Calculation network resource scheduling system and method

Abstract

The invention discloses a computing network resource scheduling system and method. Relates to the field of cloud and edge computing, and comprises a resource management layer used for managing network heterogeneous resources, the network heterogeneous resources comprising at least one of computing resources and network configuration resources; the computing network sensing layer is in communication connection with the resource management layer and is used for converting the network heterogeneous resources into resource indexes according to a preset measurement standard and monitoring the resource use state of the network heterogeneous resources in the resource management layer in real time, and the resource indexes are indexes conforming to the data format of the scheduling resources of the scheduling layer; and the arrangement scheduling layer is in communication connection with the computing network sensing layer and is used for generating a resource scheduling scheme through a preset multi-factor scheduling algorithm, the resource use state and the resource demand index of the user. Through the method and the device, the problem of low computing network resource scheduling efficiency in related technologies is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of cloud and edge computing, in particular, to an algorithm network resource scheduling system and method. BACKGROUND

[0002] In the field of fusion of algorithm network and communication network, the resource management and scheduling model in the related art faces many challenges, especially in the efficient use of heterogeneous resources and energy optimization. A two-layer architecture of "resource management layer-scheduling layer" is adopted, which can provide basic resource orchestration services, but its limitations lead to insufficient efficiency and flexibility of resource scheduling. Algorithm resources and network resources are usually managed separately, and different controllers and platforms are responsible for them. This fragmented management makes it difficult for algorithm and network resources to achieve seamless collaboration and linkage, especially in scenarios that require heterogeneous hardware resources to participate simultaneously. The coordination efficiency between resources is severely affected. The current resource scheduling mechanism relies on preset templates or static strategies, which are not effective when facing dynamic scenarios (such as sudden changes in frame rate in cloud rendering). This static scheduling method cannot automatically adapt to changes in load and requires manual intervention to adjust, thereby introducing additional latency and management burden.

[0003] The resource management scheme in the related art lacks fine control of the energy consumption of the underlying hardware, such as the lack of "fusion pressure-fusion disc" and other hardware-level energy consumption adjustment mechanisms. This directly leads to waste of energy during resource utilization, especially during high load periods, and the inability to effectively adjust hardware states to optimize energy consumption, resulting in a decrease in overall energy efficiency. There is a clear breakpoint in the scheduling collaboration between the application layer and the resource management layer, especially between the heterogeneous resource access (perception layer) and orchestration (management layer) of the algorithm network. This breakpoint in resource scheduling results in a long response time for algorithm and network resources in dealing with high-concurrency application requests, further exacerbating the delay and inefficiency of resource allocation.

[0004] Currently, there is no effective solution to the problem of low efficiency of algorithm network resource scheduling in the related art. SUMMARY

[0005] The main purpose of the present application is to provide an algorithm network resource scheduling system and method to solve the problem of low efficiency of algorithm network resource scheduling in the related art.

[0006] In order to achieve the above object, according to one aspect of the present application, a kind of algorithm network resource scheduling system is provided.The system includes: resource management layer, for managing network heterogeneous resources, wherein network heterogeneous resources include at least one of: computing resources and network configuration resources;Algorithm network perception layer is connected with resource management layer, for converting network heterogeneous resources into resource index according to preset metric standard, and real-time monitoring the resource usage state of network heterogeneous resources in resource management layer, wherein resource index is the index that conforms to the data format of the scheduling resource of arrangement scheduling layer;Arrangement scheduling layer is connected with algorithm network perception layer, for generating resource scheduling scheme by preset multi-factor scheduling algorithm, resource usage state and user's resource demand index, wherein preset multi-factor scheduling algorithm is used to generate candidate scheduling scheme randomly based on network heterogeneous resources, from candidate scheduling scheme, the resource scheduling scheme that meets resource demand index is filtered out, the computing efficiency in resource usage state is maximized, and the power consumption cost in resource usage state is minimized.

[0007] Optionally, the resource management layer includes: a resource configuration module for configuring network heterogeneous resources through virtualization and pooling operations; a resource monitoring module for monitoring the resource usage state of network heterogeneous resources and sending the resource usage state to the algorithm network perception layer; a resource scheduling module for scheduling network heterogeneous resources through the resource scheduling scheme; and a resource recycling module for recycling network heterogeneous resources when detecting that network heterogeneous resources are in an idle state.

[0008] Optionally, the algorithm network perception layer includes: an abstract modeling module for inputting network heterogeneous resources into a quantumization measurement model to obtain resource indexes, wherein the resource indexes are resource indexes divided according to a preset metric standard; an encapsulation module for encapsulating network heterogeneous resources into target interfaces according to the resource indexes; a perception module for collecting the resource usage state sent by the resource management layer and performing data preprocessing on the resource usage state to obtain resource usage state in a data processing format conforming to the arrangement scheduling layer; a configuration module for adjusting configuration information and allocation state of network heterogeneous resources according to the resource usage state; and an access module for providing the target interfaces to the arrangement management layer.

[0009] Optionally, the arrangement scheduling layer includes: a service-oriented scheduling module for providing resource scheduling services to the application layer; a resource visualization module for displaying the resource usage state of network heterogeneous resources; a resource metering module for real-time statistics of consumption and performance indicators of network heterogeneous resources, wherein the performance indicators include at least one of: processing speed and delay; a monitoring and analysis module for monitoring user resource demand satisfaction and resource service quality indicators; and an arrangement scheduling module for planning network heterogeneous resources through preset multi-factor scheduling algorithm, resource usage state and resource demand index to generate a resource scheduling scheme.

[0010] Optionally, the orchestration scheduling layer is further configured to send a resource scheduling instruction to the algorithm-network perception layer based on the resource scheduling scheme, the algorithm-network perception layer is further configured to adjust configuration information and an allocation state of the network heterogeneous resources based on the resource scheduling instruction, and send the configuration information and the allocation state to the resource management layer, and the resource management layer is further configured to schedule the network heterogeneous resources based on the configuration information and the allocation state.

[0011] Optionally, the preset multi-factor scheduling algorithm randomly generates a plurality of candidate resource scheduling schemes based on the network heterogeneous resources, eliminates candidate scheduling schemes that do not meet the resource demand index, and selects the resource scheduling scheme from the candidate scheduling schemes that meet the resource demand index according to a preset objective function, wherein the preset objective function aims to minimize power consumption cost and maximize calculation efficiency.

[0012] Optionally, the system further comprises a resource twin library configured to map the resource usage state of the network heterogeneous resources, the resource twin library is updated in real time based on the resource usage state monitored by the algorithm-network perception layer, and the orchestration scheduling layer obtains the resource usage state through the resource twin library.

[0013] Optionally, the system further comprises an application layer in communication connection with the orchestration scheduling layer, configured to receive a resource scheduling request initiated by a user, and send the resource demand index in the resource scheduling request to the orchestration scheduling layer.

[0014] According to another aspect of the present application, an algorithm-network resource scheduling method applied to the algorithm-network resource scheduling system is provided. The method comprises: obtaining a resource demand index of a user and determining a resource usage state of network heterogeneous resources; inputting the resource demand index and the resource usage state into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and scheduling the network heterogeneous resources to a client of the user based on the resource scheduling scheme.

[0015] Optionally, inputting the resource demand index and the resource usage state into the preset multi-factor scheduling algorithm to obtain the resource scheduling scheme comprises: constructing a constraint condition based on the resource demand index, constructing a first objective function between a scheduled resource index and calculation efficiency, and constructing a second objective function between the scheduled resource index and power consumption cost; combining the constraint condition, the first objective function and the second objective function into a planning model; solving the planning model by a Pareto frontier solving algorithm to obtain a target solution, wherein the target solution is a target resource index scheduling amount that meets the constraint condition, maximizes the calculation efficiency and minimizes the power consumption cost; and generating the resource scheduling scheme based on the target resource index scheduling amount.

[0016] In order to achieve the above object, according to another aspect of the present application, a network resource scheduling device is provided. The device comprises: an acquisition unit configured to acquire a resource demand index of a user and determine a resource usage state of a network heterogeneous resource; an input unit configured to input the resource demand index and the resource usage state into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and a scheduling unit configured to schedule the network heterogeneous resource to a client of the user based on the resource scheduling scheme.

[0017] In order to achieve the above object, according to another aspect of the present application, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the network resource scheduling method described in various embodiments of the present application.

[0018] According to the present application, a resource management layer is adopted to manage network heterogeneous resources, wherein the network heterogeneous resources comprise at least one of computing resources and network configuration resources; a network calculation perception layer is in communication connection with the resource management layer and is configured to convert the network heterogeneous resources into resource indexes according to a preset measurement standard and monitor a resource usage state of the network heterogeneous resources in the resource management layer in real time, wherein the resource indexes are indexes conforming to a data format of scheduling resources of an arrangement scheduling layer; and the arrangement scheduling layer is in communication connection with the network calculation perception layer and is configured to generate a resource scheduling scheme through a preset multi-factor scheduling algorithm, the resource usage state and a resource demand index of a user, wherein the preset multi-factor scheduling algorithm is configured to randomly generate a candidate scheduling scheme based on the network heterogeneous resources, and select a resource scheduling scheme from the candidate scheduling scheme, which conforms to the resource demand index, maximizes a computing efficiency in the resource usage state and minimizes a power consumption cost in the resource usage state, thereby solving the problem of low network resource scheduling efficiency in the related art. The three-layer system of the resource management layer, the network calculation perception layer and the arrangement scheduling layer is adopted, the resource management layer can effectively manage the network heterogeneous resources including the computing resources and the network configuration resources, thereby improving the integration degree and utilization rate of the resources and reducing the situation of resource fragmentation and idling. The network calculation perception layer can convert the network heterogeneous resources into unified resource indexes according to the preset measurement standard through the communication connection with the resource management layer. The arrangement scheduling layer can automatically generate the resource scheduling scheme based on the real-time monitored resource usage state and the specific resource demand index of the user through the preset multi-factor scheduling algorithm, can make the optimal resource allocation decision under the complex and changeable business environment, and thus achieves the effect of improving the network resource scheduling efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application, and are used to interpret the illustrative embodiments of the present application and their descriptions, and do not constitute improper limitations to the present application. In the drawings:

[0020] Figure 1is a schematic diagram of an algorithm network resource scheduling system according to an embodiment of the present application;

[0021] Figure 2 is a schematic diagram of an algorithm network resource scheduling flow according to an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of a resource scheduling scheme generation flow according to an embodiment of the present application;

[0023] Figure 4 is a flowchart of an algorithm network resource scheduling method according to an embodiment of the present application;

[0024] Figure 5 is a schematic diagram of an algorithm network resource scheduling device according to an embodiment of the present application;

[0025] Figure 6 is a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] It should be noted that the embodiments and features of the embodiments in the present application can be combined with each other without conflict. The technical solutions in the embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0027] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0028] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to include only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to the process, method, product or device.

[0029] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for display, analyzed data, etc.) involved in the present disclosure are all information and data authorized by the user or authorized by all parties.

[0030] It should be noted that the collected information is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards in the relevant region, necessary security measures are taken, it does not violate public order and good customs, and provides corresponding operation portal for user to choose authorization or refusal.

[0031] The application will be described below in conjunction with the preferred implementation steps, Figure 1 is a schematic diagram of the algorithm network resource scheduling system provided by the embodiment of the application, as Figure 1 shown, the system comprises:

[0032] The resource management layer is configured to manage network heterogeneous resources, wherein the network heterogeneous resources comprise at least one of the following: computing resources and network configuration resources.

[0033] Specifically, the resource management layer function is the unified management of computers and network heterogeneous resources, which is responsible for the virtualization and pool management of resources. The resources can be computing resources such as CPU, memory, storage devices, etc., or network resources such as inter-cluster network, intra-cluster network, backbone network, metropolitan area network, transmission network and other heterogeneous network resources. The following are some key functions and features of the resource management layer: according to the strategy generated by the multi-factor scheduling algorithm, the resources are allocated, including the allocation of CPU (Central Processing Unit), GPU (Graphics Processing Unit), NPU (Neural Network Processing Unit), ARM (Advanced RISC Machines) memory, storage and other computing resources, and the configuration of network resources such as intra-cloud network and inter-cloud private line. Monitor the usage and running status of algorithm network resources to ensure that the resources can meet the customer's needs continuously. According to the customer's SLA (Service Level Agreement) requirements and real-time data resources and state information, real-time resource allocation is performed. When the resources are no longer needed, they are recycled for other purposes in real time. Various strategies and technologies are used to improve the utilization efficiency of resources. Ensure that the access and use of resources comply with security policies and standards.

[0034] The algorithm network perception layer is in communication connection with the resource management layer, configured to convert the network heterogeneous resources into resource indicators according to a preset measurement standard, and monitor the resource usage state of the network heterogeneous resources in the resource management layer in real time, wherein the resource indicators are indicators in a data format conforming to the scheduling resources of the orchestration and scheduling layer.

[0035] Specifically, the algorithm network perception layer focuses on unified abstraction, definition and description of diversified heterogeneous algorithm network resources, and measurement based on preset measurement standards. Algorithm network related information is abstracted, perceived and collected in a unified format through edge cloud management, network controllers and other means, realizing unified data encapsulation. Through the quantization measurement model, the algorithm network perception layer can abstract and quantify the characteristics of heterogeneous hardware (such as CPU, GPU, NPU, network devices, etc.) such as processing power, bandwidth, delay, power consumption, etc. into a set of standardized measurement indicators. This allows different types of resources to be described uniformly, facilitating subsequent resource scheduling and optimization. The space-time compression algorithm is used to reduce the monitoring overhead of resource state, and through efficient data compression methods, real-time collection and processing of resource usage data is realized to ensure the accuracy and timeliness of resource state. This algorithm can quickly compress massive space-time data into lower dimensions while retaining key information, improving resource monitoring efficiency.

[0036] The algorithm network perception layer communicates closely with the resource management layer to obtain the usage status of network heterogeneous resources in real time, including the occupancy rate of computing resources, traffic, delay and packet loss rate of network devices. Real-time monitoring capability ensures that the orchestration and scheduling layer can obtain the latest resource usage to make quick and accurate scheduling decisions. The algorithm network perception layer converts the collected raw resource data into a data format and indicators required by the orchestration and scheduling layer. The algorithm network perception layer feeds back the transformed resource indicators and monitored resource usage status to the orchestration and scheduling layer, providing real-time resource information for the multi-factor dynamic planning algorithm, thereby realizing closed-loop control and dynamic optimization.

[0037] The orchestration and scheduling layer is in communication connection with the algorithm network perception layer, and is configured to generate a resource scheduling scheme by a preset multi-factor scheduling algorithm, resource usage status and user resource demand indicators. The preset multi-factor scheduling algorithm is configured to randomly generate a candidate scheduling scheme based on network heterogeneous resources, and to select a resource scheduling scheme that meets the resource demand indicators, maximizes the computing efficiency in the resource usage status, and minimizes the power consumption cost in the resource usage status from the candidate scheduling scheme.

[0038] Specifically, the orchestration and scheduling layer continuously perceives the current resource usage status and QoS (Quality of Service) indicators of the application during the running of the application, calculates a scheduling scheme that best matches the application demand and optimizes resource utilization according to the initial demand of the application, perceived resource information, and multiple factors such as cost, resource utilization, QoS indicators, and adjusts the resources of the application according to the scheduling scheme.

[0039] The orchestration scheduling layer receives user resource requests from the service layer and parses the SLA requirements therein, including but not limited to computing efficiency, network bandwidth, delay, cost budget, and other parameters. User demand indicators are converted into hard constraints and optimization targets for resource allocation, providing guidance for subsequent resource scheduling. The orchestration scheduling layer communicates with the algorithm and network perception layer to query the current resource usage status, including computing resource utilization, network link bandwidth and delay status, etc. Using a preset multi-factor scheduling algorithm, the orchestration scheduling layer generates multiple candidate scheduling schemes based on user resource requirements and real-time resource usage status.

[0040] The multi-factor scheduling algorithm considers multiple dimensions such as computing efficiency, power consumption cost, and network performance, ensuring that the allocation of resources can achieve optimal performance and cost balance under the premise of meeting hard constraints. Through a closed-loop mechanism of policy-driven and state feedback, the orchestration scheduling layer evaluates the applicability of each candidate scheduling scheme, i.e., whether the scheme meets the hard constraints in the user SLA requirements such as delay and service availability. Subsequently, through a dynamic weight adjustment mechanism, the orchestration scheduling layer calculates the weighted scores of each optimization item (such as computing efficiency, power consumption cost), and finally selects a resource scheduling scheme that can meet the hard constraints and maximize computing efficiency and minimize power consumption cost.

[0041] For example, Figure 2 is a schematic diagram of the algorithm and network resource scheduling process provided by an embodiment of the present application, as Figure 2 shown, a customer initiates an algorithm and network scheduling service request;

[0042] The service management platform (such as cloud rendering, cloud gaming, video, etc.) sends service request data to the algorithm and network integrated management and control scheduling platform;

[0043] The orchestration scheduling layer function block interacts with the resource twin library to query computing power and network resources;

[0044] The edge cloud management platform and network controller are called to execute the scheduling strategy, i.e., to issue resource allocation instructions, and finally to realize resource allocation.

[0045] The computing network resource scheduling system provided in this application embodiment manages heterogeneous network resources through a resource management layer, wherein the heterogeneous network resources include at least one of the following: computing resources and network configuration resources; a computing network perception layer, which is communicatively connected to the resource management layer, is used to convert heterogeneous network resources into resource indicators according to a preset metric standard, and to monitor the resource usage status of heterogeneous network resources in the resource management layer in real time, wherein the resource indicators are indicators that conform to the data format of the orchestration and scheduling layer for scheduling resources; an orchestration and scheduling layer, which is communicatively connected to the computing network perception layer, is used to generate resource scheduling schemes through a preset multi-factor scheduling algorithm, resource usage status, and user resource demand indicators, wherein the preset multi-factor scheduling algorithm is used to randomly generate candidate scheduling schemes based on heterogeneous network resources, and to select resource scheduling schemes that meet the resource demand indicators, maximize the computing efficiency in the resource usage status, and minimize the power consumption cost in the resource usage status from the candidate scheduling schemes, thereby solving the problem of low computing network resource scheduling efficiency in related technologies. The system adopts a three-layer architecture: Resource Management Layer, Network Awareness Layer, and Orchestration and Scheduling Layer. The Resource Management Layer effectively manages heterogeneous network resources, including computing and network configuration resources, improving resource integration and utilization while reducing resource fragmentation and idleness. The Network Awareness Layer, through communication with the Resource Management Layer, transforms heterogeneous network resources into unified resource indicators according to preset metrics. The Orchestration and Scheduling Layer, using a preset multi-factor scheduling algorithm, automatically generates resource scheduling schemes based on real-time monitored resource usage status and specific user resource requirements. This enables optimal resource allocation decisions in complex and ever-changing business environments, thereby improving the efficiency of network resource scheduling.

[0046] Optionally, in the computing network resource scheduling system provided in this application embodiment, the resource management layer includes: a resource configuration module, used to configure heterogeneous network resources through virtualization and pooling operations; a resource monitoring module, used to monitor the resource usage status of heterogeneous network resources and send the resource usage status to the computing network perception layer; a resource scheduling module, used to schedule heterogeneous network resources through a resource scheduling scheme; and a resource recycling module, used to recycle heterogeneous network resources when it is detected that the heterogeneous network resources are in an idle state.

[0047] In some embodiments, the resource configuration module converts physical resources into virtual resources using virtualization technology, such as creating virtual instances on computing resources like CPUs, GPUs, and NPUs, and network devices (like routers and switches), to achieve flexible resource allocation and isolation. The virtualized resources are further integrated into resource pools for unified management and elastic allocation. Pooling operations include resource classification, storage, and management, ensuring rapid and accurate allocation to users or services when needed. The resource monitoring module continuously monitors the usage of heterogeneous network resources, including key performance indicators such as compute node load, network link traffic, latency, and packet loss rate. The frequency and accuracy of monitoring data collection directly affect the accuracy of scheduling decisions. The resource monitoring module sends the collected resource usage status to the network perception layer in real time, providing a basis for subsequent resource scheduling and optimization.

[0048] The resource scheduling module, based on the resource scheduling scheme from the orchestration layer, is responsible for executing specific resource allocation and configuration operations, ensuring that resources are efficiently and rationally distributed across various computing tasks or services. The resource scheduling module can dynamically adjust resource allocation strategies based on real-time resource usage status and business needs to cope with sudden load changes or optimize resource utilization efficiency. The resource reclamation module continuously monitors the usage status of heterogeneous network resources, identifying those that are idle or inefficiently used. Once a resource is detected as idle, the resource reclamation module initiates a reclamation operation, releasing the resource and returning it to the resource pool for subsequent reallocation, thus avoiding resource waste.

[0049] In this embodiment, the resource management layer achieves unified management of different types of heterogeneous resources (such as CPUs, GPUs, and network devices) through virtualization and pooling operations, simplifying the scheduling and configuration of cross-layer resources. The dynamic resource allocation capability of the resource scheduling module enables the system to adaptively adjust resource allocation strategies based on real-time resource status and business needs, achieving intelligent resource scheduling and improving service response speed and resource utilization. The resource recycling module can promptly reclaim and reuse idle resources, which not only reduces resource idleness but also lowers overall operating costs and energy consumption, aligning with the principles of green computing and cost-effectiveness.

[0050] Optionally, in the computing network resource scheduling system provided in this application embodiment, the computing network perception layer includes: an abstract modeling module, used to input heterogeneous network resources into a quantum metric model to obtain resource indicators, wherein the resource indicators are resource indicators divided according to a preset metric standard; an encapsulation module, used to encapsulate heterogeneous network resources into target interfaces according to the resource indicators; a perception module, used to collect resource usage status sent by the resource management layer, and perform data preprocessing on the resource usage status to obtain resource usage status that conforms to the data processing format of the orchestration and scheduling layer; a configuration module, used to adjust the configuration information and allocation status of heterogeneous network resources according to the resource usage status; and an access module, used to provide target interfaces to the orchestration and management layer.

[0051] In some embodiments, the abstract modeling module employs a quantum metric model to transform the characteristics of heterogeneous network resources into resource indicators conforming to preset metric standards. These indicators may include, but are not limited to, processing power, network bandwidth, latency, and power consumption. They are classified and quantified according to preset standards to achieve a standardized description of resource characteristics. Through model calculations, the characteristics of heterogeneous resources are transformed into a series of resource indicators, which become important inputs for resource scheduling decisions, ensuring that resource scheduling algorithms can accurately understand and evaluate resource status. The encapsulation module encapsulates the generated resource indicators according to the target interface format, ensuring that this data can be correctly identified and processed by the scheduling algorithms of the orchestration layer. The target interface is designed to follow a unified data exchange protocol, facilitating cross-layer communication and data interaction. Through encapsulation operations, the characteristic information of heterogeneous network resources is adaptively transformed into the input format required by the orchestration layer, which helps improve the accuracy and efficiency of resource scheduling decisions.

[0052] The perception module continuously collects resource usage status data sent by the resource management layer, including computing resource occupancy, network link traffic, and performance. This raw data is then preprocessed, including data cleaning, format conversion, and standardization, to conform to the data processing format of the orchestration and scheduling layer. This preprocessing ensures data consistency and readability, facilitating further analysis and utilization by the orchestration and scheduling layer and supporting the generation of resource scheduling decisions. Based on the preprocessed resource usage status, the configuration module can adjust the configuration information and allocation status of heterogeneous network resources in real time, such as dynamically adjusting CPU frequency and network bandwidth allocation, to optimize resource utilization efficiency and response speed. By executing resource optimization strategies, the configuration module achieves dynamic monitoring and real-time adjustment of resource status, ensuring continuous resource optimization and efficient utilization. The access module provides a target interface to the orchestration and scheduling layer as a channel for resource information transmission, supporting real-time querying and scheduling of heterogeneous network resources. Through the target interface, the access module facilitates information flow between the orchestration and scheduling layer and the resource management layer, ensuring accurate delivery and feedback of resource scheduling decisions and achieving high efficiency in cross-layer scheduling.

[0053] The abstract modeling and encapsulation operations of the computing network perception layer in this embodiment ensure standardized descriptions of resource characteristics and unified data formats, improving the compatibility and efficiency of cross-layer scheduling and reducing the overhead of protocol conversion and data processing. The real-time monitoring and data preprocessing capabilities of the perception module enable the system to respond quickly to changes in resource status. Through the immediate adjustment of the configuration module, the response speed and processing capacity to sudden resource demands are improved. By providing standardized resource usage status data, the computing network perception layer supports the orchestration and scheduling layer in generating more accurate and optimized resource scheduling decisions, ensuring that resource allocation meets business needs while maximizing resource utilization efficiency and cost-effectiveness. The access module, through the provision of target interfaces, strengthens communication and collaboration between the orchestration and scheduling layer and the resource management layer, realizing closed-loop control of resource scheduling strategies and improving the stability and service quality of the entire computing network.

[0054] Optionally, in the network resource scheduling system provided in this application embodiment, the orchestration and scheduling layer includes: a service-oriented scheduling module for providing resource scheduling services to the application layer; a resource visualization module for displaying the resource usage status of heterogeneous network resources; a resource metering module for real-time statistics of the consumption and performance indicators of heterogeneous network resources, wherein the performance indicators include at least one of the following: processing speed and latency; a monitoring and analysis module for monitoring the user resource demand satisfaction and resource service quality indicators; and an orchestration and scheduling module for planning heterogeneous network resources and generating resource scheduling schemes through preset multi-factor scheduling algorithms, resource usage status, and resource demand indicators.

[0055] In some embodiments, the service-oriented scheduling module acts as a "service-oriented" interface for resource scheduling capabilities. It directly faces the application layer, providing highly encapsulated resource scheduling services. Users or applications can acquire or release resources through simple service calls, such as API requests, without needing to delve into the details of the underlying resources or scheduling algorithms. A flexible and easy-to-use interface is designed to ensure the transparency and convenience of resource scheduling while hiding the complexity of the underlying scheduling logic, improving the user and application experience. The resource visualization module is responsible for presenting the real-time usage status of heterogeneous network resources to users or management interfaces in an intuitive way. This includes graphical interfaces, such as dashboards, charts, or maps, displaying key indicators such as resource utilization and load, facilitating monitoring and decision-making. Through the visualization of resource pools, users can see the collection of all available resources and how they are allocated and used, helping them better understand resource distribution and usage.

[0056] The resource metering module continuously tracks and statistically analyzes the actual consumption of heterogeneous network resources, including CPU time, memory usage, storage usage, and network traffic. It also monitors resource performance metrics such as processing speed and latency. The monitoring and analysis module periodically checks the degree to which user resource needs are met, ensuring that services can respond promptly to users' actual needs and improve user satisfaction. It continuously analyzes resource service quality metrics, such as SLA achievement rate and QoS metrics, to evaluate the effectiveness of resource scheduling strategies and provide data support for subsequent strategy optimization. The orchestration and scheduling module is the brain of the computing network. Based on user SLA requirements, real-time resource usage status, and multi-factor scheduling algorithms, it intelligently plans heterogeneous network resources and generates optimal resource scheduling schemes. When generating resource scheduling schemes, the orchestration and scheduling module comprehensively considers multiple optimization objectives such as processing speed, latency, energy consumption, and cost, achieving a balance between service quality and resource efficiency. The orchestration and scheduling module works closely with the computing network perception layer to form a closed-loop control mechanism, adjusting scheduling strategies in real time to cope with dynamic changes in resource status and user needs, ensuring the flexibility and real-time nature of resource scheduling.

[0057] The service-oriented scheduling module in this embodiment greatly improves the user and application experience and reduces the complexity of resource management by providing a simple and easy-to-use resource scheduling service interface. The orchestration and scheduling module achieves an optimal balance between resource efficiency and service quality through comprehensive consideration of multi-factor scheduling algorithms and dynamic closed-loop control, improving resource utilization and user satisfaction. The monitoring and analysis module provides real-time monitoring of user resource demand satisfaction and service quality indicators, offering real-time feedback and data support for resource scheduling decisions, ensuring the real-time nature and accuracy of resource scheduling. The resource metering module's real-time statistics and performance indicator monitoring support cost-benefit analysis and the formulation of energy-saving strategies, helping to reduce operating costs and improve the system's environmental adaptability.

[0058] Optionally, in the computing network resource scheduling system provided in this application embodiment, the orchestration and scheduling layer is further used to issue resource scheduling instructions to the computing network perception layer based on the resource scheduling scheme. The computing network perception layer is further used to adjust the configuration information and allocation status of heterogeneous network resources based on the resource scheduling instructions, and send the configuration information and allocation status to the resource management layer. The resource management layer is further used to schedule heterogeneous network resources based on the configuration information and allocation status.

[0059] In some embodiments, cross-layer collaboration achieves closed-loop collaboration through top-down policy-driven approaches and bottom-up state feedback. When a user initiates a business request from the application layer (such as cloud rendering or cloud host deployment), the orchestration and scheduling layer receives SLA requirements and optimization objectives (cost / energy consumption) through the "service-oriented scheduling capability" interface, initiates a multi-factor scheduling algorithm to generate a strategy, and issues standardized instructions to the perception layer. The perception layer uses the "unified configuration of computing power and network" capability to call the multi-cluster management system (K8S / OpenStack) to coordinate underlying resources. The resource management layer performs physical resource allocation and transmits metering data back in real time. After ETL (Extract, Transform, Load) processing by the perception layer, monitoring indicators are generated and fed back to the orchestration layer to dynamically optimize the closed-loop strategy (such as scaling up or down). Finally, resource pooling visualization, cost analysis, and service delivery are realized at the application layer.

[0060] The cross-layer interaction mechanism in this embodiment forms a closed-loop control system capable of real-time monitoring of resource usage status, rapid response to changes in user demand and network conditions, dynamic adjustment and optimization of resources, and ensuring stable and efficient service quality. Through intelligent decision-making at the orchestration and scheduling layer, combined with resource abstraction at the network perception layer and direct scheduling at the resource management layer, unified management and intelligent scheduling of heterogeneous network resources are achieved, improving resource utilization efficiency and scheduling accuracy. The cross-layer mechanism ensures that resource scheduling decisions fully consider multiple factors such as user SLA, cost, and energy consumption, meeting high service quality standards while effectively controlling costs, reflecting the intelligence and economic benefits of the computing network. The security sandbox recycling mechanism at the resource management layer, combined with resource status monitoring at the network perception layer, ensures secure resource isolation while promptly clearing and protecting user data, enhancing the overall system security and data protection capabilities.

[0061] Optionally, in the computing network resource scheduling system provided in this application embodiment, a preset multi-factor scheduling algorithm randomly generates multiple candidate resource scheduling schemes based on heterogeneous network resources, eliminates candidate scheduling schemes that do not meet resource demand indicators, and selects resource scheduling schemes from the candidate scheduling schemes that meet resource demand indicators according to a preset objective function, wherein the preset objective function aims to minimize power consumption cost and maximize computing efficiency.

[0062] In some embodiments, the algorithm first analyzes the resource requirements of users or applications, including computing requirements (CPU and GPU utilization), network requirements (bandwidth and latency), and storage requirements. Based on the resource requirement analysis, the algorithm randomly generates multiple candidate resource scheduling schemes, covering potential allocation methods for heterogeneous network resources, including the allocation of computing resources, storage resources, and network link configurations. Among the generated candidate scheduling schemes, the algorithm first eliminates those that do not meet the resource requirement indicators. For example, if the CPU utilization of a scheme is lower than the application's minimum requirements, or the network latency exceeds an acceptable range, the scheme will be excluded, thereby narrowing the selection range and focusing on scheduling schemes that can theoretically meet the requirements.

[0063] For candidate scheduling schemes that pass the initial screening and meet the resource requirement indicators, the algorithm further optimizes them using a pre-defined objective function. This objective function has two objectives: minimizing power consumption and maximizing computational efficiency. It aims to find the optimal scheduling strategy that simultaneously satisfies these two optimization goals. Through the optimization calculation of the objective function, the algorithm selects the best resource scheduling scheme from the candidate schemes.

[0064] For example, Figure 3 This is a schematic diagram of the resource scheduling scheme generation process provided in the embodiments of this application, such as... Figure 3 As shown, a two-layer decision-making model is constructed, with customer SLAs as absolute constraints and multi-dimensional optimization objectives as dynamic weights. First, based on real-time monitoring data, a hard-constraint filter quickly eliminates all computing nodes that violate SLA latency, computing power, and availability requirements. Second, within the compliant node set, a multi-objective optimization function is established, focusing on power consumption cost and computational efficiency. The Pareto front algorithm is used to obtain a non-dominated solution set (the set of achievable best trade-offs in multi-objective optimization). Finally, combined with an environment-aware dynamic weight adjustment mechanism, the weighted scores of each optimization item are adaptively calculated based on real-time electricity price fluctuations, peak and valley characteristics of business load, and carbon emission intensity, achieving optimal energy efficiency cost scheduling decisions under the premise of zero SLA default. This innovative architecture, employing hierarchical constraint processing and multi-factor collaborative optimization, resolves the technical contradiction in traditional scheduling where service quality assurance and resource utilization efficiency are difficult to balance.

[0065] This embodiment achieves efficient resource scheduling in a complex heterogeneous resource environment by using a preset multi-factor scheduling algorithm, while also taking into account cost control and performance optimization to improve the efficiency of computing network resource scheduling.

[0066] Optionally, in the computing network resource scheduling system provided in this application embodiment, the system further includes: a resource twin library, used to map the resource usage status of heterogeneous network resources. The resource twin library is updated in real time based on the resource usage status monitored by the computing network perception layer, and the orchestration and scheduling layer obtains the resource usage status through the resource twin library.

[0067] In some real-time examples, one of the core functions of a resource twin library is to build digital twin models of heterogeneous network resources. These models accurately reflect the real-time status of resources, including key performance indicators such as CPU utilization, GPU usage time, network bandwidth consumption, latency, and packet loss rate. Through continuous monitoring and mapping, the resource twin library can generate "real-time snapshots" of resources, providing the data foundation needed for decision-making at the orchestration and scheduling layer. The resource twin library is closely connected to the network perception layer, which is responsible for collecting and preprocessing resource usage status data from the resource management layer. Based on the resource usage status monitored by the network perception layer, the resource twin library achieves real-time data updates. In addition to real-time status mapping, the resource twin library also stores historical resource usage data, which provides valuable information for predictive analysis of resources. Through historical data, future resource demand and performance can be predicted, providing forward-looking insights for scheduling decisions. The analysis of historical data also provides a basis for the iterative improvement of resource optimization algorithms (such as multi-factor scheduling algorithms), enabling the algorithms to adjust according to actual resource usage patterns and improve scheduling efficiency and accuracy.

[0068] The orchestration and scheduling layer obtains real-time updated resource usage status information through interaction with the resource twin library. This information is a key input for generating resource scheduling schemes, ensuring the real-time nature and accuracy of the scheduling strategy based on the actual resource status. The data in the resource twin library also supports the operation of multi-factor scheduling algorithms, helping the algorithms to find the optimal resource allocation scheme while meeting user SLAs, comprehensively considering power consumption, cost, and computational efficiency, thereby achieving efficient resource utilization and cost-effectiveness optimization.

[0069] This embodiment enables the computing network to better achieve intelligent resource scheduling by establishing and maintaining a resource twin library, ensuring efficient resource utilization, while reducing power consumption costs and improving computing efficiency.

[0070] Optionally, the system also includes an application layer, which is connected to the orchestration and scheduling layer for receiving resource scheduling requests initiated by users and sending the resource demand indicators in the resource scheduling requests to the orchestration and scheduling layer.

[0071] In some embodiments, the application layer directly faces users or business applications, receiving resource scheduling requests initiated by them. These requests may originate from different types of business scenarios, such as cloud rendering, cloud gaming, virtual reality / augmented reality applications, big data processing, etc., each with specific requirements for computing power, storage, and network resources. The requests include resource requirement metrics, which describe in detail the user's requirements for specific resource characteristics, such as CPU / GPU performance levels, maximum network bandwidth and latency, storage capacity, etc. The application layer needs to parse these metrics to ensure an accurate understanding of the user's needs.

[0072] The parsed resource demand metrics are encapsulated into formatted information by the application layer and sent to the orchestration and scheduling layer via a communication connection. The communication connection between the application layer and the orchestration and scheduling layer should also possess high reliability and low latency characteristics to support real-time or near real-time resource scheduling processes, especially for latency-sensitive application scenarios such as online games and real-time video streaming. Upon receiving the resource demand metrics, the orchestration and scheduling module of the orchestration and scheduling layer generates a resource scheduling scheme based on the resource usage status information in the resource twin library using a preset multi-factor scheduling algorithm. The orchestration and scheduling layer sends the generated resource scheduling scheme back to the application layer, or directly executes resource scheduling and feeds back the scheduling results to the application layer. The application layer is responsible for presenting the scheduling results to the user or application, ensuring that resource scheduling requests are responded to and satisfied.

[0073] The application layer in this embodiment provides an intuitive user interface, enabling users to easily submit resource scheduling requests and monitor scheduling progress and results. It parses resource requirement indicators from user requests to ensure these requirements are accurately understood by the orchestration layer. Acting as a bridge between users and the internal scheduling mechanism of the computing network, the application layer is responsible for transmitting resource requirement indicators to the orchestration layer and simultaneously feeding back scheduling results to users or applications. The application layer efficiently processes user resource scheduling requests, collaborates with the orchestration layer to generate and execute resource scheduling schemes, provides users with customized computing network resource services, and optimizes resource utilization efficiency and cost-effectiveness.

[0074] It's important to note that traditional network resource management suffers from a severe disconnect between energy consumption and physical characteristics. "Physical characteristics" refer to the inherent attributes of the underlying hardware (such as architecture and energy efficiency potential) and its real-time operating status (such as CPU / GPU core frequency and temperature, memory access patterns, and network port speed and power consumption). This disconnect is particularly pronounced under high loads: scheduling decisions focus only on macro-level resource utilization (such as CPU %), ignoring the direct impact of micro-level differences in hardware status on energy consumption (e.g., high-frequency operation under low utilization or localized overheating leading to a sharp drop in energy efficiency). For example, a CPU might have the highest energy efficiency at 30% load, but the scheduling algorithm might allow multiple nodes to idle at 10% load, resulting in higher overall power consumption.

[0075] The "hardware deep linkage" system senses the physical characteristics of hardware in real time and uses energy efficiency as a core factor when scheduling computing, storage, and network resources. Based on this, the scheduling engine generates hardware control instructions that integrate target performance (such as setting specific core frequencies, adjusting network port speeds, and optimizing memory status), and executes them precisely through an abstraction layer. This physical characteristic perception and real-time closed-loop control of hardware, combined with the coordinated optimization of global load and temperature (such as avoiding hotspots), deeply binds and finely matches the scheduling of computing power, storage, and network resources under high load with the energy consumption status of their corresponding hardware, maximizing the overall energy efficiency ratio.

[0076] According to another embodiment of this application, a network resource scheduling method applied to the above-described network resource scheduling system is also provided. Figure 4 This is a flowchart of a network resource scheduling method provided according to an embodiment of this application, such as... Figure 4 As shown, the method includes the following steps:

[0077] Step S401: Obtain the user's resource demand indicators and determine the resource usage status of heterogeneous network resources.

[0078] Specifically, the application layer communicates directly with users or business applications, receiving resource scheduling requests initiated by users. These requests contain specific resource requirement metrics, such as CPU / GPU utilization, network bandwidth, latency requirements, and storage capacity. The application layer parses these requirement metrics from the resource scheduling request to ensure an accurate understanding of user needs, such as the level of computing, network, and storage requirements. The parsed resource requirement metrics are encapsulated and forwarded to the orchestration and scheduling layer, triggering the resource scheduling decision-making process. The network perception layer monitors the usage status of heterogeneous network resources in real time, including but not limited to CPU utilization, GPU load, network link bandwidth usage, latency and packet loss rate, and storage usage status. The network perception layer updates the monitored resource usage status information to a resource twin database in real time. This resource twin database is a real-time mapping and historical record of resource usage status, supporting the orchestration and scheduling layer in making decisions based on real-time data.

[0079] Step S402: Input the resource demand index and resource usage status into the preset multi-factor scheduling algorithm to obtain the resource scheduling scheme.

[0080] Specifically, the algorithm receives user resource demand indicators from the application layer and real-time usage status of heterogeneous network resources provided by the network perception layer. Demand indicators include specific resource requirements such as CPU, GPU, network bandwidth, and latency; usage status covers real-time data such as current resource load, availability, and power consumption. Before the algorithm runs, the input data needs to be preprocessed to ensure data format consistency and validity. This includes data cleaning, format conversion, and anomaly detection to ensure the algorithm can perform calculations based on accurate data. The pre-defined multi-factor scheduling algorithm first constructs a hard constraint filtering layer, which performs preliminary resource screening based on user SLA requirements for CPU / GPU latency, computing power, and availability, eliminating all computing power nodes that do not meet the SLA indicators. Subsequently, the algorithm constructs a multi-objective optimization layer, establishing a mathematical model with power consumption cost and computational efficiency as the core optimization objectives. This model uses power consumption cost and computational efficiency as key factors, finding a set of non-dominated solutions by solving the Pareto front, i.e., the set of optimal trade-offs that can be achieved under multi-objective constraints.

[0081] The algorithm further considers environmentally sensitive factors, such as real-time electricity price fluctuations, peak and valley characteristics of business load, and carbon emission intensity. Through a dynamic weight adjustment mechanism, it assigns real-time weights to optimization items such as power consumption cost and computational efficiency to adapt to dynamic changes in the external environment. Under the influence of dynamic weights, the algorithm can adaptively recalculate the weighted scores of each optimization item, ensuring that a resource scheduling scheme that meets the optimal conditions of the environment at any given time is generated. Based on the resource set after hard constraint screening and the multi-objective optimization function after dynamic weight adjustment, a pre-set multi-factor scheduling algorithm calculates the resource scheduling scheme. This scheme details how resources are adjusted from their current state to a state that meets user needs, including resource allocation priorities, scheduling time windows, and specific resource usage patterns. The algorithm can make real-time decisions on a minute-level timescale, ensuring that the generation and execution of the resource scheduling scheme can quickly respond to changes in user needs and fluctuations in the network environment.

[0082] Step S403: Based on the resource scheduling scheme, schedule heterogeneous network resources to the user's client.

[0083] Specifically, the orchestration and scheduling layer translates resource scheduling schemes into concrete scheduling instructions. These instructions include resource allocation priorities, resource types (such as CPU, GPU, network bandwidth, etc.), resource quantities, and scheduling time windows. The orchestration and scheduling layer transmits these instructions to the resource management layer through a communication interface. The resource management layer is responsible for the actual allocation and configuration of underlying physical resources, ensuring the execution of the resource scheduling scheme. Based on the scheduling instructions, the resource management layer allocates computing resources in the central cloud and edge cloud. This includes starting or stopping specific computing nodes, adjusting the allocation ratio of computing resources, and optimizing GPU load balancing to meet user demand metrics. Simultaneously, the resource management layer also configures network resources, including adjusting network bandwidth, optimizing network paths, and setting network service quality parameters, to ensure efficient utilization of network resources and low-latency transmission of user services.

[0084] The computing network resource scheduling method provided in this application obtains user resource demand indicators and determines the resource usage status of heterogeneous network resources; inputs the resource demand indicators and resource usage status into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and schedules heterogeneous network resources to user clients based on the resource scheduling scheme, thus solving the problem of low computing network resource scheduling efficiency in related technologies. It adopts a three-layer system of "resource management layer - computing network perception layer - orchestration and scheduling layer." The resource management layer can effectively manage heterogeneous network resources, including computing resources and network configuration resources, improving resource integration and utilization, and reducing resource fragmentation and idleness. The computing network perception layer, through communication with the resource management layer, can transform heterogeneous network resources into unified resource indicators according to preset measurement standards. The orchestration and scheduling layer, through a preset multi-factor scheduling algorithm, automatically generates a resource scheduling scheme based on real-time monitored resource usage status and user specific resource demand indicators, enabling it to make optimal resource allocation decisions in complex and ever-changing business environments, thereby improving the efficiency of computing network resource scheduling.

[0085] Optionally, in the network resource scheduling method provided in this application embodiment, inputting resource demand indicators and resource usage status into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme includes: constructing constraints based on resource demand indicators; constructing a first objective function between scheduling resource indicators and computational efficiency; constructing a second objective function between scheduling resource indicators and power consumption cost; combining the constraints, the first objective function, and the second objective function into a planning model; solving the planning model using a Pareto front algorithm to obtain a target solution, wherein the target solution is a target resource indicator scheduling amount that meets the constraints, maximizes computational efficiency, and minimizes power consumption cost; and generating a resource scheduling scheme based on the target resource indicator scheduling amount.

[0086] In some embodiments, firstly, constraints are constructed based on the resource requirement metrics submitted by the user. These metrics may include, but are not limited to, CPU / GPU utilization, network bandwidth, latency, and storage capacity. They constitute the basic requirements for computing network services and are hard metrics that the resource scheduling scheme must meet. The constraints also include specific requirements in the user's SLA agreement to ensure that resource scheduling meets performance metrics while also guaranteeing service quality.

[0087] A first objective function is constructed to maximize the computational efficiency of resource scheduling. Computational efficiency may be affected by various factors such as resource type (e.g., CPU, GPU), resource allocation method (e.g., balanced allocation, priority allocation), and task parallelism. A second objective function is constructed to minimize the power consumption cost of scheduling resources. Power consumption cost includes not only direct power consumption but may also involve indirect costs such as resource cooling and maintenance. The constraints, the first objective function, and the second objective function are combined into a multi-objective programming model. This model attempts to find a resource scheduling scheme that simultaneously maximizes computational efficiency and minimizes power consumption cost while satisfying all constraints. The Pareto front algorithm is used to solve the programming model. The Pareto front algorithm is an effective method for handling multi-objective optimization problems; it can find a set of solutions that achieve an optimal balance between the objective functions, i.e., the Pareto optimal solution set.

[0088] By employing the Pareto front algorithm, a set of objective solutions that meet the constraints and achieve an optimal balance between maximizing computational efficiency and minimizing power consumption cost can be obtained. These solutions constitute a candidate set of resource scheduling schemes. The algorithm also considers environmentally sensitive factors, such as real-time electricity prices, changes in business load, and carbon emission intensity, and assigns real-time weights to the computational efficiency and power consumption cost objectives through a dynamic weight adjustment mechanism. Under the influence of dynamic weights, the algorithm adaptively selects an optimal objective solution from the Pareto optimal solution set as the resource scheduling scheme, which achieves the best trade-off between computational efficiency and power consumption cost in the current environment. Based on the selected objective solution, the orchestration and scheduling layer generates a specific resource scheduling scheme, including the priority of resource allocation, resource type (such as CPU, GPU), resource quantity, and scheduling time window.

[0089] This embodiment utilizes a resource scheduling scheme based on a multi-factor scheduling algorithm to dynamically adapt to changes in user needs and network environment, optimize resource allocation, improve computing efficiency, and reduce power consumption costs. This ensures the accuracy and efficiency of resource scheduling decisions and meets the dual requirements of computing networks in terms of resource utilization efficiency and service quality assurance.

[0090] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0091] This application also provides a computing network resource scheduling device. It should be noted that the computing network resource scheduling device of this application can be used to execute the computing network resource scheduling method provided in this application. The computing network resource scheduling device provided in this application is described below.

[0092] Figure 5 This is a schematic diagram of a network resource scheduling device provided according to an embodiment of this application. For example... Figure 5 As shown, the device includes: an acquisition unit 501, used to acquire the user's resource demand indicators and determine the resource usage status of heterogeneous network resources; an input unit 502, used to input the resource demand indicators and resource usage status into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and a scheduling unit 503, used to schedule heterogeneous network resources to the user's client based on the resource scheduling scheme.

[0093] The computing network resource scheduling device provided in this application embodiment acquires user resource demand indicators and determines the resource usage status of heterogeneous network resources through an acquisition unit 501; an input unit 502 inputs the resource demand indicators and resource usage status into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and a scheduling unit 503 schedules heterogeneous network resources to user clients based on the resource scheduling scheme. This solves the problem of low computing network resource scheduling efficiency in related technologies. It adopts a three-layer system of "resource management layer - computing network perception layer - orchestration and scheduling layer." The resource management layer can effectively manage heterogeneous network resources, including computing resources and network configuration resources, improving resource integration and utilization, and reducing resource fragmentation and idleness. The computing network perception layer, through communication with the resource management layer, can transform heterogeneous network resources into unified resource indicators according to preset measurement standards. The orchestration and scheduling layer, through a preset multi-factor scheduling algorithm, automatically generates a resource scheduling scheme based on real-time monitored resource usage status and user specific resource demand indicators. This enables it to make optimal resource allocation decisions in complex and ever-changing business environments, thereby improving the efficiency of computing network resource scheduling.

[0094] Optionally, in the computing network resource scheduling device provided in this application embodiment, the input unit 502 includes: a construction module, used to construct constraints based on resource demand indicators, construct a first objective function between scheduling resource indicators and computing efficiency, and construct a second objective function between scheduling resource indicators and power consumption cost; a combination module, used to combine the constraints, the first objective function, and the second objective function into a planning model; a solution module, used to solve the planning model using the Pareto front algorithm to obtain the objective solution, wherein the objective solution is the target resource indicator scheduling amount that meets the constraints, maximizes computing efficiency, and minimizes power consumption cost; and a generation module, used to generate a resource scheduling scheme based on the target resource indicator scheduling amount.

[0095] The network resource scheduling device includes a processor and a memory. The aforementioned acquisition unit 501, input unit 502, and scheduling unit 503 are all stored in the memory as program units. The processor executes the aforementioned program units stored in the memory to realize the corresponding functions.

[0096] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and adjusting kernel parameters can improve the efficiency of network resource scheduling.

[0097] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0098] This invention provides a computer-readable storage medium storing a program that, when executed by a processor, implements a computer network resource scheduling method.

[0099] This invention provides a processor for running a program, wherein the program executes a network resource scheduling method during runtime.

[0100] Figure 6 This is a schematic diagram of an electronic device provided according to an embodiment of this application. For example... Figure 6 As shown, electronic device 601 includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: obtaining the user's resource demand indicators and determining the resource usage status of heterogeneous network resources; inputting the resource demand indicators and resource usage status into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and scheduling heterogeneous network resources to the user's client based on the resource scheduling scheme. The device in this paper can be a server, PC, PAD, mobile phone, etc.

[0101] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program with the following method steps: obtaining the user's resource demand indicators and determining the resource usage status of heterogeneous network resources; inputting the resource demand indicators and resource usage status into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; and scheduling heterogeneous network resources to the user's client based on the resource scheduling scheme.

[0102] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0104] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0106] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0107] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0108] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0109] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0110] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A computing network resource scheduling system, characterized in that, include: A resource management layer for managing heterogeneous network resources, wherein the heterogeneous network resources include at least one of the following: computing resources and network configuration resources; The network perception layer is communicatively connected to the resource management layer. It is used to convert the heterogeneous network resources into resource indicators according to a preset measurement standard, and to monitor the resource usage status of the heterogeneous network resources in the resource management layer in real time. The resource indicators are indicators that conform to the data format of the orchestration and scheduling layer for scheduling resources. The orchestration and scheduling layer is communicatively connected to the computing network perception layer. It is used to generate a resource scheduling scheme based on a preset multi-factor scheduling algorithm, the resource usage status, and the user's resource demand indicators. The preset multi-factor scheduling algorithm is used to randomly generate candidate scheduling schemes based on the heterogeneous network resources, and select the resource scheduling scheme that meets the resource demand indicators, maximizes the computational efficiency in the resource usage status, and minimizes the power consumption cost in the resource usage status.

2. The computing network resource scheduling system according to claim 1, characterized in that, The resource management layer includes: The resource configuration module is used to configure the heterogeneous network resources through virtualization and pooling operations; The resource monitoring module is used to monitor the resource usage status of the heterogeneous network resources and send the resource usage status to the computing network perception layer. The resource scheduling module is used to schedule the heterogeneous network resources according to the resource scheduling scheme; The resource recycling module is used to recycle the heterogeneous network resources when the network is detected to be in an idle state.

3. The computing network resource scheduling system according to claim 1, characterized in that, The network perception layer includes: The abstract modeling module is used to input the heterogeneous resources of the network into the quantum metric model to obtain resource indicators, wherein the resource indicators are resource indicators divided according to a preset metric standard. The encapsulation module encapsulates the heterogeneous network resources into target interfaces according to the resource indicators; The perception module is used to collect the resource usage status sent by the resource management layer, and to preprocess the resource usage status to obtain the resource usage status that conforms to the data processing format of the orchestration and scheduling layer. A configuration module is used to adjust the configuration information and allocation status of the heterogeneous network resources according to the resource usage status; The access module is used to provide the target interface to the orchestration management layer.

4. The computing network resource scheduling system according to claim 1, characterized in that, The orchestration and scheduling layer includes: The service-oriented scheduling module is used to provide resource scheduling services to the application layer; The resource visualization module is used to display the resource usage status of the heterogeneous network resources; The resource metering module is used to statistically analyze the consumption and performance indicators of the heterogeneous network resources in real time, wherein the performance indicators include at least one of the following: processing speed and latency; The monitoring and analysis module is used to monitor the degree to which user resource needs are met and resource service quality indicators. The orchestration and scheduling module is used to plan the heterogeneous network resources through the preset multi-factor scheduling algorithm, the resource usage status, and the resource demand indicators, and generate the resource scheduling scheme.

5. The computing network resource scheduling system according to claim 1, characterized in that, The orchestration and scheduling layer is further configured to issue resource scheduling instructions to the computing network perception layer based on the resource scheduling scheme. The computing network perception layer is further configured to adjust the configuration information and allocation status of the heterogeneous network resources based on the resource scheduling instructions, and send the configuration information and allocation status to the resource management layer. The resource management layer is further configured to schedule the heterogeneous network resources based on the configuration information and allocation status.

6. The computing network resource scheduling system according to claim 1, characterized in that, The preset multi-factor scheduling algorithm randomly generates multiple candidate resource scheduling schemes based on the heterogeneous network resources, eliminates candidate scheduling schemes that do not meet the resource demand indicators, and selects the resource scheduling scheme from the candidate scheduling schemes that meet the resource demand indicators according to a preset objective function, wherein the preset objective function aims to minimize power consumption cost and maximize computational efficiency.

7. The computing network resource scheduling system according to claim 1, characterized in that, The system further includes a resource twin library for mapping the resource usage status of the heterogeneous network resources. The resource twin library is updated in real time based on the resource usage status monitored by the computing network perception layer. The orchestration and scheduling layer obtains the resource usage status through the resource twin library.

8. The computing network resource scheduling system according to claim 1, characterized in that, The system further includes an application layer, which is communicatively connected to the orchestration and scheduling layer, for receiving resource scheduling requests initiated by users and sending the resource demand indicators in the resource scheduling requests to the orchestration and scheduling layer.

9. A method for scheduling computing network resources, applied to the computing network resource scheduling system according to any one of claims 1-8, characterized in that, include: Obtain user resource demand metrics and determine the resource utilization status of heterogeneous network resources; The resource demand indicators and the resource usage status are input into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme; Based on the resource scheduling scheme, the heterogeneous network resources are scheduled to the user's client.

10. The network resource scheduling method according to claim 9, characterized in that, The resource demand index and the resource usage status are input into a preset multi-factor scheduling algorithm to obtain a resource scheduling scheme, including: Based on the resource demand indicators, constraints are constructed, a first objective function is constructed between the scheduling resource indicators and computational efficiency, and a second objective function is constructed between the scheduling resource indicators and power consumption cost. The planning model is formed by combining the constraints, the first objective function, and the second objective function; The planning model is solved by the Pareto front algorithm to obtain the objective solution, wherein the objective solution is the target resource scheduling amount that meets the constraints, maximizes the computational efficiency and minimizes the power consumption cost. The resource scheduling scheme is generated based on the target resource index scheduling amount.

Images