CN-122019150-A - Computing network energy resource scheduling processing method, storage medium and electronic equipment

CN122019150ACN 122019150 ACN122019150 ACN 122019150ACN-122019150-A

Abstract

The invention discloses a method for scheduling and processing network energy resources, a storage medium and electronic equipment. The method comprises the steps of obtaining scheduling factors corresponding to a plurality of computing power clusters, wherein the scheduling factors comprise computing power factors, network factors, energy factors and service affinity factors, the computing power factors, the network factors and the energy factors are obtained by scoring indexes of regional servers on computing network energy resources, determining scheduling factor weights corresponding to the computing power clusters based on business scene adjustment coefficients corresponding to the computing power clusters, obtaining cluster scores of the computing power clusters based on the scheduling factors and the scheduling factor weights of the computing power clusters, and determining target computing power clusters for executing current service requests from the computing power clusters according to the cluster scores. The invention solves the technical problems of signaling overload and poor flexibility of resource scheduling caused by a centralized management architecture in the algorithm network energy resource scheduling in the related technology.

Inventors

TANG YATING
SHAO ZIHAO
DING CHENGCHENG
PENG KAILAI
XIE RENCHAO

Assignees

紫金山实验室

Dates

Publication Date: 20260512
Application Date: 20260112

Claims (14)

1. The method for processing the dispatching of the network energy resources is characterized by comprising the following steps: The method comprises the steps that scheduling factors corresponding to a plurality of computing power clusters are obtained, wherein the scheduling factors comprise computing power factors, network factors, energy factors and service affinity factors, the computing power factors are obtained by scoring computing power indexes of the corresponding computing power clusters through an area server, the network factors are obtained by scoring the network indexes of the corresponding computing power clusters through the area server, the energy factors are obtained by scoring the energy indexes of the corresponding computing power clusters through the area server, and the service affinity factors are used for indicating the similarity between services provided by the corresponding computing power clusters and current service requests; Determining a scheduling factor weight corresponding to each of the plurality of computing clusters based on a business scenario adjustment coefficient corresponding to each of the plurality of computing clusters, wherein the scheduling factor weight comprises a computing factor weight, a network factor weight, an energy factor weight and a service affinity factor weight, and the business scenario adjustment coefficient is used for adjusting the corresponding scheduling factor weight according to business scenario requirements of the corresponding computing clusters; Obtaining cluster scores corresponding to the plurality of computing power clusters based on the scheduling factors corresponding to the plurality of computing power clusters and the scheduling factor weights corresponding to the plurality of computing power clusters; and determining a target computing force cluster for executing the current service request from the computing force clusters according to the cluster scores corresponding to the computing force clusters.
2. The method of claim 1, wherein the obtaining a scheduling factor for each of the plurality of computing clusters comprises: the service affinity factor for any one of the power clusters is obtained by: Determining the label similarity between a plurality of services provided by any one computing power cluster and the current service request, wherein the label similarity represents the similarity between a service label of the corresponding service provided by any one computing power cluster and the service label of the current service request, and the service label is at least used for indicating the service type of the corresponding service; determining service similarities before the plurality of services and the current service request based on the tag similarities between the plurality of services and the current service request respectively; and determining the service affinity factor of any computing power cluster based on the service similarity between the plurality of services and the service before the current service request.
3. The method of claim 2, wherein the determining the service similarity of the plurality of services to the current service request based on the tag similarity between the plurality of services to the current service request, respectively, comprises: When the service labels corresponding to each service in the plurality of services provided by any computing power cluster are multiple, and the service labels corresponding to the current service request are multiple, obtaining the service similarity between any service in the plurality of services and the current service request by the following method: Determining the maximum tag similarity corresponding to each of the plurality of service tags of the current service request from a tag similarity set corresponding to each of the plurality of service tags of the current service request, wherein the tag similarity set comprises tag similarities between the corresponding tag of the current service request and the plurality of service tags of any service; Determining service similarity between any service and the current service request based on an average value of maximum tag similarity corresponding to each of a plurality of service tags of the current service request; and obtaining the service similarity between the plurality of services and the current service request by adopting a mode of obtaining the service similarity between any one service and the current service request.
4. The method of claim 1, wherein, before the determining the scheduling factor weights for each of the plurality of computing power clusters based on the respective business scenario adjustment coefficients for each of the plurality of computing power clusters, the method further comprises: Determining a cross-domain disaster recovery scene adjustment coefficient corresponding to each of the plurality of computing force clusters based on the fault rate corresponding to each of the plurality of computing force clusters, wherein the cross-domain disaster recovery scene adjustment coefficient is used for quantifying the adjustment requirements of the corresponding computing force clusters in terms of disaster recovery and business continuity; Determining a wind power rapid-reduction scene adjustment coefficient corresponding to each of the plurality of computing power clusters based on wind power fluctuation rates corresponding to each of the plurality of computing power clusters, wherein the wind power fluctuation rates represent wind power fluctuation conditions of areas where the corresponding smooth clusters are located, and the wind power rapid-reduction scene adjustment coefficient is used for quantifying adjustment requirements of the corresponding computing power clusters under the condition of unstable renewable energy supply; and determining a carbon emission assessment scene adjustment coefficient corresponding to each of the plurality of computing force clusters based on the residual carbon emission quota corresponding to each of the plurality of computing force clusters, wherein the carbon emission assessment scene adjustment coefficient is used for quantifying the adjustment requirement of the corresponding computing force cluster in meeting the carbon emission target.
5. The method of claim 1, wherein the determining the scheduling factor weights for each of the plurality of computing force clusters based on the traffic scene adjustment coefficients for each of the plurality of computing force clusters comprises: Determining weight adjustment coefficients corresponding to the plurality of computing power clusters based on the business scene adjustment coefficients corresponding to the plurality of computing power clusters, wherein the weight adjustment coefficients comprise computing power adjustment coefficients, network adjustment coefficients, energy adjustment coefficients and service affinity adjustment coefficients; and optimizing initial factor weights corresponding to the plurality of computing power clusters based on the weight adjustment coefficients corresponding to the plurality of computing power clusters, so as to obtain scheduling factor weights corresponding to the plurality of computing power clusters, wherein the initial factor weights comprise initial computing power weights, initial network weights, initial energy weights and initial service affinity weights.
6. The method of claim 5, wherein the determining the weight adjustment coefficients for each of the plurality of computing force clusters based on the business scenario adjustment coefficients for each of the plurality of computing force clusters comprises: determining the computational power regulation coefficients, the network regulation coefficients and the service affinity regulation coefficients corresponding to the computational power clusters based on the cross-domain disaster recovery scene regulation coefficients corresponding to the computational power clusters; And determining the energy adjustment coefficients corresponding to the plurality of computing power clusters based on the wind power rapid reduction scene adjustment coefficients and the carbon emission assessment scene adjustment coefficients corresponding to the plurality of computing power clusters.
7. The method of claim 1, wherein, in the case where the traffic scene adjustment coefficients include a cross-domain disaster recovery scene adjustment coefficient, a wind-sudden-subtraction scene adjustment coefficient, and a carbon-emission-assessment scene adjustment coefficient, the determining, from the plurality of computing clusters, a target computing cluster for executing the current service request according to the cluster scores corresponding to the plurality of computing clusters, includes: acquiring a last cluster weight corresponding to each of the plurality of computing clusters, wherein the last cluster weight is obtained by executing a last service request of the current service request; summing the last cluster weight corresponding to each of the plurality of computing force clusters with the corresponding cluster score to obtain the current cluster weight corresponding to each of the plurality of computing force clusters; And determining the target computing force cluster from the computing force clusters based on the current cluster weights corresponding to the computing force clusters.
8. The method for processing the dispatching of the network energy resources is characterized by comprising the following steps: Acquiring calculation power indexes, network indexes and energy indexes corresponding to the calculation power clusters respectively; Determining the computational power factors, the network factors and the energy factors corresponding to the computational power clusters based on the computational power indexes, the network indexes and the energy indexes corresponding to the computational power clusters; The method comprises the steps of sending calculation factors, network factors and energy factors corresponding to the calculation force clusters to a main server, determining cluster scores corresponding to the calculation force clusters by the main server, wherein the cluster scores are obtained based on scheduling factors and scheduling factor weights of the calculation force clusters, the scheduling factors comprise the calculation force factors, the network factors, the energy factors and service affinity factors, the service affinity factors are used for indicating the similarity between services provided by the calculation force clusters and current service requests, the scheduling factor weights are obtained based on service scene adjustment coefficients of the calculation force clusters, the scheduling factor weights comprise the calculation force factor weights, the network factor weights, the energy factor weights and the service affinity factor weights, the service scene adjustment coefficients are used for adjusting the corresponding scheduling factor weights according to service scene requirements of the calculation force clusters, and the cluster scores corresponding to the calculation force clusters are used for determining target calculation force clusters used for executing the current service requests from the calculation force clusters.
9. The method of claim 8, wherein the determining the power factor, the network factor, the energy factor, respectively, for the plurality of power clusters based on the power index, the network index, and the energy index, respectively, for the plurality of power clusters, comprises: when the calculation force indexes comprise a plurality of calculation force indexes, the calculation force factor of any calculation force cluster is obtained based on the plurality of calculation force indexes of the any calculation force cluster by the following mode: normalizing a plurality of calculation power indexes of any calculation power cluster to obtain a plurality of normalized calculation power indexes, wherein the plurality of calculation power indexes comprise CPU utilization rate, memory utilization rate, disk input and output rate and video memory utilization rate; Performing weighted average operation on the plurality of normalized calculation indexes to obtain calculation factors of any calculation cluster; based on the calculation force indexes corresponding to the calculation force clusters, the calculation force factors corresponding to the calculation force clusters are obtained by adopting a mode of obtaining the calculation force factors of any calculation force cluster.
10. The method of claim 8, wherein the determining the power factor, the network factor, the energy factor, respectively, for the plurality of power clusters based on the power index, the network index, and the energy index, respectively, for the plurality of power clusters, comprises: In the case that the network index includes a plurality of network indexes, based on the plurality of network indexes of any computing power cluster, the network factor of any computing power cluster is obtained by the following method: Normalizing a plurality of network indexes of any computing power cluster to obtain a plurality of normalized network indexes, wherein the plurality of network indexes comprise standard network delay and network jitter; Performing weighted summation operation on the plurality of normalized network indexes to obtain weighted network indexes; Performing exponential decay operation on the weighted network index based on a preset decay intensity coefficient to obtain a network factor of any computing power cluster; Based on the network indexes corresponding to the plurality of computing power clusters, the network factors corresponding to the plurality of computing power clusters are obtained by adopting a mode of obtaining the network factors of any computing power cluster.
11. The method of claim 8, wherein the determining the power factor, the network factor, the energy factor, respectively, for the plurality of power clusters based on the power index, the network index, and the energy index, respectively, for the plurality of power clusters, comprises: when the energy indexes comprise a plurality of energy indexes, the energy factor of any computing power cluster is obtained based on the plurality of energy indexes of the computing power cluster by the following mode: normalizing a plurality of energy indexes of any computing power cluster to obtain a plurality of normalized energy indexes, wherein the plurality of energy indexes comprise carbon emission factors and electricity charge cost; Performing weighted average operation on the plurality of normalized energy indexes to obtain energy factors of any computing power cluster; And obtaining the energy factors corresponding to the plurality of computing power clusters by adopting a mode of obtaining the energy factors of any computing power cluster based on the energy indexes corresponding to the plurality of computing power clusters.
12. A non-volatile storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the method of computing network resource scheduling processing of any one of claims 1 to 11.
13. An electronic device comprising one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of computing network energy resource scheduling processing of any of claims 1-11.
14. A computer program product comprising a computer program, characterized in that the computer program when executed by a processor implements the steps of the method for processing the scheduling of computational network resources of any one of claims 1 to 11.

Description

Computing network energy resource scheduling processing method, storage medium and electronic equipment Technical Field The invention relates to the field of intelligent scheduling, in particular to a method for scheduling and processing network energy resources, a storage medium and electronic equipment. Background The technical background of the current computing network energy fusion scheduling system is deeply rooted in regional resource allocation and intelligent transformation of a novel power system. With the explosive development of 5G, artificial intelligence and Internet of things technologies, the annual average increase in the demand for computing power and the scale of data centers are increasing. However, there is a significant spatial mismatch between the distribution of the computing power resources and the energy supply, for example, a part of the areas bear most of the computing power demands in the whole area, but face the constraints of scarcity of land, high electricity price and carbon emission indexes, while another part of the areas have a higher green electricity proportion and sufficient land resources, but the computing power utilization rate is lower than 10% for a long time, resulting in high resource waste. In order to solve the contradiction, more and more high-computation hub nodes and data center clusters are aimed at constructing a scheduling network for cross-regional resource supply and demand coordination. The related art discloses a computational network energy collaborative planning method considering the time sequence scheduling of computational tasks, which is used for realizing multi-level interaction of computational resources, computational tasks, load demands and energy supply, achieving the optimal overall configuration scheme of the computational network energy and realizing the computational network energy collaborative planning of the time sequence scheduling of the computational tasks. However, the method has the following problems that 1) a differentiated measurement mechanism is lacked, the centralized management architecture needs to report node data (such as CPU load, bandwidth and electricity price) in full quantity, and control plane signaling is overloaded. When the node size exceeds 10 tens of thousands, a large amount occupies the transmission bandwidth. 2) The multi-factor fusion is insufficient, namely, a scheduling model in the related technology only optimizes a network dimension target or an energy dimension target, and consideration factors are incomplete. 3) The scheduling algorithm in the related art is usually a fixed optimized weight factor due to insufficient multi-factor coordination. When facing different demand scenes, the method is difficult to flexibly adjust and realize smooth distribution of time sequence requests. In summary, the related art has the problems of signaling overload, poor scheduling flexibility and the like caused by the centralized management architecture in the network energy resource scheduling. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the invention provides a processing method, a storage medium and electronic equipment for dispatching of computing network energy resources, which at least solve the technical problems of signaling overload and poor flexibility of resource dispatching caused by a centralized management architecture in computing network energy resource dispatching in the related technology. According to one aspect of the embodiment of the invention, a computing network energy resource scheduling processing method is provided, and the computing network energy resource scheduling processing method comprises the steps of obtaining scheduling factors corresponding to a plurality of computing force clusters respectively, wherein the scheduling factors comprise computing force factors, network factors, energy factors and service affinity factors, the computing force factors are obtained by scoring computing force indexes of the corresponding computing force clusters through a regional server, the network factors are obtained by scoring the network indexes of the corresponding computing force clusters through the regional server, the energy factors are obtained by scoring the energy indexes of the corresponding computing force clusters through the regional server, the service affinity factors are used for indicating the similarity between services provided by the corresponding computing force clusters and current service requests, determining scheduling factor weights corresponding to the computing force clusters respectively based on service scene adjustment coefficients corresponding to the computing force clusters respectively, the scheduling factor weights comprise computing force factor weights, network factor weights, energy factor weights and service affinity factor weights, the service scene adjustment weights are us