server resources of this specification can be used by Slurm when K8s is not in use, and then the specification information and the type of server resources are stored in the database. Similarly, a specification named "47-slurm" can also be created, and it can be set to occupy 1400G memory, 5 GPU cards, 100-core CPU, and 500G hard disk, and set to a shared resource type. It should be noted that the examples listed here are only illustrative and do not limit this embodiment.

[0048] In one embodiment of the present application, a possible implementation is provided. The task scheduling microservice 130 is further configured to have the computing cluster manager's integration microservice 150 request computing resources on the target physical server that are identical to the resources on the target specification server. If the request is approved, the placeholder task is determined to have been successfully created; if the request is not approved, the placeholder task is determined to have not been successfully created. In this embodiment, by having the computing cluster manager's integration microservice 150 request computing resources on the target physical server that are identical to the resources on the target specification server, the computing cluster manager is informed of the resources required by the application deployment task before the application deployment task is initiated, preventing any impact on the application deployment task execution.

[0049] In one embodiment of the present application, a possible implementation is provided. The task scheduling microservice 130 is further configured to call the computing cluster manager's integration microservice 150 if the placeholder task is not successfully created. The computing cluster manager's integration microservice 150 stops all current computing tasks on the target physical server and creates a placeholder task. Once the placeholder task is successfully created, the integration microservice 140 of the application deployment manager is called to deploy the application using the target-specification server resources on the target physical server. Once the placeholder task is successfully created, it serves only to identify resource occupancy and does not actually use the resources.

[0050] In this embodiment, the integrated microservice 150 of the computing cluster manager stops all current computing tasks on the target physical server, thereby freeing up resources for application deployment.

[0051] A possible implementation method is provided in an embodiment of the present application. The integrated microservice 150 of the computing cluster manager is also used to re-queue the placeholder task and all current computing tasks in order of priority for execution after the placeholder task is successfully created. The placeholder task will be queued to the front for execution, informing the computing cluster manager that the computing resources with the same specifications as the target server resources are occupied. However, in fact, this part of the computing resources is not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment.

[0052] The present application provides a possible implementation method in which the compute cluster manager's integrated microservice 150 is also used to wait for other available resources to execute requeued computing tasks. In this embodiment, when the application deployment manager task completes and releases resources, the computing cluster manager's computing tasks can quickly reallocate resources and continue execution. This tight coordination mechanism breaks the traditional isolation between deployment and computing, achieving seamless integration and coordinated operation between the two.

[0053] The embodiment of the present application provides a possible implementation method, in which the resource management microservice 110 is further configured to provide information about server resources to the interface terminal 120 for display;

[0054] The interface terminal 120 is used to receive a request for creating one or more specifications of a server resource submitted by a user based on the information of the server resource;

[0055] The resource management microservice 110 is also used to create one or more specifications of server resources according to the request, and configure the type of each specification of server resources as shared resources or exclusive resources.

[0056] In this embodiment, users can flexibly set resource specifications and types according to actual business needs to allocate and manage resources more accurately.

[0057] Based on the systems for coordinated optimization of deployment and computing under limited resources provided in the above embodiments, and based on the same inventive concept, an embodiment of the present application also provides a method for coordinated optimization of deployment and computing under limited resources.

[0058] like Figure 2 As shown, the embodiment of the present application provides a method for coordinated optimization of deployment and computing under limited resources, which may specifically include the following steps S201 to S206:

[0059] Step S201, using a microservice architecture to build a system for deployment and computing coordination optimization under limited resources, the system includes a resource management microservice, an interface end, a task scheduling microservice, an integrated microservice of an application deployment manager, and an integrated microservice of a computing cluster manager.

[0060] In this step, you can use a framework such as Spring Cloud for development. Spring Cloud is a microservices framework. The resource management microservice, the user interface, the task scheduling microservice, the application deployment manager integration microservice, and the compute cluster manager integration microservice can communicate with each other through RESTful APIs.

[0061] In step S202, the resource management microservice collects information about server resources, creates one or more specifications of server resources based on the information about the server resources, and configures the type of server resources of each specification as shared resources or exclusive resources.

[0062] In this step, the server resource information may include one or more of CPU usage, memory usage, GPU status, and hard disk remaining space.

[0063] Here, the resource management microservice can provide server resource information to the user interface for display. The user interface can then receive requests from users to create one or more server resource specifications based on the server resource information. The resource management microservice then creates one or more server resource specifications based on the requests and configures each server resource specification as a shared resource or an exclusive resource. This allows users to flexibly set resource specifications and types based on actual business needs, allocating and managing resources more accurately.

[0064] In step S203, the task scheduling microservice receives a work order submitted by the user through the interface to pre-occupy the target specification server resources to deploy the application, calls the integrated microservice of the application deployment manager based on the work order, and the integrated microservice of the application deployment manager declares the required target specification server resources; wherein the type of the target specification server resources is a shared resource.

[0065] In this step, the type of the target specification server resource is a shared resource, which means that the target specification server resource can be used by the computing tasks of the computing cluster manager when the application deployment manager is not using it.

[0066] In step S204, the task scheduling microservice responds to one or more real-time computing tasks, calls the integrated microservice of the computing cluster manager according to the one or more real-time computing tasks, and the integrated microservice of the computing cluster manager uses the target specification server resources of the shared resource type to execute the one or more real-time computing tasks.

[0067] In step S205, the task scheduling microservice receives an instruction from the user through the interface to use the target specification server resources for application deployment, determines the target physical server to be deployed according to the instruction, and calls the integrated microservice of the computing cluster manager. The integrated microservice of the computing cluster manager creates a placeholder task on the target physical server with the required computing resources being the same as the target specification server resources. The placeholder task only occupies the computing resource quota but does not actually consume computing resources. That is, it is used to inform the computing cluster manager that the computing resources that are the same as the target specification server resources are occupied, but in fact this part of the computing resources are not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment; the priority of the placeholder task is set to the highest.

[0068] In step S206, the task scheduling microservice determines whether the placeholder task is created successfully. If the placeholder task is created successfully, the integration microservice of the application deployment manager is called to use the target specification server resources on the target physical server to deploy the application.

[0069] This embodiment innovatively divides the types of server resources of various specifications into shared resources or exclusive resources, and implements a pre-occupancy and coordination mechanism, thereby realizing the deep integration and dynamic allocation of server resources of the application deployment manager and the computing cluster manager. In the past, in the early stage of application deployment or when the application was not fully loaded, the server resources managed by the application deployment manager were idle, and it was common for the computing cluster manager to queue up computing tasks due to insufficient resources. This technical solution breaks this resource barrier, enabling the computing cluster manager to flexibly utilize the idle resources of the application deployment manager, ensuring that every server resource can be fully utilized at any time. For example, departments use the application deployment manager to deploy business applications respectively, and use the computing cluster manager to perform computing tasks such as data analysis. After adopting this technical solution, the originally idle server resources of the application deployment manager are reasonably allocated to the computing tasks of the computing cluster manager, which greatly improves the efficiency of server resource utilization and reduces server resource waste.

[0070] In one embodiment of the present application, a possible implementation method is provided. In step S206 above, when the task scheduling microservice determines whether the placeholder task has been successfully created, the task scheduling microservice can also call the computing cluster manager's integrated microservice, which then requests computing resources on the target physical server that are identical to the target specification server resources. If the request is approved, the placeholder task is determined to have been successfully created; if not, the placeholder task is determined to have been unsuccessfully created. In this embodiment, by having the computing cluster manager's integrated microservice request computing resources on the target physical server that are identical to the target specification server resources, the computing cluster manager is informed of the resources required by the application deployment task before the application deployment task is initiated, preventing any impact on the execution of the application deployment task.

[0071] The present application provides a possible implementation method. In step S206 above, when the task scheduling microservice determines whether the placeholder task is successfully created, if the placeholder task is not successfully created, the task scheduling microservice calls the integrated microservice of the computing cluster manager, which stops all current computing tasks on the target physical server and creates a placeholder task. After the placeholder task is successfully created, the integrated microservice of the application deployment manager is called to use the target specification server resources on the target physical server for application deployment. After the placeholder task is successfully created, the placeholder task is only used to identify resource occupancy and does not actually use the resources.

[0072] In this embodiment, the integrated microservice of the computing cluster manager stops all current computing tasks on the target physical server, thereby freeing up resources for application deployment.

[0073] In an embodiment of the present application, a possible implementation method is provided. After the placeholder task is successfully created, the integrated microservice of the computing cluster manager will re-queue the placeholder task and all current computing tasks in order of priority for execution. The placeholder task will be put to the front for execution, informing the computing cluster manager that the computing resources with the same target specification server resources are occupied. However, in fact, this part of the computing resources is not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment.

[0074] This embodiment of the present application provides a possible implementation method, in which the integrated microservice of the computing cluster manager waits for other available resources to execute the requeued computing tasks. In this embodiment, when the application deployment manager task completes and releases resources, the computing cluster manager's computing tasks can quickly reallocate resources and continue execution. This tight coordination mechanism breaks the traditional isolation between deployment and computing, achieving seamless integration and coordinated operation of the two.

[0075] The above introduces Figure 2There are multiple implementation methods for each link of the embodiment shown. The following will further illustrate the method of deployment and computing coordination optimization under limited resources of the embodiment of the present application through specific embodiments.

[0076] In this specific embodiment, the application deployment manager is Kubernetes (abbreviated as K8s) and the computing cluster manager is Slurm as an example.

[0077] Currently, many computing power scheduling and deployment platforms combine the application deployment capabilities of K8s with the computational task scheduling capabilities of Slurm, allowing users to deploy their own applications through K8s and execute tasks on Slurm clusters. However, this seemingly perfect combination actually exposes the problem of underutilized server resources. Under the current operating model, K8s runs deployment tasks on the servers it manages, while Slurm runs computation tasks on its own servers. This results in relative independence of resources between the two and a lack of effective coordination. For example, when K8s' server resources are idle, Slurm's server resources may be in a state of shortage, and computation tasks can only wait in line, unable to promptly utilize K8s' idle resources. This irrational resource allocation not only results in resource waste, but can also lead to inefficient task execution, increased user waiting time, and increased overall server operating costs.

[0078] Breaking down the resource barriers between Kubernetes and Slurm, coordinating them with limited resources to achieve dynamic resource sharing and maximize utilization, has become a key issue urgently needed for computing scheduling and deployment platforms. Specifically, the current technical challenge is the lack of coordination and communication between the deployment and compute clusters to maximize server resource utilization.

[0079] Based on an in-depth analysis of the characteristics of K8s and Slurm, the inventors found that K8s is primarily responsible for application deployment. Once deployed, to ensure stability and continuity, applications should not be disturbed unless the deployment ends or the user manually stops them. This makes their resource usage relatively fixed. Slurm, on the other hand, is responsible for managing computing tasks, which have the flexibility to be interrupted and re-run, and can better adapt to dynamic changes in resources. Based on the differences in their characteristics, this application has constructed an innovative mechanism that uses K8s idle resources to support Slurm computing tasks. The specific improvement measures are as follows 1) to 4):

[0080] 1) Build a unified management system

[0081] A system for deploying and coordinating compute optimization under limited resources was built, deeply integrating Kubernetes and Slurm. This system, serving as the core hub for resource coordination, provides unified management of one or more server resources and displays the resource status of these servers in real time. This integration breaks the existing independent management model of Kubernetes and Slurm, enabling the interoperability and coordinated deployment of resource information.

[0082] 2) Create specifications and divide resource types

[0083] This application innovatively manages server resources in the form of "specifications", where each specification brings together different types of server resources, such as CPU, memory, GPU, hard disk, etc. For K8s, different resource types are set for each specification, which are divided into shared resources and exclusive resources. Shared resources can be used by both K8s and Slurm, while exclusive resources are only for K8s to use. This division of resource types not only ensures the special needs of K8s application deployment, but also provides the possibility for Slurm to utilize K8s idle resources. By flexibly setting resource specifications and types, users can allocate and manage resources more accurately according to actual business needs.

[0084] 3) Implement resource pre-occupancy strategy

[0085] When a user submits a work order requesting a specific shared resource type in Kubernetes, although the required resources are declared in the Kubernetes namespace, they are not actually used. Slurm can then utilize these temporarily idle resources to run the computation task. This strategy fully utilizes idle resource periods without affecting Kubernetes application deployment plans, effectively avoiding resource waste and significantly improving overall resource utilization.

[0086] 4) Improve resource coordination mechanisms

[0087] When a user deploys a task using a certain shared type specification of K8s, the system responds quickly. First, a Slurm placeholder task with the highest runtime priority and requiring the same resources is created on the server (this task only serves as a resource occupation indicator and does not actually use resources). This notifies Slurm that some resources on the current server have been occupied by K8s, preventing new Slurm computing tasks from entering. The system then uses K8s to create a deployment application on the target physical server for which resources have been pre-applied. Here, a placeholder task is created in a 1:1 ratio, and as the deployment task's life cycle progresses, the placeholder task is also generated and eliminated.

[0088] If the Slurm placeholder task is successfully created on the target physical server, it indicates that the current server resources are sufficient to meet the needs of the K8s application, and K8s can successfully create the application. If the Slurm placeholder task is not successfully created on the target physical server and is in a queued state, it indicates that the current server resources are insufficient. The system will immediately stop all Slurm computing tasks running on the target physical server to free up resources for creating the K8s application (K8s will continue to retry if the creation fails).

[0089] At this point, all Slurm computing tasks for the target physical server are stopped, and all Slurm computing tasks are requeued. The highest-priority Slurm placeholder task is executed first, and then successfully created and notified to the Slurm cluster. Some resources on the current server are already occupied, and K8s can try to create it successfully. Other stopped Slurm computing tasks wait to run on other available resources. This resource coordination mechanism ensures the smooth progress of K8s application deployment tasks, while also ensuring the reasonable execution of Slurm computing tasks under dynamic resource changes, and achieving effective coordination of deployment and computing. In other words, the advantage of the above innovative measures is that, without affecting the normal deployment of K8s, the retry characteristics of the computing cluster itself are utilized to achieve the utilization of server resources without interfering with the normal deployment of K8s and the computing of Slurm.

[0090] The specific implementations are as follows A1 to A4:

[0091] A1, build a system for coordinated optimization of deployment and computing under limited resources.

[0092] System Architecture Design: A microservices architecture was adopted to build a system for deployment and computational coordination optimization under limited resources, developed using frameworks such as Spring Cloud. The system primarily includes a resource management microservice, a user interface, a task scheduling microservice, an application deployment manager integration microservice, and a compute cluster manager integration microservice. These microservices communicate via RESTful APIs, enabling integration of Kubernetes and Slurm, and unified management of server resources.

[0093] Interface Development and Integration: Develop interfaces for interacting with Kubernetes and Slurm. For Kubernetes, use the official Kubernetes client libraries (such as the Kubernetes Python Client) to operate the Kubernetes cluster, enabling functions such as creating and deleting pods and querying resources. For Slurm, use command-line tools (such as squeue and sbatch) or its APIs to submit, query, and cancel Slurm tasks. These interfaces are then encapsulated into corresponding microservices for invocation by other modules.

[0094] Resource Monitoring and Display: The resource management microservice regularly collects server resource information, including CPU usage, memory usage, GPU status, and available hard disk space. A monitoring system is built using Prometheus and Grafana to store and visualize this collected resource data, allowing administrators to understand server resource status in real time. Prometheus and Grafana are open source monitoring tools. For example, for a server with the IP address 192.168.101.47, real-time usage of its 1600GB of memory, 8 GPU cards, 144 CPU cores, and 10TB of hard disk space can be displayed.

[0095] A2, create specifications and divide resource types.

[0096] Specification Definition Module Development: Developing a specification definition module for systems that coordinate and optimize deployment and computation under limited resources. Users access the specification definition page through the system's web interface or API. Here, they can create new specifications. For example, a user enters the specification name "47-Shared Computing Power-k8s" and sets the required resource quantities for the specification, namely, 200GB of memory, 4 GPU cards, 44 CPU cores, and 1TB of hard drive. Furthermore, the resource type is set for the K8s-related specification. Selecting "Shared Resources" here means that the resources in this specification can be used by Slurm when not in use by K8s. The system stores this specification information and resource type in the database. Similarly, a user can create a specification named "47-slurm," specifying 1400GB of memory, 5 GPU cards, 100 CPU cores, and 500GB of hard drive, and setting it as a shared resource type.

[0097] Resource availability check: After a user creates a specification, the system automatically checks whether the server's resources meet the requirements. The resource management microservice queries the current server's resource status. For example, if the total resources of server 192.168.101.47 do not meet the "47-Shared Computing Power-k8s" specification, the system will prompt the user and return a failure message when launching subsequent related tasks.

[0098] A3, resource pre-occupancy.

[0099] Work Order Submission and Processing: Users submit work orders for the "47-Shared Computing Power-k8s" specification through the web interface of the system for deployment and computational coordination optimization under limited resources. After receiving the work order, the system stores the work order information in the database and, through the Kubernetes integration microservice, declares the required resources in the Kubernetes cluster using a namespace. However, these resources are not actually allocated to the user's Kubernetes application at this point. At this point, the resources are not actually used; they are merely declared through the Kubernetes namespace so that Slurm can run the computation task.

[0100] Slurm task scheduling: For example, on the server 192.168.101.47, the "47-slurm" specification divides two tasks. Task A occupies 300GB of memory, 3 GPU cards, 50 CPU cores, and a 200GB hard disk; Task B occupies 300GB of memory, 2 GPU cards, 40 CPU cores, and a 200GB hard disk. Both tasks are running normally at the same time, but K8s has no tasks running at this time. In this case, Slurm can use the idle resources corresponding to the "47-Shared Computing Power-k8s" specification to run other computing tasks.

[0101] A4, resource coordination.

[0102] Receiving and processing deployment instructions: When a user issues a system instruction to deploy a task using the "47-Shared Computing Power-k8s" specification, the Task Scheduling microservice receives the instruction. First, the Task Scheduling microservice determines the target physical server for application deployment based on the instruction and calls the Slurm Integration microservice. This generates a Slurm script command to execute, requesting a Slurm placeholder task on the target physical server with the highest priority and the same resources as the "47-Shared Computing Power-k8s" specification (for identification purposes only; actual resources are not used). If the Slurm placeholder task is successfully created, the Task Scheduling microservice then calls the K8s Integration microservice to create and deploy the application on a node on the 192.168.101.47 server for which resources have been pre-allocated.

[0103] The actual Slurm script is as follows (it just declares how many resources are needed and does not run any tasks):

[0104] “#! / bin / bash / / Specifies that the script is executed using Bash (a command line interpreter) to ensure that the Bash syntax in the script can be parsed correctly;

[0105] #SBATCH -J share-job-20250630100931714 / / Set the job name to share-job-20250630100931714 for easy identification and management;

[0106] #SBATCH -p defs / / Specify that the job is submitted to the partition named defs;

[0107] #SBATCH -N 1 / / Apply for 1 computing node;

[0108] #SBATCH --nodelist=node-47 / / Explicitly specify that the job runs on node-47, bypassing the scheduler's automatic allocation;

[0109] #SBATCH -o job.%j.out / / Define the log file for standard output, %j will be replaced by the job ID;

[0110] #SBATCH -e job.%j.err / / Define the log file for error output, %j will be replaced by the job ID;

[0111] #SBATCH --requeue / / Allows jobs to be automatically requeued when they fail;

[0112] #SBATCH -A admin / / Specify the account to which the job belongs as admin, which is used for resource billing or permission control;

[0113] #SBATCH --qos=height / / Set the quality of service (qos) to height, associated priority or resource limit;

[0114] #SBATCH -c 44 / / Apply for 44 CPU cores;

[0115] #SBATCH --mem=200G / / Apply for 200G memory;

[0116] #SBATCH --gres=gpu:4 / / Apply for 4 GPUs;

[0117] while true; do

[0118] sleep 1000

[0119] done / / Infinite loop, sleeping every 1000 seconds".

[0120] Application creation results: When a user deploys a task on server 192.168.101.47 using the "47-Shared Computing Power-k8s" specification, a script is generated to create a Slurm placeholder task with the same resource requirements, but no resources are actually used. The Slurm placeholder task fails to be created because it requires four GPUs, but Slurm Tasks A and B together use five GPUs, while the server only has eight. The Task Scheduling microservice immediately calls the Slurm Integration microservice to stop all Slurm computing tasks (Tasks A and B) on the server and then creates the Slurm placeholder task. After the Slurm placeholder task is successfully created, it requeues the Slurm placeholder task, along with Tasks A and B, for execution, with the placeholder task being placed at the front of the queue. The Task Scheduling microservice then calls the K8s Integration microservice, which recreates the K8s application on the server. Meanwhile, Tasks A and B wait for execution on other available resources. Since there are still 4 idle cards at this time, only one of Task A and Task B can be executed normally, so one can be executed before the other one.

[0121] If the Slurm placeholder task is created successfully, the K8s deployment task is started directly.

[0122] This embodiment can achieve the following technical effects:

[0123] (1) Maximizing resource utilization

[0124] By building a unified management system (a system for coordinated optimization of deployment and computation under limited resources), innovatively classifying resource types and implementing resource pre-occupation and coordination mechanisms, we achieve deep integration and dynamic allocation of K8s and Slurm resources. Previously, during the early stages of application deployment or when applications were not fully loaded, K8s often left its managed server resources idle, while Slurm often had computing tasks waiting in queues due to insufficient resources. This approach breaks down this resource barrier, enabling Slurm to flexibly utilize K8s's idle resources and ensuring that every server resource is fully utilized at all times. For example, departments may use K8s to deploy business applications while simultaneously utilizing Slurm for computing tasks such as data analysis. With this approach, previously idle K8s resources are rationally allocated to Slurm tasks, significantly improving resource utilization efficiency and reducing resource waste.

[0125] (2) Save server overhead costs

[0126] Traditionally, separate server clusters are required to meet the needs of both K8s deployment and Slurm computing tasks, increasing costs in hardware procurement, equipment room rental, power consumption, and maintenance. This new approach enables resource sharing and collaboration between K8s and Slurm on the same server, significantly reducing the number of servers required while still meeting the same business needs.

[0127] (3) Facilitate deployment and coordinated computing operations

[0128] This method effectively solves the problem of resource unawareness and scheduling incoordination between K8s and Slurm by introducing an innovative resource coordination mechanism. When K8s is deploying applications, it can promptly identify resource occupancy by applying for Slurm placeholder tasks without actually using resources, ensuring that the deployment tasks proceed smoothly without interference; and when the K8s task is completed and resources are released, the Slurm task can quickly reallocate resources and continue execution. This close coordination mechanism breaks the previous situation where deployment and computing were isolated from each other, and realizes seamless connection and collaborative operation between the two. With the help of this method, when application deployment and data analysis tasks are carried out simultaneously, the system can dynamically adjust according to resource requirements, ensuring the efficient and stable operation of the business and greatly improving the user experience.

[0129] Based on the same inventive concept, an embodiment of the present application also provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the method of deployment and computing coordination optimization under limited resources of any of the above embodiments.

[0130] In an exemplary embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The electronic device 300 shown includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may further include a transceiver 304. It should be noted that in actual applications, the number of transceivers 304 is not limited to one, and the structure of the electronic device 300 does not constitute a limitation on the embodiments of the present application.

[0131] Processor 301 may be a CPU, GPU, DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, or a combination of a DSP and a microprocessor.

[0132] Bus 302 may include a path for transmitting information between the above components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. Bus 302 may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0133] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0134] The memory 303 is used to store computer program codes for executing the solution of the present application, and the execution is controlled by the processor 301. The processor 301 is used to execute the computer program codes stored in the memory 303 to implement the contents shown in the above method embodiment.

[0135] Among them, electronic devices include but are not limited to: mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0136] Based on the same inventive concept, an embodiment of the present application also provides a storage medium storing a computer program, wherein the computer program is configured to execute the method for deployment and computing coordination optimization under limited resources of any of the above embodiments at runtime.

[0137] Those skilled in the art will clearly understand that the specific working processes of the systems, devices, and modules described above can refer to the corresponding processes in the aforementioned method embodiments, and for the sake of brevity, they will not be further described here.

[0138] Those skilled in the art will understand that the technical solution of the present application, in essence, or in whole or in part, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes a number of program instructions that cause an electronic device (such as a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application when the program instructions are executed. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0139] Alternatively, all or part of the steps of implementing the aforementioned method embodiments may be accomplished by hardware related to program instructions (such as electronic devices such as personal computers, servers, or network devices), and the program instructions may be stored in a computer-readable storage medium. When the program instructions are executed by a processor of an electronic device, the electronic device executes all or part of the steps of the methods described in the various embodiments of the present application.

[0140] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that, within the spirit and principles of the present application, they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate from the protection scope of the present application.

Claims

1. A system for deployment and computational coordination optimization under limited resources, characterized in that: The system is built using a microservice architecture, which includes a resource management microservice, an interface terminal, a task scheduling microservice, an integrated microservice of an application deployment manager, and an integrated microservice of a computing cluster manager; The resource management microservice is used to collect information about server resources, create one or more specifications of server resources based on the information about server resources, and configure the type of server resources of each specification as shared resources or exclusive resources; The task scheduling microservice is configured to receive a work order submitted by a user through the interface to pre-occupy target-specification server resources for application deployment, and to call the application deployment manager's integrated microservice based on the work order, so that the application deployment manager's integrated microservice declares the required target-specification server resources; wherein the target-specification server resources are of a shared resource type; The task scheduling microservice is further configured to respond to one or more real-time computing tasks and call the integrated microservice of the computing cluster manager according to the one or more real-time computing tasks, so that the integrated microservice of the computing cluster manager uses the target specification server resources of the shared resource type to execute the one or more real-time computing tasks; The task scheduling microservice is further configured to receive an instruction from a user through the interface end to use the target specification server resources for application deployment, determine the target physical server to which the application is to be deployed based on the instruction, and call the integrated microservice of the computing cluster manager, so that the integrated microservice of the computing cluster manager creates a placeholder task on the target physical server with the same required computing resources as the target specification server resources. The placeholder task only occupies the computing resource quota but does not actually consume computing resources, that is, it is used to inform the computing cluster manager that the computing resources that are the same as the target specification server resources are occupied, but in fact, this part of the computing resources are not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment; the priority of the placeholder task is set to the highest; The task scheduling microservice is further configured to determine whether the placeholder task is created successfully. If the placeholder task is created successfully, the integration microservice of the application deployment manager is called to use the target specification server resources on the target physical server to deploy the application.

2. The system according to claim 1, wherein: The task scheduling microservice is also used by the integrated microservice of the computing cluster manager to apply for computing resources on the target physical server that are the same as the target specification server resources. If the application is approved, it is determined that the placeholder task is created successfully; if the application is not approved, it is determined that the placeholder task is not created successfully.

3. The system according to claim 1 or 2, characterized in that The task scheduling microservice is also used to call the integrated microservice of the computing cluster manager if the placeholder task is not created successfully, and the integrated microservice of the computing cluster manager stops all current computing tasks on the target physical server and creates the placeholder task. After the placeholder task is successfully created, the integrated microservice of the application deployment manager is called to use the target specification server resources on the target physical server to deploy the application.

4. The system according to claim 3, characterized in that The integrated microservice of the computing cluster manager is also used to re-queue the placeholder task and all the current computing tasks in order of priority for execution after the placeholder task is successfully created. The placeholder task will be put at the front for execution, informing the computing cluster manager that the computing resources with the same specifications as the target server resources are occupied. However, in fact, this part of the computing resources is not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment.

5. The system according to claim 4, characterized in that The integrated microservice of the computing cluster manager is also used to wait for other available resources to execute the re-queued computing tasks.

6. The system according to claim 1, wherein: The server resource information includes one or more of CPU usage, memory usage, GPU status, and available hard disk space.

7. The system according to claim 1, wherein: The resource management microservice is also used to provide information about server resources to the interface for display; The interface is configured to receive a request submitted by a user for creating one or more specifications of a server resource based on information about the server resource; The resource management microservice is further used to create one or more specifications of server resources according to the request, and configure the type of each specification of server resources as shared resources or exclusive resources.

8. A method for deployment and computation coordination optimization under limited resources, characterized in that: The method comprises: A system for deployment and computing coordination optimization under limited resources is built using a microservices architecture. The system includes a resource management microservice, an interface terminal, a task scheduling microservice, an integrated microservice for an application deployment manager, and an integrated microservice for a computing cluster manager. The resource management microservice collects information about server resources, creates one or more specifications of server resources based on the information about the server resources, and configures the type of server resources of each specification as shared resources or exclusive resources; The task scheduling microservice receives a work order submitted by a user through the interface to pre-occupy target-specification server resources for application deployment, calls the application deployment manager's integrated microservice based on the work order, and the application deployment manager's integrated microservice declares the required target-specification server resources; wherein the target-specification server resources are shared resources; The task scheduling microservice responds to one or more real-time computing tasks and calls the integration microservice of the computing cluster manager according to the one or more real-time computing tasks, and the integration microservice of the computing cluster manager uses the target specification server resources of the shared resource type to execute the one or more real-time computing tasks; The task scheduling microservice receives an instruction from the user through the interface end to use the target specification server resources for application deployment, determines the target physical server to be deployed on the basis of the instruction, and calls the integrated microservice of the computing cluster manager. The integrated microservice of the computing cluster manager creates a placeholder task on the target physical server with the same required computing resources as the target specification server resources. The placeholder task only occupies the computing resource quota but does not actually consume computing resources. That is, it is used to inform the computing cluster manager that the computing resources that are the same as the target specification server resources are occupied, but in fact, this part of the computing resources are not used by the placeholder task, but will be used by the integrated microservice of the application deployment manager for application deployment; the priority of the placeholder task is set to the highest; The task scheduling microservice determines whether the placeholder task is created successfully. If the placeholder task is created successfully, the integration microservice of the application deployment manager is called to use the target specification server resources on the target physical server to deploy the application.

9. An electronic device, characterized in that: The system comprises a processor and a memory, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the method for deployment and computing coordination optimization under limited resources according to claim 8.

10. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method for coordinated optimization of deployment and computing under limited resources according to claim 8 when running.

Images