A method and system for scheduling boundary computing power routing based on OAM information publishing

Abstract

This invention discloses a boundary computing power routing scheduling method and system based on OAM information publishing. The method includes: acquiring a computing request to be processed and node awareness data sent by multiple computing power routes; the computing power routes are communicatively connected to multiple computing power nodes and determine the node awareness data based on communication interaction; determining the scheduling route corresponding to the computing request from the multiple computing power routes based on the node awareness data; determining the node scheduling strategy corresponding to the scheduling route based on the computing request and the node awareness data corresponding to the scheduling route; generating an OAM information message based on the computing request and the node scheduling strategy, and sending the OAM information message to the corresponding computing power node through the scheduling route. Therefore, this invention can achieve accurate task scheduling based on real-time awareness data, thereby effectively improving the processing efficiency and response speed of computing requests among multiple computing power nodes.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a boundary computing power routing and scheduling method and system based on OAM information publication. Background Technology

[0002] With the continuous evolution of computing networks (CFN) and computing routing technologies, how to dynamically schedule tasks based on network status and computing node load has become a core issue in improving the performance of distributed computing systems. However, existing computing scheduling technologies are typically based on static configuration or periodic resource reporting mechanisms. They acquire computing requests and distribute tasks along preset node paths. Due to the lack of technical support for real-time acquisition of node-aware data from computing requests and computing routing, and for determining scheduling routes and node scheduling strategies accordingly, it is impossible to generate and send OAM (Operation, Administration, and Maintenance) information messages containing the strategy to computing nodes for execution. This makes it difficult to achieve accurate task scheduling based on real-time awareness data, leading to insufficient processing efficiency and response speed of computing requests across multiple computing nodes. This limits the resource utilization and system stability of computing networks in complex and dynamic business scenarios. Therefore, existing technologies have shortcomings that urgently need to be addressed. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a boundary computing power routing and scheduling method and system based on OAM information publication, which can realize accurate task scheduling based on real-time sensing data, thereby effectively improving the processing efficiency and response speed of computing requests among multiple computing power nodes.

[0004] To address the aforementioned technical problems, the first aspect of this invention discloses a boundary computing power routing and scheduling method based on OAM information publishing, the method comprising: The system acquires computing requests to be processed and node-aware data sent by multiple computing power routes; the computing power routes communicate with multiple computing power nodes and determine the node-aware data based on communication interactions. Based on the node perception data, determine the scheduling route corresponding to the computing request from the plurality of computing power routes; Based on the computation request and the node awareness data corresponding to the scheduling route, determine the node scheduling strategy corresponding to the scheduling route; Based on the computation request and the node scheduling policy, an OAM information message is generated, and the OAM information message is sent to the corresponding computing power node through the scheduling route; the computing power node receives the OAM information message and executes the computation request according to the node scheduling policy.

[0005] As an optional implementation, in the first aspect of the present invention, the node sensing data includes node data corresponding to each of the computing power nodes and communication data between at least two of the computing power nodes; the node data includes at least one of node hardware data, node software data, node computing power, node network link status, and node sensing data.

[0006] As an optional implementation, in the first aspect of the present invention, determining the scheduling route corresponding to the computing request from the plurality of computing power routes based on the node-aware data includes: For each computing power route, the computing scheduling capability corresponding to the computing power route is determined based on the node perception data corresponding to the computing power route and the prediction model. Based on the computing scheduling capability, the scheduling route corresponding to the computing request is determined from the plurality of computing power routes.

[0007] As an optional implementation, in the first aspect of the present invention, determining the computing scheduling capability corresponding to the computing power route based on the node-aware data corresponding to the computing power route and using a prediction model includes: The node perception data corresponding to the computing power route is input into the trained overall computing power prediction model to obtain the overall predicted computing power corresponding to the computing power route; The node perception data corresponding to the computing power route is input into the trained average computing power prediction model to obtain the average predicted computing power corresponding to the computing power route. Calculate the logarithm of the difference between the overall prediction computing power and the average prediction computing power to obtain the average computing power weight; Calculate the product of the average predicted computing power and the average computing power weight, and the sum of the product and the overall predicted computing power to obtain the computing scheduling capability corresponding to the computing power route.

[0008] As an optional implementation, in the first aspect of the present invention, determining the scheduling route corresponding to the computing request from the plurality of computing power routes based on the computing scheduling capability includes: From all the computing power routes, determine a preset number of computing power routes with the highest computing scheduling capacity to obtain multiple candidate scheduling routes; Calculate the average of the differences between the computational scheduling capabilities of every two candidate scheduling routes to obtain the routing differentiation parameters; Determine whether the routing differentiation parameter is greater than a preset first parameter threshold; If so, the candidate scheduling route with the highest computing scheduling capability is determined as the scheduling route corresponding to the computing request; If not, calculate the product of the computational scheduling capability and the computational weight corresponding to each candidate scheduling route to obtain the corrected scheduling capability corresponding to each candidate scheduling route; the computational weight is proportional to the amount of computational requests scheduled by the candidate scheduling route in the historical time period; The candidate scheduling route with the highest corrected scheduling capability is determined as the scheduling route corresponding to the calculation request.

[0009] As an optional implementation, in the first aspect of the present invention, determining the node scheduling strategy corresponding to the scheduling route based on the computation request and the node-aware data corresponding to the scheduling route includes: For each computing node corresponding to the scheduling route, the node capability parameters corresponding to the computing node are determined based on the node data and the communication data corresponding to the computing node. From all computing power nodes corresponding to the scheduling route, nodes whose node capability parameters are greater than a preset second parameter threshold are selected to obtain multiple allocation nodes; The objective function is set to minimize the computation time corresponding to the scheduling strategy; the computation time is predicted by inputting the node parameters and the computation task of each allocated node corresponding to the scheduling strategy into a trained computation time prediction model. The scheduling strategy for all the allocated nodes relative to the computation request is iteratively calculated based on the dynamic optimization algorithm until the objective function is satisfied, so as to obtain the node scheduling strategy corresponding to the scheduling route.

[0010] As an optional implementation, in the first aspect of the present invention, determining the node capability parameters corresponding to the computing power node based on the node data and the communication data corresponding to the computing power node includes: The node data corresponding to the computing power node is input into the trained computing power performance prediction model to obtain the output predicted computing power performance. The communication data between the computing power node and all other computing power nodes is input into the trained node collaboration prediction model to obtain the output node collaboration performance. The weighted sum of the predicted computing power performance and the node collaboration performance is calculated to obtain the node capability parameters corresponding to the computing power node.

[0011] As an optional implementation, in the first aspect of the present invention, the constraints of the dynamic programming algorithm include: In the scheduling strategy, the amount of allocated tasks corresponding to each allocation node is proportional to the node capability parameter corresponding to that allocation node. In the scheduling strategy, the difference between the amount of allocated tasks corresponding to any two allocation nodes is less than a preset difference threshold. In the scheduling strategy, the scheduling time corresponding to each allocation node is inversely proportional to the node capability parameter corresponding to that allocation node; the scheduling time is predicted by inputting the scheduling route and the allocation task of the allocation node into a trained data transmission time prediction model.

[0012] A second aspect of this invention discloses a boundary computing power routing and scheduling system based on OAM information publishing, the system comprising: The acquisition module is used to acquire computing requests to be processed and node-aware data sent by multiple computing power routes; the computing power routes are communicatively connected to multiple computing power nodes and determine the node-aware data based on communication interactions; The first determining module is used to determine the scheduling route corresponding to the computing request from the plurality of computing power routes based on the node perception data; The second determining module is used to determine the node scheduling strategy corresponding to the scheduling route based on the calculation request and the node awareness data corresponding to the scheduling route. The scheduling module is used to generate an OAM information message based on the computing request and the node scheduling policy, and send the OAM information message to the corresponding computing power node through the scheduling route; the computing power node receives the OAM information message and executes the computing request according to the node scheduling policy.

[0013] As an optional implementation, in a second aspect of the present invention, the node sensing data includes node data corresponding to each of the computing power nodes and communication data between at least two of the computing power nodes; the node data includes at least one of node hardware data, node software data, node computing power, node network link status, and node sensing data.

[0014] As an optional implementation, in a second aspect of the invention, the first determining module determines the specific method by which it determines the scheduling route corresponding to the computing request from the plurality of computing power routes based on the node-aware data, including: For each computing power route, the computing scheduling capability corresponding to the computing power route is determined based on the node perception data corresponding to the computing power route and the prediction model. Based on the computing scheduling capability, the scheduling route corresponding to the computing request is determined from the plurality of computing power routes.

[0015] As an optional implementation, in a second aspect of the invention, the first determining module determines the specific method of the computing scheduling capability corresponding to the computing power route based on the node-aware data corresponding to the computing power route and a prediction model, including: The node perception data corresponding to the computing power route is input into the trained overall computing power prediction model to obtain the overall predicted computing power corresponding to the computing power route; The node perception data corresponding to the computing power route is input into the trained average computing power prediction model to obtain the average predicted computing power corresponding to the computing power route. Calculate the logarithm of the difference between the overall prediction computing power and the average prediction computing power to obtain the average computing power weight; Calculate the product of the average predicted computing power and the average computing power weight, and the sum of the product and the overall predicted computing power to obtain the computing scheduling capability corresponding to the computing power route.

[0016] As an optional implementation, in a second aspect of the invention, the specific method by which the first determining module determines the scheduling route corresponding to the computing request from the plurality of computing power routes based on the computing scheduling capability includes: From all the computing power routes, determine a preset number of computing power routes with the highest computing scheduling capacity to obtain multiple candidate scheduling routes; Calculate the average of the differences between the computational scheduling capabilities of every two candidate scheduling routes to obtain the routing differentiation parameters; Determine whether the routing differentiation parameter is greater than a preset first parameter threshold; If so, the candidate scheduling route with the highest computing scheduling capability is determined as the scheduling route corresponding to the computing request; If not, calculate the product of the computational scheduling capability and the computational weight corresponding to each candidate scheduling route to obtain the corrected scheduling capability corresponding to each candidate scheduling route; the computational weight is proportional to the amount of computational requests scheduled by the candidate scheduling route in the historical time period; The candidate scheduling route with the highest corrected scheduling capability is determined as the scheduling route corresponding to the calculation request.

[0017] As an optional implementation, in a second aspect of the invention, the second determining module determines the specific method of the node scheduling strategy corresponding to the scheduling route based on the calculation request and the node-aware data corresponding to the scheduling route, including: For each computing node corresponding to the scheduling route, the node capability parameters corresponding to the computing node are determined based on the node data and the communication data corresponding to the computing node. From all computing power nodes corresponding to the scheduling route, nodes whose node capability parameters are greater than a preset second parameter threshold are selected to obtain multiple allocation nodes; The objective function is set to minimize the computation time corresponding to the scheduling strategy; the computation time is predicted by inputting the node parameters and the computation task of each allocated node corresponding to the scheduling strategy into a trained computation time prediction model. The scheduling strategy for all the allocated nodes relative to the computation request is iteratively calculated based on the dynamic optimization algorithm until the objective function is satisfied, so as to obtain the node scheduling strategy corresponding to the scheduling route.

[0018] As an optional implementation, in a second aspect of the present invention, the specific method by which the second determining module determines the node capability parameters corresponding to the computing power node based on the node data and the communication data corresponding to the computing power node includes: The node data corresponding to the computing power node is input into the trained computing power performance prediction model to obtain the output predicted computing power performance. The communication data between the computing power node and all other computing power nodes is input into the trained node collaboration prediction model to obtain the output node collaboration performance. The weighted sum of the predicted computing power performance and the node collaboration performance is calculated to obtain the node capability parameters corresponding to the computing power node.

[0019] As an optional implementation, in a second aspect of the invention, the constraints of the dynamic programming algorithm include: In the scheduling strategy, the amount of allocated tasks corresponding to each allocation node is proportional to the node capability parameter corresponding to that allocation node. In the scheduling strategy, the difference between the amount of allocated tasks corresponding to any two allocation nodes is less than a preset difference threshold. In the scheduling strategy, the scheduling time corresponding to each allocation node is inversely proportional to the node capability parameter corresponding to that allocation node; the scheduling time is predicted by inputting the scheduling route and the allocation task of the allocation node into a trained data transmission time prediction model.

[0020] A third aspect of this invention discloses another boundary computing power routing and scheduling system based on OAM information publishing, the system comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in the boundary computing power routing scheduling method based on OAM information publication disclosed in the first aspect of the present invention.

[0021] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the boundary computing power routing scheduling method based on OAM information publication disclosed in the first aspect of the present invention.

[0022] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: This invention acquires node-aware data of computing requests and computing power routing, determines scheduling routes and node scheduling strategies accordingly, and then generates and sends OAM information messages containing the strategies to computing power nodes for execution. This enables precise task scheduling based on real-time awareness data, thereby effectively improving the processing efficiency and response speed of computing requests among multiple computing power nodes. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a boundary computing power routing and scheduling method based on OAM information publishing disclosed in an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the structure of a boundary computing power routing and scheduling system based on OAM information publishing disclosed in an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of another boundary computing power routing and scheduling system based on OAM information publishing disclosed in an embodiment of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0029] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0030] This invention discloses a boundary computing power routing and scheduling method and system based on OAM information dissemination. By acquiring node-aware data of computing requests and computing power routing transmissions, and determining scheduling routes and node scheduling strategies accordingly, it generates and sends OAM information messages containing the strategy to the computing power nodes for execution. This enables precise task scheduling based on real-time awareness data, effectively improving the processing efficiency and response speed of computing requests across multiple computing power nodes. Detailed descriptions follow.

[0031] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a boundary computing power routing and scheduling method based on OAM information publishing disclosed in an embodiment of the present invention. Figure 1 The described boundary computing power routing and scheduling method based on OAM information publishing can be applied to data processing systems / data processing equipment / data processing servers (wherein, the server includes local processing servers or cloud processing servers). For example... Figure 1 As shown, the boundary computing power routing and scheduling method based on OAM information publication can include the following operations: 101. Obtain the computing requests to be processed and the node-aware data sent by multiple computing power routes.

[0032] Optionally, the computing power routing communication connects to multiple computing power nodes and determines node-aware data based on communication interactions.

[0033] Optionally, the computation request can be an artificial intelligence model training request, a large-scale graphics rendering request, a real-time video encoding and decoding request, a big data analysis request, a blockchain encrypted computation request, or an autonomous driving path planning request; this invention does not limit the scope of the request.

[0034] Optionally, the computing power routing can be a Layer 3 switch with computing power awareness, a computing power gateway device, a software-defined networking (SDN) controller, an edge computing power relay node, or a virtualized computing power scheduling center; this invention does not limit the scope of the invention.

[0035] Optionally, the node's sensing data can be real-time status data obtained through BGP extension (BGP-LS), network management protocol (SNMP), segmented routing (SRv6), or in-band telemetry (INT) technology; this invention does not impose any limitations.

[0036] Optionally, the node-aware data includes node data corresponding to each computing power node and communication data between at least two computing power nodes.

[0037] Optionally, node data may include at least one of node hardware data, node software data, node computing power, node network link status, and node sensor data.

[0038] Optionally, the node hardware data can be the number of CPU cores, GPU floating memory computing power, NPU tensor cores, FPGA logic resources, memory frequency, or NVMe storage IOPS parameters; this invention does not limit these parameters.

[0039] Optionally, the node software data can be the operating system kernel version, container engine (Docker / Containerd) status, CUDA version number, pre-installed algorithm library (MKL / OpenBLAS) version, or the number of running task processes; this invention does not impose any limitations on this.

[0040] Optionally, the network link status of the node can be port throughput, link error rate, network card physical bandwidth utilization, routing table entry capacity, or current concurrent TCP connection count; this invention does not impose any limitations.

[0041] Optionally, the node sensing data can be real-time data such as processor core temperature, motherboard power supply voltage, computer room ambient humidity, fan speed, or total power consumption (TDP), and this invention does not limit it.

[0042] Optionally, the communication data may be round-trip time (RTT) between computing nodes, link jitter, packet loss rate, physical spacing, or signal-to-noise ratio of the fiber optic link; this invention does not limit the data.

[0043] 102. Based on node perception data, determine the scheduling route corresponding to the computing request from multiple computing power routes.

[0044] 103. Based on the node awareness data corresponding to the computation request and the scheduling route, determine the node scheduling strategy corresponding to the scheduling route.

[0045] 104. Generate OAM information messages based on computing requests and node scheduling policies, and send the OAM information messages to the corresponding computing power nodes through scheduling routing.

[0046] Optionally, the computing node receives OAM information messages and executes computing requests according to the node scheduling policy.

[0047] Optionally, the OAM information message can be an Ethernet OAM (802.3ah) based message, an MPLS based OAM message, an SRv6 based path tracing message, a bidirectional forwarding detection (BFD) message, or a custom computing power awareness description message; the present invention does not impose any limitations.

[0048] As can be seen, the above-described embodiments of the invention obtain node-aware data of computing requests and computing power routing, determine scheduling routes and node scheduling strategies accordingly, and then generate and send OAM information messages containing the strategies to computing power nodes for execution. This enables precise task scheduling based on real-time awareness data, thereby effectively improving the processing efficiency and response speed of computing requests among multiple computing power nodes.

[0049] As an optional embodiment, the step above, determining the scheduling route corresponding to the computing request from multiple computing power routes based on node-aware data, includes: For each computing power route, the computing scheduling capability corresponding to the computing power route is determined based on the node perception data corresponding to the computing power route and the prediction model. Based on computing scheduling capabilities, the scheduling route corresponding to the computing request is determined from multiple computing power routes.

[0050] Optionally, the prediction model can be a multilayer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), extreme gradient boosting (XGBoost) or support vector regression (SVR), and this invention does not limit it.

[0051] As can be seen, by using the above optional embodiments to analyze the node-aware data of each computing power route using a predictive model to determine its computing scheduling capability and select the target scheduling route accordingly, it is possible to achieve a quantitative assessment and scientific decision-making of the scheduling potential of the routing nodes, thereby significantly enhancing the predictability and reliability of computing resource allocation.

[0052] As an optional embodiment, the step above, determining the computing scheduling capability corresponding to the computing power route based on the node-aware data corresponding to the computing power route and using a prediction model, includes: The node perception data corresponding to the computing power route is input into the trained overall computing power prediction model to obtain the overall predicted computing power corresponding to the computing power route; The node perception data corresponding to the computing power route is input into the trained average computing power prediction model to obtain the average predicted computing power corresponding to the computing power route; Calculate the logarithm of the difference between the overall prediction computing power and the average prediction computing power to obtain the average computing power weight; Calculate the product of the average predicted computing power and the average computing power weight, and the sum of the product and the overall predicted computing power to obtain the computing scheduling capability corresponding to the computing power route.

[0053] Optionally, the overall computing power prediction model or the average computing power prediction model can be a spatial topology computing power prediction model based on graph neural networks (GNN), a feature fusion model based on Transformer, or a residual prediction model based on ResNet. This invention does not limit the model.

[0054] As can be seen, by inputting node-aware data into the overall and average computing power prediction models respectively and combining the logarithmic weight of the difference between the two to correct the prediction results and obtain the computing scheduling capability, it is possible to achieve refined modeling of computing power distribution characteristics, thereby greatly improving the accuracy and robustness of computing power evaluation results.

[0055] As an optional embodiment, the step above, determining the scheduling route corresponding to the computing request from multiple computing power routes based on computing scheduling capabilities, includes: From all computing power routes, a predetermined number of computing power routes with the highest computing scheduling capacity are determined to obtain multiple candidate scheduling routes; Calculate the average of the differences in computational scheduling capabilities between every two candidate routes to obtain the route differentiation parameters; Determine whether the routing differentiation parameter is greater than the preset first parameter threshold; If so, the candidate scheduling route with the highest computational scheduling capability is determined as the scheduling route corresponding to the computation request; If not, calculate the product of the computational scheduling capability and computational weight corresponding to each candidate scheduling route to obtain the corrected scheduling capability corresponding to each candidate scheduling route. The candidate scheduling route with the highest modification scheduling capability is determined as the scheduling route corresponding to the computation request.

[0056] Optionally, the computation weight is proportional to the amount of computation requests scheduled by the candidate scheduling route in the historical time period.

[0057] Optionally, the preset percentage can be the top 10%, top 20%, top 5%, or a percentage value dynamically determined according to the scale of the computing power cluster; this invention does not impose any limitations.

[0058] As can be seen, by selecting high-capacity candidate routes and combining them with route differentiation parameters or introducing historical computation weights to determine the final route, it is possible to achieve flexible scheduling that selects the best option when differences are significant and balances load when differences are small. This can effectively avoid overloading of a single route and improve system throughput.

[0059] As an optional embodiment, the step above, determining the node scheduling strategy corresponding to the scheduling route based on the node-aware data corresponding to the computation request and the scheduling route, includes: For each computing node corresponding to the scheduling route, determine the node capability parameters corresponding to the computing node based on the node data and communication data of that computing node. Nodes whose node capability parameters are greater than the preset second parameter threshold are selected from all computing power nodes corresponding to the scheduling route to obtain multiple allocation nodes; The objective function is set to minimize the computation time corresponding to the scheduling strategy. The scheduling strategy for all allocated nodes relative to the computation request is iteratively calculated based on the dynamic optimization algorithm until the objective function is satisfied, so as to obtain the node scheduling strategy corresponding to the scheduling route.

[0060] Optionally, the computation time is predicted by inputting the node parameters and the computation task of each allocated node corresponding to the scheduling strategy into a trained computation time prediction model.

[0061] Optionally, the dynamic optimization algorithm can be a genetic algorithm, particle swarm optimization algorithm, simulated annealing algorithm, ant colony optimization algorithm, improved differential evolution algorithm, or a search algorithm based on reinforcement learning (RL), and this invention does not limit it.

[0062] As can be seen, through the above optional embodiments, by calculating the node capability parameters of computing nodes and selecting high-quality allocation nodes, taking the minimization of computing time as the objective function and using dynamic optimization algorithms to iteratively calculate the node scheduling strategy, it is possible to achieve a scientific allocation of computing tasks among nodes, thereby significantly shortening the overall completion time of large-scale computing requests.

[0063] As an optional embodiment, the step above, determining the node capability parameters corresponding to the computing power node based on the node data and communication data corresponding to the computing power node, includes: The node data corresponding to the computing power node is input into the trained computing power performance prediction model to obtain the output predicted computing power performance. The communication data between the computing node and all other computing nodes is input into the trained node collaboration prediction model to obtain the output node collaboration performance. The node capability parameters corresponding to the predicted computing power node are obtained by calculating the weighted sum of the predicted computing power performance and the node collaboration performance.

[0064] Optionally, the computing power performance prediction model or node collaboration prediction model can be a feature extraction model based on attention mechanism, a performance evolution prediction model based on long short-term memory network (LSTM) or a regression model based on random forest, and this invention does not limit it.

[0065] As can be seen, through the above optional embodiments, by inputting node data and communication data into the computing power performance and node collaboration prediction model respectively and weighting and summing the output results to determine the node capability parameters, it is possible to comprehensively evaluate nodes from two dimensions: individual computing performance and group collaboration potential. This provides an objective and comprehensive quantitative basis for efficient node selection.

[0066] As an optional embodiment, the constraints of the dynamic programming algorithm in the above steps include: In the scheduling strategy, the number of tasks assigned to each allocation node is proportional to the node's capacity parameter. In the scheduling strategy, the difference between the amount of tasks allocated to any two allocation nodes is less than a preset difference threshold. In the scheduling strategy, the scheduling time for each assigned node is inversely proportional to the node capability parameter corresponding to that assigned node.

[0067] Optionally, the scheduling time is predicted by inputting the scheduling route and the allocation task of the allocation node into a trained data transmission time prediction model.

[0068] Optionally, the data transmission time prediction model can be a queue theory-based modeling model, a link transmission delay prediction model based on graph convolutional networks (GCN), or a time series model based on deep belief networks (DBN), and this invention does not limit it.

[0069] As can be seen, by introducing constraints such as the task volume being proportional to the node's capacity, the task volume difference being limited, and the scheduling time being inversely proportional to the capacity into the dynamic programming algorithm, the scheduling scheme can ensure that it fully leverages the advantages of high-performance nodes while taking into account load balancing and transmission efficiency among nodes, thereby achieving the optimal utilization of computing resources.

[0070] Example 2 Please see Figure 2 , Figure 2This is a schematic diagram of the structure of a boundary computing power routing and scheduling system based on OAM information publishing, as disclosed in an embodiment of the present invention. Figure 2 The described boundary computing power routing and scheduling system based on OAM information publishing can be applied to data processing systems / data processing equipment / data processing servers (wherein, the server includes local processing servers or cloud processing servers). For example... Figure 2 As shown, the boundary computing power routing and scheduling system based on OAM information publishing can include: The acquisition module 201 is used to acquire computing requests to be processed and node-aware data sent by multiple computing power routes.

[0071] Optionally, the computing power routing communication connects to multiple computing power nodes and determines node-aware data based on communication interactions.

[0072] The first determining module 202 is used to determine the scheduling route corresponding to the computing request from multiple computing power routes based on node perception data.

[0073] The second determining module 203 is used to determine the node scheduling strategy corresponding to the scheduling route based on the node perception data corresponding to the computing request and the scheduling route.

[0074] The scheduling module 204 is used to generate OAM information messages based on computing requests and node scheduling policies, and send the OAM information messages to the corresponding computing power nodes through scheduling routing.

[0075] Optionally, the computing node receives OAM information messages and executes computing requests according to the node scheduling policy.

[0076] As can be seen, the above-described embodiments of the invention obtain node-aware data of computing requests and computing power routing, determine scheduling routes and node scheduling strategies accordingly, and then generate and send OAM information messages containing the strategies to computing power nodes for execution. This enables precise task scheduling based on real-time awareness data, thereby effectively improving the processing efficiency and response speed of computing requests among multiple computing power nodes.

[0077] As an optional embodiment, the node-aware data includes node data corresponding to each computing power node and communication data between at least two computing power nodes; the node data includes at least one of node hardware data, node software data, node computing power, node network link status, and node sensing data.

[0078] As can be seen, the above optional embodiments limit the details of the data content of the node-aware data to comprehensively characterize the overall node characteristics of the computing power node cluster, assist in the realization of accurate task scheduling based on real-time awareness data, and improve the processing efficiency and response speed of computing requests among multiple computing power nodes.

[0079] As an optional embodiment, the first determining module determines the specific method of the scheduling route corresponding to the computing request from multiple computing power routes based on node-aware data, including: For each computing power route, the computing scheduling capability corresponding to the computing power route is determined based on the node perception data corresponding to the computing power route and the prediction model. Based on computing scheduling capabilities, the scheduling route corresponding to the computing request is determined from multiple computing power routes.

[0080] As can be seen, by using the above optional embodiments to analyze the node-aware data of each computing power route using a predictive model to determine its computing scheduling capability and select the target scheduling route accordingly, it is possible to achieve a quantitative assessment and scientific decision-making of the scheduling potential of the routing nodes, thereby significantly enhancing the predictability and reliability of computing resource allocation.

[0081] As an optional embodiment, the first determining module determines the specific method of computing scheduling capability corresponding to the computing power route based on the node-aware data corresponding to the computing power route and a prediction model, including: The node perception data corresponding to the computing power route is input into the trained overall computing power prediction model to obtain the overall predicted computing power corresponding to the computing power route; The node perception data corresponding to the computing power route is input into the trained average computing power prediction model to obtain the average predicted computing power corresponding to the computing power route; Calculate the logarithm of the difference between the overall prediction computing power and the average prediction computing power to obtain the average computing power weight; Calculate the product of the average predicted computing power and the average computing power weight, and the sum of the product and the overall predicted computing power to obtain the computing scheduling capability corresponding to the computing power route.

[0082] As can be seen, by inputting node-aware data into the overall and average computing power prediction models respectively and combining the logarithmic weight of the difference between the two to correct the prediction results and obtain the computing scheduling capability, it is possible to achieve refined modeling of computing power distribution characteristics, thereby greatly improving the accuracy and robustness of computing power evaluation results.

[0083] As an optional embodiment, the first determining module determines the specific method by which it determines the scheduling route corresponding to the computing request from multiple computing power routes based on computing scheduling capabilities, including: From all computing power routes, a predetermined number of computing power routes with the highest computing scheduling capacity are determined to obtain multiple candidate scheduling routes; Calculate the average of the differences in computational scheduling capabilities between every two candidate routes to obtain the route differentiation parameters; Determine whether the routing differentiation parameter is greater than the preset first parameter threshold; If so, the candidate scheduling route with the highest computational scheduling capability is determined as the scheduling route corresponding to the computation request; If not, calculate the product of the computational scheduling capacity and the computational weight for each candidate scheduling route to obtain the corrected scheduling capacity for each candidate scheduling route; optionally, the computational weight is proportional to the amount of computational requests scheduled by the candidate scheduling route in the historical time period. The candidate scheduling route with the highest modification scheduling capability is determined as the scheduling route corresponding to the computation request.

[0084] As can be seen, by selecting high-capacity candidate routes and combining them with route differentiation parameters or introducing historical computation weights to determine the final route, it is possible to achieve flexible scheduling that selects the best option when differences are significant and balances load when differences are small. This can effectively avoid overloading of a single route and improve system throughput.

[0085] As an optional embodiment, the second determining module determines the specific method of the node scheduling strategy corresponding to the scheduling route based on the computation request and the node-aware data corresponding to the scheduling route, including: For each computing node corresponding to the scheduling route, determine the node capability parameters corresponding to the computing node based on the node data and communication data of that computing node. Nodes whose node capability parameters are greater than the preset second parameter threshold are selected from all computing power nodes corresponding to the scheduling route to obtain multiple allocation nodes; The objective function is set to minimize the computation time corresponding to the scheduling strategy; optionally, the computation time is predicted by inputting the node parameters and the computation task of each allocated node corresponding to the scheduling strategy into a trained computation time prediction model. The scheduling strategy for all allocated nodes relative to the computation request is iteratively calculated based on the dynamic optimization algorithm until the objective function is satisfied, so as to obtain the node scheduling strategy corresponding to the scheduling route.

[0086] As can be seen, through the above optional embodiments, by calculating the node capability parameters of computing nodes and selecting high-quality allocation nodes, taking the minimization of computing time as the objective function and using dynamic optimization algorithms to iteratively calculate the node scheduling strategy, it is possible to achieve a scientific allocation of computing tasks among nodes, thereby significantly shortening the overall completion time of large-scale computing requests.

[0087] As an optional embodiment, the second determining module determines the specific method of the node capability parameters corresponding to the computing power node based on the node data and communication data corresponding to the computing power node, including: The node data corresponding to the computing power node is input into the trained computing power performance prediction model to obtain the output predicted computing power performance. The communication data between the computing node and all other computing nodes is input into the trained node collaboration prediction model to obtain the output node collaboration performance. The node capability parameters corresponding to the predicted computing power node are obtained by calculating the weighted sum of the predicted computing power performance and the node collaboration performance.

[0088] As can be seen, through the above optional embodiments, by inputting node data and communication data into the computing power performance and node collaboration prediction model respectively and weighting and summing the output results to determine the node capability parameters, it is possible to comprehensively evaluate nodes from two dimensions: individual computing performance and group collaboration potential. This provides an objective and comprehensive quantitative basis for efficient node selection.

[0089] As an optional implementation, the constraints of the dynamic programming algorithm include: In the scheduling strategy, the number of tasks assigned to each allocation node is proportional to the node's capacity parameter. In the scheduling strategy, the difference between the amount of tasks allocated to any two allocation nodes is less than a preset difference threshold. In the scheduling strategy, the scheduling time for each allocation node is inversely proportional to the node capability parameter corresponding to that allocation node; the scheduling time is predicted by inputting the scheduling route and the allocation task of the allocation node into a trained data transmission time prediction model.

[0090] As can be seen, by introducing constraints such as the task volume being proportional to the node's capacity, the task volume difference being limited, and the scheduling time being inversely proportional to the capacity into the dynamic programming algorithm, the scheduling scheme can ensure that it fully leverages the advantages of high-performance nodes while taking into account load balancing and transmission efficiency among nodes, thereby achieving the optimal utilization of computing resources.

[0091] Example 3 Please see Figure 3 , Figure 3 This is another boundary computing power routing and scheduling system based on OAM information publishing disclosed in the embodiments of the present invention. Figure 3 The described boundary computing power routing and scheduling system based on OAM information publishing is applied in data processing systems / data processing equipment / data processing servers (wherein, the server includes local processing servers or cloud processing servers). For example... Figure 3 As shown, the boundary computing power routing and scheduling system based on OAM information publishing can include: Memory 301 storing executable program code; Processor 302 coupled to memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the boundary computing power routing scheduling method based on OAM information publication described in Embodiment 1.

[0092] Example 4 This invention discloses a computer read storage medium that stores a computer program for electronic data interchange, wherein the computer program causes a computer to execute the steps of the boundary computing power routing scheduling method based on OAM information publication described in Embodiment 1.

[0093] Example 5 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of the boundary computing power routing scheduling method based on OAM information publication described in Embodiment 1.

[0094] The foregoing has described specific embodiments of this specification; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0096] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components.

[0097] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented 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.

[0098] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. 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, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0099] 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.

[0100] 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.

[0101] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0102] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0103] 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.

[0104] 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 a process, method, article, or apparatus. Without further limitation, 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 said element.

[0105] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0106] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0107] Finally, it should be noted that the boundary computing power routing and scheduling method and system based on OAM information publication disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for border computing power routing and scheduling based on OAM information publishing, characterized in that, The method includes: The system acquires computing requests to be processed and node-aware data sent by multiple computing power routes; the computing power routes communicate with multiple computing power nodes and determine the node-aware data based on communication interactions. Based on the node perception data, determine the scheduling route corresponding to the computing request from the plurality of computing power routes; Based on the computation request and the node awareness data corresponding to the scheduling route, determine the node scheduling strategy corresponding to the scheduling route; Based on the computation request and the node scheduling policy, an OAM information message is generated, and the OAM information message is sent to the corresponding computing power node through the scheduling route; the computing power node receives the OAM information message and executes the computation request according to the node scheduling policy.

2. The boundary computing power routing and scheduling method based on OAM information publication according to claim 1, characterized in that, The node sensing data includes node data corresponding to each of the computing power nodes and communication data between at least two of the computing power nodes; the node data includes at least one of node hardware data, node software data, node computing power, node network link status, and node sensing data.

3. The boundary computing power routing and scheduling method based on OAM information publication according to claim 1, characterized in that, The step of determining the scheduling route corresponding to the computing request from the plurality of computing power routes based on the node-aware data includes: For each computing power route, the computing scheduling capability corresponding to the computing power route is determined based on the node perception data corresponding to the computing power route and the prediction model. Based on the computing scheduling capability, the scheduling route corresponding to the computing request is determined from the plurality of computing power routes.

4. The boundary computing power routing and scheduling method based on OAM information publication according to claim 3, characterized in that, The step of determining the computing scheduling capability corresponding to the computing power route based on the node perception data corresponding to the computing power route and using a prediction model includes: The node perception data corresponding to the computing power route is input into the trained overall computing power prediction model to obtain the overall predicted computing power corresponding to the computing power route; The node perception data corresponding to the computing power route is input into the trained average computing power prediction model to obtain the average predicted computing power corresponding to the computing power route. Calculate the logarithm of the difference between the overall prediction computing power and the average prediction computing power to obtain the average computing power weight; Calculate the product of the average predicted computing power and the average computing power weight, and the sum of the product and the overall predicted computing power to obtain the computing scheduling capability corresponding to the computing power route.

5. The boundary computing power routing and scheduling method based on OAM information publication according to claim 3, characterized in that, The step of determining the scheduling route corresponding to the computing request from the plurality of computing power routes based on the computing scheduling capability includes: From all the computing power routes, determine a preset number of computing power routes with the highest computing scheduling capacity to obtain multiple candidate scheduling routes; Calculate the average of the differences between the computational scheduling capabilities of every two candidate scheduling routes to obtain the routing differentiation parameters; Determine whether the routing differentiation parameter is greater than a preset first parameter threshold; If so, the candidate scheduling route with the highest computing scheduling capability is determined as the scheduling route corresponding to the computing request; If not, calculate the product of the computational scheduling capability and the computational weight corresponding to each candidate scheduling route to obtain the corrected scheduling capability corresponding to each candidate scheduling route; the computational weight is proportional to the amount of computational requests scheduled by the candidate scheduling route in the historical time period; The candidate scheduling route with the highest corrected scheduling capability is determined as the scheduling route corresponding to the calculation request.

6. The boundary computing power routing and scheduling method based on OAM information publication according to claim 2, characterized in that, The step of determining the node scheduling strategy corresponding to the scheduling route based on the computation request and the node-aware data corresponding to the scheduling route includes: For each computing node corresponding to the scheduling route, the node capability parameters corresponding to the computing node are determined based on the node data and the communication data corresponding to the computing node. From all computing power nodes corresponding to the scheduling route, nodes whose node capability parameters are greater than a preset second parameter threshold are selected to obtain multiple allocation nodes; The objective function is set to minimize the computation time corresponding to the scheduling strategy; the computation time is predicted by inputting the node parameters and the computation task of each allocated node corresponding to the scheduling strategy into a trained computation time prediction model. The scheduling strategy for all the allocated nodes relative to the computation request is iteratively calculated based on the dynamic optimization algorithm until the objective function is satisfied, so as to obtain the node scheduling strategy corresponding to the scheduling route.

7. The boundary computing power routing and scheduling method based on OAM information publication according to claim 6, characterized in that, The step of determining the node capability parameters corresponding to the computing power node based on the node data and the communication data corresponding to the computing power node includes: The node data corresponding to the computing power node is input into the trained computing power performance prediction model to obtain the output predicted computing power performance. The communication data between the computing power node and all other computing power nodes is input into the trained node collaboration prediction model to obtain the output node collaboration performance. The weighted sum of the predicted computing power performance and the node collaboration performance is calculated to obtain the node capability parameters corresponding to the computing power node.

8. The boundary computing power routing and scheduling method based on OAM information publication according to claim 7, characterized in that, The constraints of the dynamic programming algorithm include: In the scheduling strategy, the amount of allocated tasks corresponding to each allocation node is proportional to the node capability parameter corresponding to that allocation node. In the scheduling strategy, the difference between the amount of allocated tasks corresponding to any two allocation nodes is less than a preset difference threshold. In the scheduling strategy, the scheduling time corresponding to each allocation node is inversely proportional to the node capability parameter corresponding to that allocation node; the scheduling time is predicted by inputting the scheduling route and the allocation task of the allocation node into a trained data transmission time prediction model.

9. A boundary computing power routing and scheduling system based on OAM information publishing, characterized in that, The system includes: The acquisition module is used to acquire computing requests to be processed and node-aware data sent by multiple computing power routes; the computing power routes are communicatively connected to multiple computing power nodes and determine the node-aware data based on communication interactions; The first determining module is used to determine the scheduling route corresponding to the computing request from the plurality of computing power routes based on the node perception data; The second determining module is used to determine the node scheduling strategy corresponding to the scheduling route based on the calculation request and the node awareness data corresponding to the scheduling route. The scheduling module is used to generate an OAM information message based on the computing request and the node scheduling policy, and send the OAM information message to the corresponding computing power node through the scheduling route; the computing power node receives the OAM information message and executes the computing request according to the node scheduling policy.

10. A boundary computing power routing and scheduling system based on OAM information publishing, characterized in that, The system includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the boundary computing power routing scheduling method based on OAM information publication as described in any one of claims 1-8.

Images