CN-122027702-A - Calculation force request scheduling method and system based on optimization calculation force routing protocol

CN122027702ACN 122027702 ACN122027702 ACN 122027702ACN-122027702-A

Abstract

The invention discloses a calculation force request scheduling method and a calculation force request scheduling system based on an optimized calculation force routing protocol, wherein the method comprises the steps of obtaining a calculation force request to be scheduled and a plurality of corresponding historical transit routes; the method comprises the steps of setting a preset protocol for carrying out demand analysis on the calculation force requests in a history transfer route, obtaining corresponding demand analysis results of the calculation force requests from each history transfer route request, determining calculation force demand parameters corresponding to the calculation force requests according to all the demand analysis results, and determining calculation force node transfer paths corresponding to the calculation force requests based on a dynamic programming algorithm according to the calculation force demand parameters. Therefore, the method and the system can realize accurate calculation power scheduling path planning based on historical route cooperative analysis and demand parameters, improve the distribution efficiency of distributed calculation power resources and the task completion rate, and reduce calculation power waste or delay risk caused by inaccurate demand analysis or suboptimal path.

Inventors

YANG JIANREN
JI JUNGANG
ZHU CHUNXIA

Assignees

深圳市证通电子股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260226

Claims (10)

1. A power request scheduling method based on an optimized power routing protocol, the method comprising: Acquiring a calculation force request to be scheduled and a plurality of corresponding historical transfer routes, wherein the historical transfer routes are internally provided with a preset protocol for carrying out demand analysis on the calculation force request; Forwarding a routing request to each history to obtain a corresponding demand analysis result of the calculation request; according to all the demand analysis results, determining the corresponding calculation force demand parameters of the calculation force request; and determining a computational force node transfer path corresponding to the computational force request based on a dynamic programming algorithm according to the computational force demand parameters, wherein the computational force node transfer path is used for driving the computational force request to be processed and transferred among a plurality of computational force nodes until a calculation result is obtained.
2. The optimal power routing protocol-based power request scheduling method according to claim 1, wherein the protocol built in the history transit route is used for executing the following steps: Judging whether the current historical transit route is the first route reached by the calculation request or not, and obtaining a judging result: When the judgment result is yes, inputting the calculation force request into a trained demand analysis model to obtain a demand analysis result; When the judgment result is negative, determining all the historical transit routes which the computing power request passes before reaching the current historical transit route, and obtaining at least one previous route; and determining a demand analysis result of the transit route of the computing power request in the current history according to the previous demand analysis result of the previous route.
3. The method for scheduling a power request based on an optimized power routing protocol according to claim 2, wherein the determining a demand analysis result of the power request for transit routing in the current history according to a previous demand analysis result of the previous route comprises: inputting the calculation force request into the demand analysis model to obtain a first demand analysis result when the number of previous routes is 1; calculating a similarity between the first demand analysis result and a previous demand analysis result of the previous route; When the similarity is larger than a first similarity threshold, determining the first demand analysis result as a demand analysis result of the transit route of the computing power request in the current history; when the similarity is smaller than a first similarity threshold, determining an intersection of the previous demand analysis result and the first demand analysis result as a demand analysis result of the transit route of the calculation request in the current history; when the number of the previous routes is larger than 1, calculating the collective similarity corresponding to the previous demand analysis results of all the previous routes; Determining a result intersection of previous demand analysis results for all of the previous routes as a demand analysis result for the transit route of the power request in the current history when the collective similarity is greater than a second similarity threshold; when the collective similarity is smaller than a second similarity threshold, inputting the calculation force request into the demand analysis model to obtain a second demand analysis result; and calculating the union of the second demand analysis result and the result intersection to obtain the demand analysis result of the computation power request transit route in the current history.
4. The optimal power routing protocol-based power request scheduling method according to claim 1, wherein the demand analysis result includes at least one of a power hardware demand, a power software demand, a power goal, a power task type, and a power calculation effect demand.
5. The method for scheduling the power request based on the optimized power routing protocol according to claim 1, wherein the determining the power demand parameter corresponding to the power request according to all the demand analysis results includes: determining a credibility parameter corresponding to each historical transit route; And calculating a weighted intersection of all the demand analysis results to obtain a power calculation demand parameter corresponding to the power calculation request, wherein the duty ratio weight of each demand analysis result in the power calculation demand parameter is in direct proportion to the corresponding credibility parameter.
6. The method for scheduling power requests based on optimized power routing protocol according to claim 5, wherein determining the reliability parameter corresponding to each historical transit route comprises: For each historical transit route, calculating the matching degree between the demand analysis result in each historical analysis record of the historical transit route and the final demand feedback of the corresponding historical calculation force request; And calculating a weighted sum average value of all the matching degrees corresponding to the historical transit routes to obtain a credibility parameter corresponding to the historical transit routes, wherein the calculation weight corresponding to each matching degree is inversely proportional to the number of routes passed by the historical calculation force request in the corresponding historical analysis record before reaching the current historical transit route.
7. The method for scheduling the computing power request based on the optimized computing power routing protocol according to claim 1, wherein the determining the computing power node transfer path corresponding to the computing power request based on the dynamic programming algorithm according to the computing power demand parameter comprises: Screening nodes matched with the calculation force demand parameters from all candidate calculation force nodes to obtain a plurality of matched calculation force nodes; And determining a computational force node transfer path which corresponds to the computational force request and is calculated by transmission among the plurality of matched computational force nodes based on a path planning algorithm.
8. The optimal power routing protocol-based power request scheduling method of claim 7, wherein the algorithm function of the path planning algorithm comprises: the transmission order precedent degree of each matched computing force node in the computing force node transfer path is in direct proportion to the matching degree between the corresponding node historical calculation requirement and the computing force requirement parameter; The node network distance between any two matched computing nodes adjacent in transmission order in the computing node transfer path is smaller than a preset distance threshold; And the performance difference degree between the computing power performances of the matched computing power nodes adjacent to any two transmission orders in the computing power node transfer paths is smaller than a preset difference degree threshold value.
9. A power request scheduling system based on an optimized power routing protocol, the system comprising: the system comprises an acquisition module, a scheduling module and a scheduling module, wherein the acquisition module is used for acquiring a calculation force request to be scheduled and a plurality of corresponding historical transfer routes, and a preset protocol for carrying out demand analysis on the calculation force request is arranged in the historical transfer routes; the request module is used for transferring the route request to each history to obtain a corresponding demand analysis result of the calculation force request; the determining module is used for determining the calculation force demand parameters corresponding to the calculation force request according to all the demand analysis results; The power calculation request processing module is used for processing the power calculation request according to the power calculation demand parameters, determining a power calculation node transfer path corresponding to the power calculation request based on a dynamic programming algorithm, and the power calculation node transfer path is used for driving the power calculation request to be processed and transferred among a plurality of power calculation nodes until a calculation result is obtained.
10. A power request scheduling system based on an optimized power routing protocol, the system comprising: a memory storing executable program code; a processor coupled to the memory; The processor invokes the executable program code stored in the memory to perform the optimal power routing protocol based power request scheduling method of any one of claims 1-8.

Description

Calculation force request scheduling method and system based on optimization calculation force routing protocol Technical Field The invention relates to the technical field of data processing, in particular to a computing power request scheduling method and system based on an optimized computing power routing protocol. Background With the rapid increase of the demand for distributed computing resources, cloud computing enterprises and data center operators increasingly pay attention to optimizing resource allocation and task execution efficiency through precise scheduling path planning, wherein how to realize efficient node transfer so as to reduce computing waste or delay risk becomes a key technical problem. In the prior art, the computing force request data to be scheduled is generally obtained, a fixed routing table or a simple greedy algorithm is adopted to distribute computing force nodes, and a transfer path is generated based on static demand estimation so as to support basic resource scheduling. The existing solution is lack of comprehensive demand analysis results of a plurality of historical transit routes and accurate determination of calculation force demand parameters, so that it is difficult to apply a dynamic programming algorithm to generate node transfer paths adapting to a complex network, the resource allocation efficiency and the task completion rate are insufficient, calculation force waste or execution delay is easily caused by inaccurate demand analysis or suboptimal paths, and the overall performance and reliability of a distributed calculation force system are limited. It can be seen that the prior art has defects and needs to be solved. Disclosure of Invention The technical problem to be solved by the invention is to provide the computational power request scheduling method and system based on the optimized computational power routing protocol, which can realize accurate computational power scheduling path planning based on historical route collaborative analysis and demand parameters, improve the distribution efficiency of distributed computational power resources and the task completion rate, and reduce the computational power waste or delay risk caused by inaccurate demand analysis or suboptimal path. In order to solve the technical problem, the first aspect of the present invention discloses a computing power request scheduling method based on an optimized computing power routing protocol, which comprises the following steps: Acquiring a calculation force request to be scheduled and a plurality of corresponding historical transfer routes, wherein the historical transfer routes are internally provided with a preset protocol for carrying out demand analysis on the calculation force request; Forwarding a routing request to each history to obtain a corresponding demand analysis result of the calculation request; according to all the demand analysis results, determining the corresponding calculation force demand parameters of the calculation force request; and determining a computational force node transfer path corresponding to the computational force request based on a dynamic programming algorithm according to the computational force demand parameters, wherein the computational force node transfer path is used for driving the computational force request to be processed and transferred among a plurality of computational force nodes until a calculation result is obtained. As an optional implementation manner, in the first aspect of the present invention, a protocol built in the history transit route is used to perform the following steps: Judging whether the current historical transit route is the first route reached by the calculation request or not, and obtaining a judging result: When the judgment result is yes, inputting the calculation force request into a trained demand analysis model to obtain a demand analysis result; When the judgment result is negative, determining all the historical transit routes which the computing power request passes before reaching the current historical transit route, and obtaining at least one previous route; and determining a demand analysis result of the transit route of the computing power request in the current history according to the previous demand analysis result of the previous route. As an optional implementation manner, in the first aspect of the present invention, the determining, according to a previous demand analysis result of the previous route, a demand analysis result of the transit route of the computing power request in the current history includes: inputting the calculation force request into the demand analysis model to obtain a first demand analysis result when the number of previous routes is 1; calculating a similarity between the first demand analysis result and a previous demand analysis result of the previous route; When the similarity is larger than a first similarity threshold, determining the first demand analysis r