CN114385345A

CN114385345A - Resource scheduling method, intelligent identification resource scheduling method and related equipment

Info

Publication number: CN114385345A
Application number: CN202011139984.3A
Authority: CN
Inventors: 吴南南; 薛天泊; 马艳芳; 吴凡
Original assignee: Nuctech Co Ltd
Current assignee: Nuctech Co Ltd
Priority date: 2020-10-22
Filing date: 2020-10-22
Publication date: 2022-04-22
Anticipated expiration: 2040-10-22

Abstract

The embodiment of the disclosure provides a resource scheduling method, an intelligent identification resource scheduling method and related equipment. The resource scheduling method comprises the following steps: acquiring a current service load; if the first service processing resource is in a full working load state at the current moment, obtaining first current working load information and second current working load information; acquiring first network connection information and second network connection information; acquiring first historical service data of a second service processing resource and acquiring second historical service data of a third service processing resource; determining a target service processing resource from a second service processing resource and a third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

Description

Resource scheduling method, intelligent identification resource scheduling method and related equipment

Technical Field

The present disclosure relates to the field of computer and communication technologies, and in particular, to a resource scheduling method and apparatus, an intelligent identification resource scheduling method, an electronic device, and a computer-readable storage medium.

Background

At present, in a service scene of intelligent identification of security inspection images, the following two situations exist:

firstly, each security check device or security check point is provided with a single intelligent identification device, hardware and software which are required by security check images acquired by the intelligent identification security check device or security check point are carried, and the single intelligent identification device is used for processing business images or security check images generated by local security check devices.

The first situation is limited by physical isolation among the single intelligent identification devices, and the single intelligent identification devices cannot process the service loads of other busy security check devices or security check points when the single intelligent identification devices are idle. Such architectures directly result in the computing power of these monolithic intelligent identification devices not being efficiently utilized, such that some computing power has to be overloaded and some computing power is directly wasted. The waste of resources directly causes the consumption of cost, and the input-output ratio cannot reach the expected standard.

In addition, if a single intelligent identification device cannot process a new service due to the fact that a full load of computing power is reached, the subsequent service cannot be continuously developed, normal circulation of the service is affected, and service stagnation or errors occur, for example, all service images or security inspection images to be subjected to security inspection of security inspection devices or security inspection points corresponding to the single intelligent identification device cannot be intelligently identified or can only be delayed, security inspection efficiency is affected, service value is reduced or service meaning is lost, and missed service images or security inspection images may be generated.

Secondly, all or part of the security check equipment or security check points are connected to an intelligent identification service center or cluster, and intelligent identification service is completed through the calculation force of service nodes in the center or cluster.

However, in some service scenarios, a centralized intelligent identification service center or a cluster cannot be deployed in an engineering structure, and it is necessary to distribute computing power to different physical locations to develop intelligent identification services using a single intelligent identification device as a unit.

It is to be noted that the information disclosed in the above background section is only for enhancement of understanding of the background of the present disclosure.

Disclosure of Invention

The embodiment of the disclosure provides a resource scheduling method and device, an electronic device and a computer-readable storage medium, which can solve the technical problem of how to improve the scheduling efficiency of business processing resources when the business processing resources are dispersed to different physical locations in the related art.

Additional features and advantages of the disclosure will be set forth in the detailed description which follows, or in part will be obvious from the description, or may be learned by practice of the disclosure.

The disclosed embodiment provides a resource scheduling method, which is applied to a resource scheduling system, wherein the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device. The method comprises the following steps: acquiring a current service load generated by the first service generation equipment at the current moment; if the first service processing resource is in a full workload state at the current moment, acquiring first current workload information of the second service processing resource and second current workload information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; obtaining first historical service data of the second service processing resource and obtaining second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

The disclosed embodiment provides a resource scheduling method, which is applied to a resource scheduling system, wherein the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device. The method comprises the following steps: acquiring a current service load generated by the first service generation equipment at the current moment; if the first service processing resource is in a first working load state at the current moment, obtaining first current working load information of the second service processing resource and second current working load information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information and the second network connection information; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

The disclosed embodiment provides a resource scheduling method, which is applied to a resource scheduling system, wherein the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device. The method comprises the following steps: acquiring a current service load generated by the first service generation equipment at the current moment; if the first service processing resource is in a first working load state at the current moment, acquiring first historical service data of the second service processing resource and acquiring second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first historical service data and the second historical service data; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

The embodiment of the disclosure provides an intelligent identification resource scheduling method, which is applied to an intelligent identification resource scheduling system, wherein the intelligent identification resource scheduling system comprises a first agent cluster, the first agent cluster comprises a first single intelligent identification device agent service node, the first single intelligent identification device agent service node corresponds to a first single intelligent identification device, a second single intelligent identification device and a third single intelligent identification device, the first single intelligent identification device corresponds to a first security inspection device, the second single intelligent identification device corresponds to a second security inspection device, and the third single intelligent identification device corresponds to a third security inspection device. The method comprises the following steps: acquiring a security inspection image acquired by the first security inspection equipment at the current moment; generating a current service load according to the security check image; if the first single intelligent identification device is in a full workload state at the current moment, acquiring first current workload information of the second single intelligent identification device and second current workload information of the third service processing resource; acquiring first network connection information between the second single intelligent identification device and the first proxy service node, and acquiring second network connection information between the third single intelligent identification device and the second proxy service node; acquiring first historical service data of the second single intelligent identification device, and acquiring second historical service data of the third single intelligent identification device; determining target single intelligent identification equipment from the second single intelligent identification equipment and the third single intelligent identification equipment according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; and distributing the current service load to the target single intelligent identification equipment so as to enable the target single intelligent identification equipment to identify the suspected object in the current service load.

The disclosed embodiment provides a resource scheduling device, the device includes: the device is applied to a resource scheduling system, the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generating device, the second service processing resource corresponds to a second service generating device, and the third service processing resource corresponds to a third service generating device. The device comprises: a current service load obtaining unit, configured to obtain a current service load generated by the first service generation device at the current time; a workload information obtaining unit, configured to obtain first current workload information of the second service processing resource and second current workload information of the third service processing resource if the first service processing resource is in a full workload state at the current time; a network connection information obtaining unit, configured to obtain first network connection information between the second service processing resource and the first proxy service node, and obtain second network connection information between the third service processing resource and the second proxy service node; a historical service data obtaining unit, configured to obtain first historical service data of the second service processing resource, and obtain second historical service data of the third service processing resource; a target processing resource determining unit, configured to determine a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical service data and the second historical service data; a current service load allocation unit, configured to allocate the current service load to the target service processing resource, so that the target service processing resource processes the current service load.

The embodiment of the disclosure provides a resource scheduling system, which includes a first proxy cluster, where the first proxy cluster includes a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device. The first proxy service node is used for acquiring the current service load generated at the current moment from the first service generation equipment; when the first service processing resource is in a full working load state at the current moment, acquiring first current working load information of the second service processing resource and second current working load information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; obtaining first historical service data of the second service processing resource and obtaining second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; distributing the current service load to the target service processing resource; the target traffic processing resource is configured to receive and process the current traffic load from the first proxy service node.

The disclosed embodiments provide a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method as described in the above embodiments.

An embodiment of the present disclosure provides an electronic device, including: at least one processor; a storage device configured to store at least one program that, when executed by the at least one processor, causes the at least one processor to implement the method as described in the above embodiments.

In the technical solutions provided in some embodiments of the present disclosure, in a case that service processing resources are dispersed to different physical locations, here, a first service processing resource corresponds to a first service generation device, a second service processing resource corresponds to a second service generation device, and a third service processing resource corresponds to a third service generation device. The method comprises the steps of abstracting a functional elastic layer from business processing resources, wherein the functional elastic layer is called as a proxy service node, each proxy service node corresponds to at least one business processing resource, the proxy service node has the capacity of forming a cluster through local self-organization, one or more business processing resources can be called as required, the edge cooperation effect is achieved, and the weak capacity of a single business processing resource is converted into the strong capacity with group addition (group intelligence formed by group relation and system). Taking the first proxy service node as an example, assuming that the first proxy service node corresponds to the first service processing resource, the second service processing resource and the third service processing resource, if the first service processing resource does not have enough computing power to process the current service load generated by the corresponding first service generation device at the current time, the first proxy service node may select a target service processing resource suitable for processing the current service load according to the first current workload information, the first network connection information and the first historical service data of the second service processing resource, and the second current workload information, the second network connection information and the second historical service data of the third service processing resource, on one hand, it is avoided that a certain service processing resource cannot process a new service load due to overload, so that the new service load cannot be processed or can only be delayed to be processed, but reduces its business value or loses its business meaning. On the other hand, the scheduling strategy of the current workload information, the network connection information and the historical service data of each service processing resource is comprehensively considered, so that the optimal target service processing resource can be selected from a plurality of service processing resources, and the technical effect of optimal overall effect is achieved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and together with the description, serve to explain the principles of the disclosure. It is to be understood that the drawings in the following description are merely exemplary of the disclosure, and that other drawings may be derived from those drawings by one of ordinary skill in the art without the exercise of inventive faculty. In the drawings:

FIG. 1 illustrates an architectural diagram of a resource scheduling system to which an embodiment of the present disclosure may be applied;

FIG. 2 illustrates an architectural diagram of a resource scheduling system to which an embodiment of the present disclosure may be applied;

FIG. 3 schematically shows a flow chart of a resource scheduling method according to an embodiment of the present disclosure;

fig. 4 schematically shows a schematic diagram of a resource scheduling method according to an embodiment of the present disclosure;

fig. 5 schematically shows a schematic diagram of a resource scheduling method according to an embodiment of the present disclosure;

FIG. 6 shows an architectural diagram of a resource scheduling system to which an embodiment of the present disclosure may be applied;

FIG. 7 shows an architectural diagram of a resource scheduling system to which an embodiment of the present disclosure may be applied;

FIG. 8 schematically shows a flow chart of a resource scheduling method according to an embodiment of the present disclosure;

fig. 9 schematically shows a schematic diagram of a resource scheduling method according to an embodiment of the present disclosure;

FIG. 10 schematically shows a flow chart of a resource scheduling method according to an embodiment of the present disclosure;

fig. 11 schematically shows a schematic diagram of a resource scheduling method according to an embodiment of the present disclosure;

fig. 12 schematically shows a schematic diagram of a resource scheduling method according to an embodiment of the present disclosure;

FIG. 13 schematically illustrates a flow diagram of a method of intelligently identifying resource scheduling according to an embodiment of the present disclosure;

fig. 14 schematically shows a block diagram of a resource scheduling apparatus according to an embodiment of the present disclosure;

fig. 15 schematically shows a structural schematic diagram of an electronic device according to an embodiment of the present disclosure.

Detailed Description

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The same reference numerals denote the same or similar parts in the drawings, and thus, a repetitive description thereof will be omitted.

The described features, structures, or characteristics of the disclosure may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the subject matter of the present disclosure can be practiced without one or more of the specific details, or with other methods, components, devices, steps, and the like. In other instances, well-known methods, devices, implementations, or operations have not been shown or described in detail to avoid obscuring aspects of the disclosure.

The drawings are merely schematic illustrations of the present disclosure, in which the same reference numerals denote the same or similar parts, and thus, a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in the form of software, or in at least one hardware module or integrated circuit, or in different networks and/or processor means and/or microcontroller means.

In this specification, the terms "a", "an", "the", "said" and "at least one" are used to indicate the presence of at least one element/component/etc.; the terms "comprising," "including," and "having" are intended to be inclusive and mean that there may be additional elements/components/etc. other than the listed elements/components/etc.; the terms "first," "second," and "third," etc. are used merely as labels, and are not limiting on the number of their objects.

Some terms referred to in the embodiments of the present disclosure are explained first.

Omission detection: certain suspected objects are missed in security inspection, so that the suspected objects successfully pass through the inspection and enter places needing safety guarantee.

Suspicion article: in a security inspection scene, illegal articles, non-compliant articles, articles not on a list, articles exceeding a pre-reported number, and the like, which need to be identified, may be specifically determined according to an actual scene.

AI (Artificial Intelligence) monomer: the single intelligent identification device carries hardware and software required by intelligent identification, and is one of service processing resources. When the single intelligent identification device and the agent service node corresponding to the single intelligent identification device are in a one-to-one relationship, the AI single body and the agent service node thereof mentioned in the embodiment of the present disclosure may be disposed in the same physical device, or may be disposed in independent physical devices, respectively.

Intelligent identification: and (3) identifying the service load by adopting an AI technology, for example, identifying the suspected object in the security inspection image by adopting a machine learning model, a deep learning model and the like.

AI monomer proxy service node (AI Device Agent): is one type of proxy service node. The AI monomer agent service node is used as an avatar of hardware or single-point software service of the intelligent identification equipment, and is used for communicating with the service generation equipment, the intelligent identification equipment or service, calling intelligent identification computing resources, managing the service generation equipment, the intelligent identification equipment or service and the like. In addition, the AI monomer agent service node also comprises the distributed function of the AI monomer agent service node as one node in the distributed cluster.

Service load: the task load generated by the business system. For example, a scanned perspective image (a kind of security image) of an item of baggage generated by a security inspection device or a security inspection point in a security inspection system corresponds to an identification task or a mapping task, which is a traffic load.

The service generating place: the location or position of the traffic load may be a hardware device or a software service, and the form is not limited. For example, the venue may be a security point or security device in a security service.

Business meaning: some traffic loads only have meaning within a certain time frame, and if the time frame is exceeded, the timeliness is lost, and the traffic load will no longer have the corresponding traffic meaning.

Edge synergy: all the nodes as a whole cooperate with each other, and resources are dynamically allocated according to needs or priorities, so that the working load of each node is reasonable and balanced.

Network distance: the transmission distance between two network nodes can be measured by PING (Packet Internet Groper, Internet Packet explorer, program for testing network connection quantity) delay, time consumption, routing hop count, and other data.

PING: the method is mainly used for detecting whether a target host is connectable or not, a source host sends a request message To the target host, waits for a response message returned by the target host, and according To the receiving and sending conditions of the message, the result contains byte number, reaction Time, Time To Live (TTL) and the like. The delay in the transceiving process, the round trip time consumption in the whole process and the route hop number between the source host and the target host mapped according to the survival time can be obtained.

Data screening, feature analysis, modeling analysis, learning and training: some big data, machine learning related techniques.

The following detailed description of exemplary embodiments of the disclosure refers to the accompanying drawings.

Fig. 1 shows an architectural diagram of a resource scheduling system to which an embodiment of the present disclosure may be applied. As shown in fig. 1, the resource scheduling system 100 provided in the embodiment of fig. 1 may include a first proxy cluster. The first proxy cluster may include at least one proxy service node, and the at least one proxy service node may include the first proxy service node. It should be noted that, although only the first proxy service node is illustrated here, the number of proxy service nodes in the first proxy cluster is not limited in this disclosure, and the first proxy service node may be any one proxy service node in the first proxy cluster.

As shown in fig. 1, the first proxy service node may correspond to a first traffic processing resource of a first traffic generation site, a second traffic processing resource of a second traffic generation site, and a third traffic processing resource of a third traffic generation site. The first service processing resource may correspond to a first service generation device at a first service generation site, the second service processing resource may correspond to a second service generation device at a second service generation site, and the third service processing resource may correspond to a third service generation device at a third service generation site, that is, the service generation device and the corresponding service processing resource are distributed at different physical locations.

It should be noted that, although the first proxy service node shown in fig. 1 corresponds to three service processing resources (i.e., a first service processing resource, a second service processing resource, and a third service processing resource), other proxy service nodes corresponding to one service processing resource, two service processing resources, or more than three service processing resources may also exist in the first proxy cluster, and the number of the service processing resources corresponding to each proxy service node is not limited in the present disclosure.

In the embodiment of the present disclosure, a service processing resource refers to any resource capable of processing a corresponding service, and the service processing resources used are different according to different service scenarios. For example, the present disclosure may refer to computing resources for performing image recognition, image processing, voice recognition, character recognition, numerical calculation, machine translation, machine question answering, sequence processing, and the like, and may also refer to network resources required for performing data transmission, and the present disclosure does not limit specific types and meanings of service processing resources. The service generation device is a device for acquiring a service load or a service task to be processed.

In this embodiment of the present disclosure, the first proxy service node of the first proxy cluster may be an independent physical device, and may also be disposed in any one or more service generation devices and/or service processing resources in the first proxy cluster, which is not limited in this disclosure.

As shown in fig. 1, the first proxy service node may include a service interface, and through the service interface, the first proxy service node may obtain, from any corresponding service generating device (for example, the first to third service generating devices), a service load collected or generated by the service generating device, may also obtain, from any corresponding service processing resource, a workload state, workload information, network connection information, and historical service data of each service processing resource, and may also select, according to the obtained workload state, workload information, network connection information, and historical service data of each service processing resource, an appropriate target service processing resource from the corresponding service processing resource, and send the service load to the selected target service processing resource through the service interface. In some embodiments, the first proxy service node may further receive, through the service interface, a service processing result (e.g., an identification result of the security check image) returned after the target service processing resource processes the service load.

In the embodiment of fig. 1, the first proxy service node directly corresponds to the first to third traffic processing resources. In other embodiments, the first proxy service node may also indirectly correspond to some of the traffic processing resources through other proxy service nodes.

For example, as shown in fig. 2, the difference from the above-mentioned fig. 1 embodiment is that the first proxy cluster in the resource scheduling system 200 may further include a second proxy service node, where the second proxy service node directly corresponds to a third traffic processing resource of a third traffic generation site, and the third traffic processing resource corresponds to a third traffic generation device of the third traffic generation site. At this time, the first proxy service node communicates with the second proxy service node, and may obtain the workload state, the workload information, the network connection information, and the historical service data of the third service processing resource through the service interface of the second proxy service node, and obtain the service load collected or generated by the third service generation device through the service interface of the second proxy service node.

The resource scheduling system provided by the embodiment of the disclosure integrates the computational power of all the single business processing resources through the customized business interface in the proxy service node or the unified standard interface under the scene that deployment is required according to the single business processing resources, and manages and schedules in a unified manner.

In the embodiment of the present disclosure, the single service processing resource refers to that hardware and software resources required for completing service processing are independently set in a physical space, rather than that the hardware and software resources required for completing service processing are set in a centralized manner.

The interfaces of the proxy service node may include an interface for performing data communication with each service processing resource and the service generation device, an interface for acquiring current workload information of each service processing resource, an interface for acquiring network connection information of each service processing resource, and an interface for acquiring historical service data (for example, historical service load generation data and historical service load distribution data) of each service processing resource (all of the above are service interfaces).

In an exemplary embodiment, the traffic interfaces of the proxy service node may further include an interface for calculating a network distance between the proxy service node and each traffic processing resource, and the like.

In an exemplary embodiment, when the resource scheduling system is a distributed cluster including a plurality of agent clusters, each agent cluster may further include a control node, and the control node may be used for agent service node management and scheduling of the distributed cluster.

In an exemplary embodiment, each proxy service node in the respective proxy cluster may include a traffic interface and a control interface. In each agent cluster, a control node of the agent cluster is elected through a control interface of each agent service node in the agent cluster. The communication between different proxy service nodes can also be realized through the control interface corresponding to the proxy service node, and the service interface is the interface for realizing the communication between the corresponding service generating equipment and the service processing resource.

In the embodiment of the present disclosure, the function elastic layer is abstracted from the corresponding at least one service processing resource through the interface, which may be referred to as a proxy service node. One proxy service node can correspond to one or more service processing resources to form a proxy cluster, and can fully schedule and utilize the computing power of the service processing resources to convert the weak capability of a single hardware device into the strong capability with group addition. The agent has the capability of forming a cluster by local self-organization, and can call hardware computing resources corresponding to one or more agent service nodes as required to achieve the effect of edge cooperation. The method avoids that a certain service processing resource cannot process the new service load due to overload, so that the new service load is not processed or only can be delayed to complete the processing, thereby reducing the service value or losing the service meaning.

Fig. 3 schematically shows a flow chart of a resource scheduling method according to an embodiment of the present disclosure. The method provided in fig. 3 can be applied to the resource scheduling system shown in fig. 1 or fig. 2. As shown in fig. 3, the method provided by the embodiment of the present disclosure may include the following steps.

In step S310, a current service load generated by the first service generating device at the current time t is obtained.

For example, also taking a security check plan scenario as an example, the current traffic load may be a security check image or a part of the security check image acquired by the first traffic generating device (e.g., a certain security check point or one or more security check devices of the security check point) in real time.

In step S320, if the first service processing resource is in a full workload state at the current time, obtain first current workload information of the second service processing resource and second current workload information of the third service processing resource.

In this embodiment of the present disclosure, the full workload state refers to a state in which a local service processing resource corresponding to a service generation device that currently generates a current service load cannot process the current service load.

For example, taking a security inspection scene as an example, a single intelligent identification device is used as a service processing resource, a first service generation device collects a frame of security inspection image at the current time, and since the current workload of the corresponding first service processing resource is completely full or although not completely full, the frame of security inspection image is processed in real time without much computational power, the first service processing resource is in a full workload state at this time. Or, the security inspection image may be divided into a plurality of image segments, that is, a plurality of traffic loads are generated, and if the first traffic processing resource can only process a part of the traffic loads in the plurality of traffic loads at the current time, or cannot process the plurality of traffic loads, it is determined that the first traffic processing resource is in a full workload state, and a traffic load that cannot be processed by the first traffic processing resource at the current time in the plurality of traffic loads is referred to as a current traffic load.

In an exemplary embodiment, obtaining the first current workload information of the second traffic processing resource and the second current workload information of the third traffic processing resource may include: acquiring a first historical service load processed by the second service processing resource and first historical processing time information thereof, and a second historical service load processed by the third service processing resource and second historical processing time information thereof; acquiring first historical service processing duration information of the second service processing resource according to the first historical service load and the first historical processing time information thereof, and acquiring second historical service processing duration information of the third service processing resource according to the second historical service load and the second historical processing time information thereof; obtaining first current to-be-processed service load data of the second service processing resource and second current to-be-processed service load data of the third service processing resource; determining the first current working load information according to the first historical service processing duration information and the first current service load data to be processed; and determining the second current working load information according to the second historical service processing duration information and the second current service load data to be processed.

In the embodiment of the present disclosure, the same weight may be set for the first historical service processing duration information and the first current to-be-processed service load data, and the first historical service processing duration information and the first current to-be-processed service load data are subjected to weighted summation to determine the first current workload information. And setting the same weight for the second historical service processing time length information and the second current service load data to be processed, carrying out weighted summation on the second historical service processing time length information and the second current service load data to be processed, and determining the second current work load information. However, the calculation manner of the first current workload information and the second current workload information in the embodiment of the present disclosure is not limited thereto.

In step S330, first network connection information between the second service processing resource and the first proxy service node and second network connection information between the third service processing resource and the second proxy service node are obtained.

In an exemplary embodiment, the first network connection information may include a first network distance, and the second network connection information may include a second network distance. The obtaining first network connection information between the second service processing resource and the first proxy service node, and the second network connection information between the third service processing resource and the second proxy service node may include: respectively sending target network requests to the second service processing resource and the third service processing resource; receiving target network responses generated by the second service processing resource and the third service processing resource responding to the target network request respectively; obtaining a first network distance between the second business processing resource and the first proxy service node according to the target network request and the target network response of the second business processing resource; and obtaining a second network distance between the third business processing resource and the first proxy service node according to the target network request and the target network response of the third business processing resource.

In an exemplary embodiment, obtaining the first network distance between the second traffic processing resource and the first proxy service node according to the target network request and the target network response of the second traffic processing resource may include: acquiring first network time consumption, first network delay and first network survival time between the second service processing resource and the first proxy service node according to the target network request and the target network response of the second service processing resource; obtaining the first network survival time mapping to obtain a first routing hop count; acquiring a first weight of the first routing hop count, a second weight of the first network time consumption and a third weight of the first network time delay; and obtaining the first network distance according to the first routing hop count and the first weight thereof, the first network time consumption and the second weight thereof, and the first network delay and the third weight thereof.

In the disclosed embodiment, the first weight may be greater than the second weight and the third weight.

In an exemplary embodiment, the second weight may be equal to the third weight.

In an exemplary embodiment, obtaining the second network distance between the third traffic processing resource and the first proxy service node according to the target network request and the target network response of the third traffic processing resource may include: acquiring second network consumed time, second network delay and second network survival time between the third business processing resource and the first proxy service node according to the target network request and the target network response of the third business processing resource; obtaining the second network survival time mapping to obtain a second routing hop count; acquiring a fourth weight of the second routing hop count, a fifth weight of the second network time consumption and a sixth weight of the second network time delay; and obtaining the second network distance according to the second routing hop count and the fourth weight thereof, the second network time consumption and the fifth weight thereof, and the second network delay and the sixth weight thereof. The second network distance is calculated in a similar manner to the first network distance.

In the embodiment of the present disclosure, after the proxy service node shown in fig. 1 and 2 is abstracted, as a policy and a method for scheduling, when a certain service generation place (taking a first service generation place as an example) generates a new service load, for example, when a security inspection device of a certain security inspection point generates a new security inspection image and needs to perform intelligent identification to determine whether the new security inspection image contains a suspected object, a service processing resource with an intelligent identification service function may be called as needed by using the proxy service node. For example, the first proxy service node may schedule the newly generated traffic load according to the network connection status and the workload status of the currently connectable traffic processing resource.

Specifically, the scheduling policy may include:

(1) the first proxy service node corresponding to the first traffic generation site generating the current traffic load sends a target network request (a PING request is taken as an example for illustration, but the disclosure is not limited thereto) to all the traffic processing resources which are set to be capable of being allocated with the traffic load, so as to determine the network connection state between each traffic processing resource and the first proxy service node. And judging which service processing resources are normally connected according to whether the target network response returned by each service processing resource can be received or not. For the service processing resources which can be normally connected, calculating the network distance between the first proxy service node and each service processing resource, for example, the first network distance between the first proxy service node and the second service processing resource, and the second network distance between the first proxy service node and the third service processing resource.

In other embodiments, the target network request may also be sent to all the traffic processing resources to which the traffic load may be allocated by the first traffic generation device corresponding to the first traffic generation location that generates the current traffic load, so as to determine the network connection state between each traffic processing resource and the first traffic generation location. And judging which service processing resources are normally connected according to whether the target network response returned by each service processing resource can be received or not. For the service processing resources which can be normally connected, calculating the network distance between the first service generating place and each service processing resource, for example, the first network distance between the first service generating place and the second service processing resource, and the second network distance between the first service generating place and the third service processing resource.

In the embodiment of the disclosure, the relative network distance between the service generating place and each service processing resource can be measured by PING time consumption and delay between the service generating place and the service processing resource and the routing hop number obtained by mapping PING survival time.

In the embodiment of the present disclosure, when the current service load is generated, the service processing resource with lower time consumption and delay and fewer route hops corresponding to PING is obtained, and compared with other service processing resources, the priority of the current service load is higher without considering other factors.

For example, also taking the first network distance as an example, a first weight of 2 may be set for the first routing hop count, and a second weight and a third weight of 1 may be set for both the first network time consumption and the first network delay. The first routing hop count, the first network time consumption and the first network delay can be weighted and summed according to the first weight, the second weight and the third weight, and the first network distance is obtained through calculation.

It is understood that the network connection information is not limited to the network distance, but may also be one or more other information such as the network transmission rate, the network stability condition, etc., and the network distance is merely an example.

It should be noted that the manner of obtaining the network distance is not limited to the calculation using PING, and other service or electronic physical manners may be adopted to obtain the network distance. The embodiment of the present disclosure does not limit the calculation method of the network distance.

(2) For the service processing resources which can be normally connected in the step (1), the current to-be-processed service load data which are currently processed by the service processing resources and the time information consumed by the service processing resources for processing historical service loads in the past are obtained to calculate and measure the distribution priority of the current service load.

The current to-be-processed service load data may include first current to-be-processed service load data of a second service processing resource and second current to-be-processed service load data of a third service processing resource. The current to-be-processed traffic load data may be, for example, the current to-be-processed traffic load number, but the present disclosure is not limited thereto, and for example, the current to-be-processed traffic load data may also be one data obtained by calculation by comprehensively considering the current to-be-processed traffic load number of each traffic processing resource and the resource required by each current to-be-processed traffic load data.

The time information consumed by the past processing of the historical service load by each service processing resource can be referred to as historical processing time information (including first historical processing time information and second historical processing time information), and can include the receiving time when the service processing resource receives each historical service load, the starting time when the service processing resource starts to process each service load, and the ending time when each service load is processed, and the historical service processing time length information of each historical service load (including first historical service load and second historical service load) of each service processing resource can be calculated according to the time information.

In the embodiment of the present disclosure, the arithmetic mean value of the historical service processing duration information of each historical service load of each service processing resource may be calculated as the historical service processing duration information of each service processing resource (including the first historical service processing duration information and the second historical service processing duration information), but the present disclosure is not limited thereto.

For example, the smaller the number of the current pending traffic load being processed, the smaller the historical traffic processing duration information consumed by the past processing, and the higher the priority without considering other factors.

Generally, as long as a service processing resource is locally allocated to a service generation site and the service processing resource has a computational margin, a traffic load locally generated by the service generation site is allocated to the local service processing resource. For example, the current traffic load collected by the first traffic generation site is preferentially allocated to the local first traffic processing resource.

However, if the local service processing resource is fully loaded, other service processing resources (for example, service processing resources that can be normally connected to the first service and do not belong to the local service processing resource) may be sorted in descending order according to the current workload information (including the first current workload information and the second current workload information), if two or more service processing resources have the same current workload information, then sorting according to the network connection information (including the first network connection information and the second network connection information) in descending order, preferentially allocating the current service load to at least one service processing resource with high sorting order as a target service processing resource (target service processing resource), for example, the first service processing resource is ranked as the target service processing resource, or the first service processing resources are ranked as the target service processing resources. I.e. the priority of the current workload information factor is higher than the priority of the network connection information factor.

If the current workload information and the network connection information of two or more service processing resources are the same, the two or more service processing resources may be randomly allocated, or the two or more service processing resources may be simultaneously used as target service processing resources, and the current service load may be simultaneously allocated to the multiple target service processing resources for parallel processing.

It should be noted that, when determining the allocation priority of the current service load according to the current workload information (including the first current workload information and the second current workload information) of the service processing resource, not only the above-described factors of the current to-be-processed service load data (including the first current to-be-processed service load data and the second current to-be-processed service load data) and the historical service processing duration information, but also other different measurement manners may be adopted. For example, only a single factor of current pending traffic load data or historical traffic processing duration information may be considered. Alternatively, only the current amount of traffic load to be processed may be considered.

In step S340, first historical service data of the second service processing resource is obtained, and second historical service data of the third service processing resource is obtained.

In an exemplary embodiment, the first historical traffic data may include first historical traffic load generation data of a second traffic generation device and first historical traffic load allocation data of a second traffic processing resource corresponding to the second traffic processing resource, and the second historical traffic data may include second historical traffic load generation data of a third traffic generation device and second historical traffic load allocation data of a third traffic processing resource corresponding to the third traffic processing resource.

In the embodiment of the present disclosure, the historical traffic load generation data (including the first historical traffic load generation data and the second historical traffic load generation data) refers to information related to the historical traffic load generated or collected by the traffic generation device or the traffic generation location corresponding to the traffic processing resource before the current time, for example, if the second traffic processing resource corresponds to the second traffic generation device and the second traffic generation location, the second traffic generation device transmits the generated or collected historical traffic load to the second traffic processing resource, and the second traffic processing resource receives the distributed historical traffic load as the first historical traffic load generation data of the second traffic processing resource. For another example, if the third service processing resource corresponds to a third service generation device and a third service generation site, the third service generation device sends the generated or collected historical service load to the third service processing resource, and the third service processing resource receives the allocated historical service load and uses the historical service load as second historical service load generation data of the third service processing resource. In the embodiment of the present disclosure, the historical service load before the current time and in a period of time closest to the current time may be selected. The historical traffic load distribution data (including the first historical traffic load distribution data and the second historical traffic load distribution data) refers to information about historical traffic loads that the traffic processing resources are distributed to need to process before the current time.

Wherein the method may further comprise: according to the first historical service load generation data, obtaining a first generated service load processing probability of a second service generation device corresponding to the second service processing resource in a first historical time period before the current time and a first generated service load probability of the second service generation device corresponding to the second service processing resource in a first future time period after the current time; obtaining a probability of a traffic load occurring in a first future period of the second traffic processing resource according to the first historical traffic load distribution data, the first generated traffic load processing probability, and the first to-be-generated traffic load probability; according to the second historical service load generation data, obtaining a second generated service load processing probability of a third service generation device corresponding to the third service processing resource in the first historical time period and a second generated service load probability of the third service generation device corresponding to the third service processing resource in the first future time period; and obtaining the probability of the service load occurring in the second future period of the third service processing resource according to the second historical service load distribution data, the second generated service load processing probability and the second to-be-generated service load probability.

In the embodiment of the present disclosure, as a policy and a method for scheduling after a proxy service node is abstracted, when a certain service generation place (for example, a first service generation place) generates a new service load (current service load), service processing resources may be called as needed, and the newly generated service load may be scheduled according to past historical service load generation data (including first historical service load generation data and second historical service load generation data) and historical service load distribution data (including first historical service load distribution data and second historical service load distribution data) of each service generation place.

Specifically, in general, each service occurrence location is configured with a service processing resource, and according to the current workload information and the calculation method of the network distance, the range of the service processing resources to which the service load generated by a certain service occurrence location can be allocated is obtained, and the service processing resources configured by the local service occurrence location, the adjacent service occurrence location and the service occurrence location near the local service occurrence location and the adjacent service occurrence location will be concentrated with a high probability. Therefore, since the generation rule of the traffic load generated for each traffic generation place can be known from the historical traffic load generation data in the historical traffic data, when a new traffic load is generated at the current time t, for a certain traffic generation place, the probability that the traffic load has been generated and is being distributed or intelligently identified in the first historical time period before the current time (for example, [ t- Δ t1, t ]) of the adjacent and nearby traffic generation places (for example, the second traffic generation place and the third traffic generation place of the first traffic generation place) (referred to as the generated traffic load processing probability (including the first generated traffic load processing probability and the second generated traffic load processing probability), and the probability that the new traffic load will be generated in the first future time period after the current time (for example, [ t, t + Δ t2 ]) of the adjacent and nearby traffic generation places (referred to as the traffic load generation probability (packet load generation probability) Including a first to-be-generated traffic load probability and a second to-be-generated traffic load probability).

According to the historical traffic load distribution data in the historical traffic data, the distribution rule of the traffic load generated by each traffic place to each traffic processing resource can be known, so that for a certain traffic place, when a new traffic load is generated at the current time t, the probability that the traffic load generated or to be generated by the traffic generation equipment or the traffic place corresponding to each traffic processing resource is distributed to the traffic processing resources configured by the adjacent and nearby traffic places can be predicted according to the generated traffic load processing probability and the traffic load probability to be generated of the adjacent and nearby traffic places, and the probability is called the traffic load probability in the future period (including the traffic load probability in the first future period and the traffic load probability in the second future period).

The generated traffic load processing probability, the probability of traffic load to be generated, and the probability of traffic load occurring in a future time period in the above embodiments may be obtained by record statistics in a system database or a log.

For example, when each traffic load is generated, the database records the generation time, the generation location (which traffic generation device collects), the distribution time, the distribution location (to which traffic processing resource is distributed), and so on, and the analysis service can count up the data.

The embodiment of the present disclosure uses the generation and distribution rules obtained according to the historical traffic load generation data and the historical traffic load distribution data of the adjacent and nearby traffic generation sites to perform modeling analysis, but what kind of rules are used in the determination process of the scheduling policy, and how to use the rules are not limited to the above example.

In step S350, a target service processing resource is determined from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical service data and the second historical service data.

In step S360, the current service load is allocated to the target service processing resource, so that the target service processing resource processes the current service load.

As shown in fig. 4, the embodiment of the present disclosure may perform modeling analysis to predict the optimal allocation of real-time traffic load in combination with the real-time workload of the traffic processing resource (e.g., the first current workload information and the second current workload information), the network connection information between the traffic generation site and the traffic processing resource (e.g., the first network distance and the second network distance determined according to the real-time first routing hop count and the second routing hop count, etc.), the historical traffic load generation data of the traffic generation site, and the historical traffic load allocation data (factors of other dimensions may also be considered).

In an exemplary embodiment, determining a target traffic processing resource from the second traffic processing resource and the third traffic processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical traffic data and the second historical traffic data may include: comparing the first to-be-generated traffic load probability with the second to-be-generated traffic load probability if the first current workload information is superior to the second current workload information, or if the first current workload information is equal to the second current workload information and the first network distance is less than or equal to the second network distance; and if the first probability of generating the service load is greater than the second probability of generating the service load, determining the third service processing resource as the target service processing resource.

In the embodiment of the present disclosure, the first current workload information being better than the second current workload information may include, but is not limited to, any one of the following cases: the first current workload quantity of the second service processing resource is less than the second current workload quantity of the third service processing resource, for example, the current to-be-processed service loads of the second service processing resource are 1, and the current to-be-processed service loads of the third service processing resource are 10; although the first current workload quantity of the second service processing resource is more than the second current workload quantity of the third service processing resource, the current service load to be processed of the second service processing resource is simpler, and the occupied processing resource is less than that of the current service load to be processed of the third service processing resource; although the number of the first current workloads of the second service processing resources is greater than that of the second current workloads of the third service processing resources, and the occupied processing resources are more than the processing resources occupied by the service loads to be processed currently by the third service processing resources, the second service processing resources can process more service loads in parallel because the processing resources owned by the second service processing resources are more than the processing resources owned by the third service processing resources, and so on.

For example, as shown in fig. 5, or taking a security inspection scene as an example, a first service generation device at a first service generation site collects a frame of security inspection image at the current time, and it is assumed that the frame of security inspection image is divided into three image segments, which correspond to a service load 1, a service load 2, and a service load 3. And assuming that the residual computing power of the first AI monomer of the first service place can only process the service load 3 and cannot process the service load 1 and the service load 2, taking the service load 1 and the service load 2 as the current service load.

And judging the most appropriate AI monomer as a second AI monomer of a second service occurrence place adjacent to the first service occurrence place according to the comparison of the first current working load information, the second current working load information, the first network distance and the second network distance. However, since the second service generation devices of the adjacent second service generation sites are also scanning concurrently, security inspection images required to be intelligently identified are synchronously generated and are also distributed to the second AI monomer for intelligent identification.

According to the first historical service data of the second AI monomer, the probability that the first to-be-generated service load of a second service generating place about to generate a new service load is higher, and the probability that the service load occurs in a first future time period is also higher, so that after model learning and analysis and comprehensive consideration, the service load 1 and the service load 2 can be sent to a third AI monomer of a third service generating place adjacent to the first service generating place. Although the network distance of the third AI monomer is not as close as that of the second AI monomer, the probability that a new service load will be generated at the place where the third service corresponding to the third AI monomer is generated is lower, and the intelligent identification work of only one service load is supposed to be currently in progress but will be completed soon.

In the embodiment of the disclosure, when the resource is scheduled, modeling and learning are performed based on the multidimensional influence factor, and the target service processing resource with the most reasonable current service load is analyzed and distributed.

In the resource scheduling method provided in this disclosure, when the service processing resources are distributed to different physical locations, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device. The method comprises the steps of abstracting a functional elastic layer from business processing resources, wherein the functional elastic layer is called as a proxy service node, each proxy service node corresponds to at least one business processing resource, the proxy service node has the capacity of forming a cluster through local self-organization, one or more business processing resources can be called as required, the edge cooperation effect is achieved, and the weak capacity of a single business processing resource is converted into the strong capacity with group addition (group intelligence formed by group relation and system). Taking the first proxy service node as an example, assuming that the first proxy service node corresponds to the first service processing resource, the second service processing resource and the third service processing resource, if the first service processing resource does not have enough computing power to process the current service load generated by the corresponding first service generation device at the current time, the first proxy service node may select a target service processing resource suitable for processing the current service load according to the first current workload information, the first network connection information and the first historical service data of the second service processing resource, and the second current workload information, the second network connection information and the second historical service data of the third service processing resource, on one hand, it is avoided that a certain service processing resource cannot process a new service load due to overload, so that the new service load cannot be processed or can only be delayed to be processed, but reduces its business value or loses its business meaning. On the other hand, the scheduling strategy of the current workload information, the network connection information and the historical service data of each service processing resource is comprehensively considered, so that the optimal target service processing resource can be selected from a plurality of service processing resources, and the technical effect of optimal overall effect is achieved.

In an exemplary embodiment, the resource scheduling system may include a plurality of agent clusters, and interconnection between different agent clusters may be implemented through a cloud. Each agent cluster comprises a control node, and different agent clusters can realize resource scheduling through the control nodes. The control node may organize the service agent nodes in the local agent cluster, communicate with the control nodes in other agent clusters in the cloud, and may allocate the traffic load generated in one agent cluster to a certain or some traffic processing resources in another agent cluster for processing.

If only one proxy service node is contained in a certain proxy cluster, the proxy service node is the control node of the proxy cluster.

If a proxy cluster contains multiple (more than two) proxy service nodes, it can be randomly assigned or determined by election which proxy service node is the control node. And voting, namely voting all the proxy service nodes in the proxy cluster to select one of the proxy service nodes as a control node.

For example, as shown in fig. 6, the difference from the above embodiment of fig. 1 is that the resource scheduling system may further include a second proxy cluster, and the second proxy cluster may include at least one proxy service node, such as a third proxy service node, where the third proxy service node corresponds to a fourth service processing resource, and the fourth service processing resource corresponds to a fourth service generation device at a fourth service generation site. In fig. 6, it is assumed that the third proxy service node is a control node of the second proxy cluster, the first proxy service node is a control node of the first proxy cluster, each control node includes a respective control interface, and communication between the first proxy cluster and the second proxy cluster can be achieved through the control interface of the first proxy service node and the control interface of the third proxy service node.

It is understood that other proxy service nodes may also be included in the second proxy cluster and the first proxy cluster.

In the embodiment of fig. 6, the first service node is assumed to be a control node of the first proxy cluster, but the disclosure is not limited thereto. For example, as shown in fig. 7, the first proxy cluster may further include a fourth proxy service node, where the fourth proxy service node corresponds to a fifth service processing resource, and the fifth service processing resource corresponds to a fifth service generation device at a fifth service generation location. In the embodiment of fig. 7, it is assumed that the fourth proxy service node is a control node of the first proxy cluster, both the first proxy service node and the second proxy service node may communicate with the fourth proxy service node, and the communication between the first proxy cluster and the second proxy cluster may be implemented through a control interface of the fourth proxy service node and a control interface of the third proxy service node.

In an exemplary embodiment, the method may further include: if the service processing resources in the first agent cluster are all in the full working load state, the current service load is sent to the second agent cluster, so that the target service processing resource is determined from the service processing resources corresponding to at least one agent service node of the second agent cluster.

For example, as shown in fig. 6, if the first proxy service node is a control node, the first proxy service node may obtain current workload information of each service processing resource in the first proxy cluster, and if it is found that each service processing resource in the first proxy cluster is in a full workload state and cannot process the current service load in time according to the current workload information of each service processing resource in the first proxy cluster, the first proxy service node may send the current service load to a control interface of a third proxy service node of the second proxy cluster through the control interface. The service interface of the third proxy service node can acquire the current workload information, the network connection information and the historical service data of each service processing resource in the second proxy cluster, analyze which service processing resource in the second proxy cluster is most suitable as the target service processing resource of the current service load according to the current workload information, and distribute the current service load to the target service processing resource to perform service processing.

Fig. 8 schematically shows a flow chart of a resource scheduling method according to an embodiment of the present disclosure. The method provided in the embodiment of fig. 8 may be applied to a resource scheduling system, where the resource scheduling system may include a first proxy cluster, the first proxy cluster may include a first proxy service node, the first proxy service node may correspond to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource may correspond to a first service generation device, the second service processing resource may correspond to a second service generation device, and the third service processing resource may correspond to a third service generation device. Reference may be made to the description of the embodiments of fig. 1 and 2 above.

As shown in fig. 8, the method provided by the embodiment of the present disclosure may include the following steps.

In step S810, a current service load generated by the first service generation device at the current time is obtained.

In step S820, if the first service processing resource is in the first workload state at the current time, obtain first current workload information of the second service processing resource and second current workload information of the third service processing resource.

In step S830, first network connection information between the second service processing resource and the first proxy service node and second network connection information between the third service processing resource and the second proxy service node are obtained.

In step S840, a target service processing resource is determined from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, and the second network connection information.

In an exemplary embodiment, determining a target traffic processing resource from the second traffic processing resource and the third traffic processing resource according to the first current workload information, the second current workload information, the first network connection information, and the second network connection information may include: if the first current workload information is superior to the second current workload information, determining the second service processing resource as the target service processing resource; comparing the first network connection information with the second network connection information if the first current workload information is equal to the second current workload information; and if the first network connection information is smaller than the second network connection information, determining the second service processing resource as the target service processing resource.

For example, as shown in fig. 9, a first network distance is determined according to a first network consumed time, a first network delay and a first routing hop count of a second service occurrence location, a second network distance is determined according to a second network consumed time, a second network delay and a second routing hop count of a third service occurrence location, and a target service processing resource can be determined by comparing first current workload information of a second service processing resource of the second service occurrence location with second current workload information of the third service occurrence location, and the first network distance with the second network distance. Reference may be made specifically to the description of the embodiment of fig. 3 above regarding prioritization based on current workload information and network connection information.

In step S850, the current traffic load is allocated to the target traffic processing resource, so that the target traffic processing resource processes the current traffic load.

The resource scheduling method provided by the embodiment of the disclosure is used as a strategy and a method for scheduling after an agent service node is abstracted, when a new service load is generated in a service, for example, when a security inspection device generates a new security inspection image and needs to perform intelligent identification to judge whether the security inspection device contains a suspected object, the resource scheduling method can schedule and arrange the newly generated service load according to the network connection condition and the working load condition of the currently connectable service processing resource when the service processing resource is called as required.

Fig. 10 schematically shows a flow chart of a resource scheduling method according to an embodiment of the present disclosure. The method provided in the embodiment of fig. 10 may be applied to a resource scheduling system, where the resource scheduling system may include a first proxy cluster, the first proxy cluster may include a first proxy service node, the first proxy service node may correspond to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource may correspond to a first service generation device, the second service processing resource may correspond to a second service generation device, and the third service processing resource may correspond to a third service generation device. Reference may be made in particular to the description of the embodiment of figures 1 and 2 above.

As shown in fig. 10, the method provided by the embodiment of the present disclosure may include the following steps.

In step S1010, a current service load generated by the first service generating device at the current time is obtained.

In step S1020, if the first service processing resource is in the first workload state at the current time, obtain first historical service data of the second service processing resource, and obtain second historical service data of the third service processing resource.

In an exemplary embodiment, the first historical traffic data may include first historical traffic load generation data of a second traffic generation device corresponding to the second traffic processing resource and first historical traffic load allocation data of the second traffic processing resource, and the second historical traffic data may include second historical traffic load generation data of a third traffic generation device corresponding to the third traffic processing resource and second historical traffic load allocation data of the third traffic processing resource.

Wherein the method may further comprise: according to the first historical service load generation data, obtaining a first generated service load processing probability of a second service generation device corresponding to the second service processing resource in a first historical time period before the current time and a first generated service load probability of the second service generation device corresponding to the second service processing resource in a first future time period after the current time; obtaining a probability of a traffic load occurring in a first future period of the second traffic processing resource according to the first historical traffic load distribution data, the first generated traffic load processing probability, and the first to-be-generated traffic load probability; according to the second historical service load generation data, obtaining a second generated service load processing probability of a third service generation device corresponding to the third service processing resource in the first historical time period and a second generated service load probability of the third service generation device corresponding to the third service processing resource in a first future time period; and obtaining the probability of the service load occurring in the second future period of the third service processing resource according to the second historical service load distribution data, the second generated service load processing probability and the second to-be-generated service load probability.

In step S1030, a target service processing resource is determined from the second service processing resource and the third service processing resource according to the first historical service data and the second historical service data.

In an exemplary embodiment, determining a target traffic processing resource from the second traffic processing resource and the third traffic processing resource according to the first historical traffic data and the second historical traffic data may include: and if the first probability of generating the service load is greater than the second probability of generating the service load, determining the third service processing resource as the target service processing resource.

For example, as shown in fig. 11, a modeling analysis may be performed according to the historical traffic load generation data and the historical traffic load distribution data to predict the optimal distribution of real-time traffic load, so as to find the most suitable adjacent and nearby traffic processing resources for distribution when the traffic processing resources configured locally at the traffic generation site are fully loaded. Reference may be made to the description above in connection with the embodiment of fig. 3.

The generated traffic load processing probability (input value 1), the generated traffic load probability (input value 2), and the historical statistical data of the historical traffic load distribution data (input value 3) obtained in the above embodiment may be used as input values, and the target value may be the probability of the occurrence of the traffic load at a certain traffic place, that is, the probability of the occurrence of the traffic load in the future period. And forming a numerical prediction model which can be time sequence prediction or regression prediction or time sequence combination regression prediction.

For example, the accuracy of the predicted target value is evaluated by using a simple one-time moving average, where the predicted target value is (input value 1+ input value 2+ input value 3)/3, or a linear regression equation, where the predicted target value is f (input value), using a decision coefficient R to evaluate the model quality, and using indexes such as MAE (Mean Absolute Error), RMSE (Root Mean Square Error), MAPE (Mean Absolute Percentage Error), and the like.

And continuously inputting historical statistical data into the model, adjusting the model when the quality and the accuracy do not reach the standard, for example, changing to quadratic moving average or adjusting a regression equation, learning and training the model, and taking the model as a first-stage model when the quality and the accuracy reach the standard, wherein the model can be continuously adjusted along with data accumulation.

For example, taking a security inspection scene as an example, as shown in fig. 12, when a first service generation device generates a new security inspection image, the new security inspection image needs to be divided into three image segments for intelligent identification, which respectively need 3 times of interface calls of AI monomers corresponding to service loads 1-3, and a local first AI monomer cannot simultaneously process the service loads of 3 intelligent identification tasks. At this time, other second service generating devices and third service generating devices are also arranged near the first service generating device, so that the probability that the second AI monomer and the third AI monomer may have service loads concurrently, that is, the probability that the service load occurs in the first future period and the probability that the service load occurs in the second future period, can be predicted according to the method, and the AI monomer corresponding to the service generating device with low probability of generating the service load is selected as the target AI monomer to perform allocation of the intelligent identification task.

In step S1040, the current service load is allocated to the target service processing resource, so that the target service processing resource processes the current service load.

The resource scheduling method provided by the embodiment of the present disclosure, after abstracting the proxy service node, is used as a policy and method for scheduling, when a new service load is generated in a service, a service processing resource can be called as needed, and scheduling arrangement is performed on the newly generated service load according to historical service generation data generated by past service loads in a service generation place and historical service load distribution data distributed to corresponding service processing resources by past service loads.

In the following embodiments, a security inspection scene is taken as an example for illustration, in the security inspection scene, the service generation device may be a security inspection point or at least one security inspection device set at the security inspection point, and the service load collected by the service generation device is a security inspection image. The service processing resource for identifying the suspected object in the security inspection image is called intelligent identification equipment, and when at least part of security inspection points or security inspection equipment corresponds to each intelligent identification equipment one by one, the intelligent identification equipment is called single intelligent identification equipment (AI single for short). The proxy service node at this time is called an AI monolithic proxy service node.

Fig. 13 schematically shows a flowchart of a method for intelligently identifying resource scheduling according to an embodiment of the present disclosure. The method provided in the embodiment of fig. 13 may be applied to an intelligent resource scheduling system, where the intelligent resource scheduling system may include a first agent cluster, the first agent cluster may include a first single intelligent identification device agent service node, the first single intelligent identification device agent service node corresponds to a first single intelligent identification device, a second single intelligent identification device, and a third single intelligent identification device, the first single intelligent identification device corresponds to a first security inspection device, the second single intelligent identification device corresponds to a second security inspection device, and the third single intelligent identification device corresponds to a third security inspection device.

As shown in fig. 13, the method provided by the embodiment of the present disclosure may include the following steps.

In step S1310, a security inspection image acquired by the first security inspection device at the current time is acquired.

In step S1320, a current traffic load is generated according to the security check image.

In step S1330, if the first single intelligent recognition device is in the full workload state at the current time, obtain first current workload information of the second single intelligent recognition device and second current workload information of the third service processing resource.

In step S1340, first network connection information between the second single intelligent recognition device and the first proxy service node and second network connection information between the third single intelligent recognition device and the second proxy service node are obtained.

In step S1350, first historical service data of the second single intelligent identification device is obtained, and second historical service data of the third single intelligent identification device is obtained.

In step S1360, a target individual smart identification device is determined from the second individual smart identification device and the third individual smart identification device according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical service data and the second historical service data.

In step S1370, the current service load is distributed to the target individual intelligent identification device, so that the target individual intelligent identification device identifies a suspected object in the current service load.

Specific implementations may refer to the description of the above embodiments.

On one hand, the intelligent identification resource scheduling method provided by the embodiment of the disclosure abstracts a functional elastic layer from an AI monomer, which is called as an AI monomer proxy service node. An AI single-body agent service node can correspond to a single-body intelligent identification device or a plurality of single-body intelligent identification devices to form an agent cluster, so that AI computing resources can be fully scheduled and utilized, and the weak capability of a single hardware device is converted into the strong capability with group addition. The AI single agent service node has the capability of forming a cluster by local self-organization, and can call hardware computing resources corresponding to one AI single agent service node or a plurality of AI single agent service nodes as required to achieve the effect of edge cooperation. On the other hand, whether the current traffic load should be sent to the AI monomer can be calculated according to the current network connection condition between the first traffic occurrence place and each AI monomer and the current workload of the AI monomer. And secondly, predicting the rule of the generation of the service load in a future service scene according to historical service data, including historical service load generation data of a service generation place and historical service load distribution data, and determining which AI monomer the service load is to be distributed to by influencing the strategy and logic of service load scheduling.

The intelligent identification resource scheduling method provided by the embodiment of the disclosure can also combine the network connection condition, the workload and the historical service data, and assist with influence factors of other dimensions, train and learn the data model through modeling analysis, and give the final distribution result of each service load.

According to the intelligent identification resource scheduling method provided by the embodiment of the disclosure, the single intelligent identification devices with weak computing power are logically gathered, and the computing power resources are scheduled as required, so that the defects that the single intelligent identification devices cannot cope with large-scale service integration and each single cannot cope with service load changes can be effectively overcome, and flexible and efficient intelligent identification capability is provided for services of customers. The method avoids the blocking of service development or the invalidation of service caused by resource limitation. And the cost brought by the original resources and the resource consumption is effectively controlled through the load balance of the resources. And finally, the service value of the client is improved.

The present disclosure is an intelligent recognition service application layer, and is formed based on a distributed cluster and machine learning technology, which is different from the learning and training of a bottom intelligent recognition algorithm and a recognition library.

Fig. 14 schematically shows a block diagram of a resource scheduling apparatus according to an embodiment of the present disclosure. The resource scheduling apparatus 1400 provided in the embodiment of fig. 14 may be applied to a resource scheduling system, where the resource scheduling system may include a first proxy cluster, the first proxy cluster may include a first proxy service node, the first proxy service node may correspond to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource may correspond to a first service generation device, the second service processing resource may correspond to a second service generation device, and the third service processing resource may correspond to a third service generation device.

As shown in fig. 14, a resource scheduling apparatus 1400 provided by the present disclosure may include: a current traffic load obtaining unit 1410, a workload information obtaining unit 1420, a network connection information obtaining unit 1430, a historical traffic data obtaining unit 1440, a target processing resource determining unit 1450, and a current traffic load allocating unit 1460.

The current traffic load obtaining unit 1410 may be configured to obtain a current traffic load generated by the first traffic generating device at the current time. The workload information obtaining unit 1420 may be configured to obtain first current workload information of the second service processing resource and second current workload information of the third service processing resource, if the first service processing resource is in a full workload state at the current time. The network connection information obtaining unit 1430 may be configured to obtain first network connection information between the second service processing resource and the first proxy service node, and obtain second network connection information between the third service processing resource and the second proxy service node. The historical service data obtaining unit 1440 may be configured to obtain first historical service data of the second service processing resource, and obtain second historical service data of the third service processing resource. The target processing resource determining unit 1450 may be configured to determine a target traffic processing resource from the second traffic processing resource and the third traffic processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical traffic data and the second historical traffic data. The current traffic load allocating unit 1460 may be configured to allocate the current traffic load to the target traffic processing resource, so that the target traffic processing resource processes the current traffic load.

In an exemplary embodiment, the workload information obtaining unit 1420 may include: a historical service load information obtaining unit, configured to obtain a first historical service load and first historical processing time information thereof processed by the second service processing resource, and a second historical service load and second historical processing time information thereof processed by the third service processing resource; a historical service duration information obtaining unit, configured to obtain first historical service processing duration information of the second service processing resource according to the first historical service load and the first historical processing time information thereof, and obtain second historical service processing duration information of the third service processing resource according to the second historical service load and the second historical processing time information thereof; a to-be-processed service load data obtaining unit, configured to obtain first current to-be-processed service load data of the second service processing resource and second current to-be-processed service load data of the third service processing resource; a first current workload determining unit, configured to determine the first current workload information according to the first historical service processing duration information and the first current to-be-processed service load data; the second current workload determining unit may be configured to determine the second current workload information according to the second historical service processing duration information and the second current to-be-processed service load data.

In an exemplary embodiment, the first network connection information may include a first network distance, and the second network connection information may include a second network distance. The network connection information obtaining unit 1430 may include: a target network request sending unit, configured to send a target network request to the second service processing resource and the third service processing resource respectively; a target network response receiving unit, configured to receive target network responses generated by the second service processing resource and the third service processing resource in response to the target network request, respectively; a first network distance obtaining unit, configured to obtain a first network distance between the second service processing resource and the first proxy service node according to a target network request and a target network response of the second service processing resource; the second network distance obtaining unit may be configured to obtain a second network distance between the third service processing resource and the first proxy service node according to the target network request and the target network response of the third service processing resource.

In an exemplary embodiment, the first network distance obtaining unit may include: a time-consuming delay time-to-live obtaining unit, configured to obtain, according to the target network request and the target network response of the second service processing resource, a first network time-consuming, a first network delay, and a first network time-to-live between the second service processing resource and the first proxy service node; a route hop count obtaining unit, configured to obtain the first network lifetime mapping to obtain a first route hop count; a weight obtaining unit, configured to obtain a first weight of the first routing hop count, a second weight of the first network time consumption, and a third weight of the first network delay; the first network distance determining unit may be configured to obtain the first network distance according to the first routing hop count and the first weight thereof, the first network time consumption and the second weight thereof, and the first network delay and the third weight thereof.

In an exemplary embodiment, the first historical traffic data may include first historical traffic load generation data of a second traffic generation device corresponding to the second traffic processing resource and first historical traffic load allocation data of the second traffic processing resource, and the second historical traffic data may include second historical traffic load generation data of a third traffic generation device corresponding to the third traffic processing resource and second historical traffic load allocation data of the third traffic processing resource. The resource scheduling apparatus 1400 may further include: a first traffic load probability obtaining unit, configured to obtain, according to the first historical traffic load generation data, a first generated traffic load processing probability of a first historical time period before the current time of a second traffic generation device corresponding to the second traffic processing resource, and a first to-be-generated traffic load probability of a first future time period after the current time of the second traffic generation device corresponding to the second traffic processing resource; a first occurrence traffic load probability obtaining unit, configured to obtain a first future period occurrence traffic load probability of the second traffic processing resource according to the first historical traffic load distribution data, the first generated traffic load processing probability, and the first to-be-generated traffic load probability; a second traffic load probability obtaining unit, configured to obtain, according to the second historical traffic load generation data, a second generated traffic load processing probability of a third traffic generation device corresponding to the third traffic processing resource in the first historical time period, and a second to-be-generated traffic load probability of the third traffic generation device corresponding to the third traffic processing resource in a first future time period; a second occurrence traffic load probability obtaining unit, configured to obtain a second future period occurrence traffic load probability of the third traffic processing resource according to the second historical traffic load distribution data, the second generated traffic load processing probability, and the second to-be-generated traffic load probability.

In an exemplary embodiment, the target processing resource determining unit 1450 may include: a traffic load probability comparing unit, configured to compare the first to-be-generated traffic load probability and the second to-be-generated traffic load probability if the first current workload information is better than the second current workload information, or if the first current workload information is equal to the second current workload information and the first network distance is less than or equal to the second network distance; a target processing resource determining unit, configured to determine that the third service processing resource is the target service processing resource if the first probability of generating a service load is greater than the second probability of generating a service load.

In an exemplary embodiment, the resource scheduling system may further include a second proxy cluster, which may include at least one proxy service node. The resource scheduling apparatus 1400 further includes: the current service load sending unit may be configured to send the current service load to the second proxy cluster if all the service processing resources in the first proxy cluster are in the full workload state, so as to determine the target service processing resource from the service processing resources corresponding to at least one proxy service node of the second proxy cluster.

The specific implementation of each unit in the resource scheduling apparatus provided in the embodiment of the present disclosure may refer to the content in the resource scheduling method, and is not described herein again.

In an aspect of the present disclosure, a resource scheduling apparatus is provided, where the resource scheduling apparatus provided in the present disclosure may be applied to a resource scheduling system, and the resource scheduling system may include a first proxy cluster, where the first proxy cluster may include a first proxy service node, the first proxy service node may correspond to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource may correspond to a first service generation device, the second service processing resource may correspond to a second service generation device, and the third service processing resource may correspond to a third service generation device.

The resource scheduling device provided by the embodiment of the present disclosure may include a current service load obtaining unit, a current work load obtaining unit, a current network connection obtaining unit, a target service processing resource selecting unit, and a current service allocation configuring unit.

The current service load obtaining unit may be configured to obtain a current service load generated by the first service generation device at the current time. The current workload obtaining unit may be configured to obtain first current workload information of the second service processing resource and second current workload information of the third service processing resource, if the first service processing resource is in the first workload state at the current time. The current network connection obtaining unit may be configured to obtain first network connection information between the second service processing resource and the first proxy service node, and obtain second network connection information between the third service processing resource and the second proxy service node. The target service processing resource selecting unit may be configured to determine a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, and the second network connection information. The current service allocation configuration unit may be configured to allocate the current service load to the target service processing resource, so that the target service processing resource processes the current service load.

In an exemplary embodiment, the target traffic processing resource selection unit may include: a workload priority allocation unit, configured to determine that the second service processing resource is the target service processing resource if the first current workload information is better than the second current workload information; a network condition allocating unit, configured to compare the first network connection information with the second network connection information if the first current workload information is equal to the second current workload information; the target service processing resource selecting unit may be configured to determine that the second service processing resource is the target service processing resource if the first network connection information is smaller than the second network connection information.

According to an aspect of the present disclosure, a resource scheduling apparatus is provided, where the resource scheduling apparatus provided in the present disclosure may be applied to a resource scheduling system, and the resource scheduling system may include a first proxy cluster, where the first proxy cluster may include a first proxy service node, the first proxy service node may correspond to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource may correspond to a first service generation device, the second service processing resource may correspond to a second service generation device, and the third service processing resource may correspond to a third service generation device.

The resource scheduling device provided by the embodiment of the disclosure may include a current service load determining unit, a historical service data obtaining unit, a target service processing resource distinguishing unit, and a current service load processing unit.

The current service load determining unit may be configured to obtain a current service load generated by the first service generation device at the current time. The historical service data obtaining unit may be configured to obtain first historical service data of the second service processing resource and obtain second historical service data of the third service processing resource, if the first service processing resource is in the first workload state at the current time. The target service processing resource distinguishing unit may be configured to determine a target service processing resource from the second service processing resource and the third service processing resource according to the first historical service data and the second historical service data. The current traffic load processing unit may be configured to allocate the current traffic load to the target traffic processing resource, so that the target traffic processing resource processes the current traffic load.

In an exemplary embodiment, the resource scheduling apparatus may further include: a first traffic load processing generation probability calculating unit, configured to obtain, according to the first historical traffic load generation data, a first generated traffic load processing probability of a first historical time period before the current time of a second traffic generation device corresponding to the second traffic processing resource, and a first to-be-generated traffic load probability of a first future time period after the current time of the second traffic generation device corresponding to the second traffic processing resource; a first future period occurrence traffic load probability obtaining unit, configured to obtain a first future period occurrence traffic load probability of the second traffic processing resource according to the first historical traffic load distribution data, the first generated traffic load processing probability, and the first to-be-generated traffic load probability; a second traffic load processing generation probability calculating unit, configured to obtain, according to the second historical traffic load generation data, a second generated traffic load processing probability of a third traffic generation device corresponding to the third traffic processing resource in the first historical time period, and a second to-be-generated traffic load probability of the third traffic generation device corresponding to the third traffic processing resource in a first future time period; a second future time period occurrence traffic load probability obtaining unit, configured to obtain a second future time period occurrence traffic load probability of the third traffic processing resource according to the second historical traffic load distribution data, the second generated traffic load processing probability, and the second to-be-generated traffic load probability.

In an exemplary embodiment, the target traffic processing resource discriminating unit may include: the target service processing resource selecting unit may be configured to determine that the third service processing resource is the target service processing resource if the first probability of generating a service load is greater than the second probability of generating a service load.

In an aspect of the present disclosure, an intelligent resource scheduling apparatus is provided, where the intelligent resource scheduling apparatus provided in the present disclosure may be applied to an intelligent resource scheduling system, the intelligent resource scheduling system may include a first agent cluster, the first agent cluster may include a first single intelligent identification device agent service node, the first single intelligent identification device agent service node may correspond to a first single intelligent identification device, a second single intelligent identification device, and a third single intelligent identification device, the first single intelligent identification device may correspond to a first security inspection device, the second single intelligent identification device may correspond to a second security inspection device, and the third single intelligent identification device may correspond to a third security inspection device.

The intelligent identification resource scheduling device provided by the embodiment of the disclosure may include: the system comprises a security inspection image acquisition unit, a current service load generation unit, a current working load information extraction unit, a current network condition extraction unit, a historical service data extraction unit, a target single intelligent identification equipment determination unit and a suspected object identification unit.

The security check image acquisition unit may be configured to acquire a security check image acquired by the first security check device at the current moment. The current service load generating unit may be configured to generate a current service load according to the security check image. The current workload information extraction unit may be configured to obtain first current workload information of the second individual intelligent recognition device and second current workload information of the third service processing resource, if the first individual intelligent recognition device is in a full workload state at the current time. The current network condition extracting unit may be configured to acquire first network connection information between the second single intelligent recognition device and the first proxy service node, and acquire second network connection information between the third single intelligent recognition device and the second proxy service node. The historical service data extracting unit may be configured to obtain first historical service data of the second single intelligent identification device, and obtain second historical service data of the third single intelligent identification device. The target individual intelligent identification device determination unit may be configured to determine the target individual intelligent identification device from the second individual intelligent identification device and the third individual intelligent identification device according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical service data and the second historical service data. The suspected object identification unit may be configured to allocate the current service load to the target individual intelligent identification device, so that the target individual intelligent identification device identifies a suspected object in the current service load.

The specific implementation of each unit in the intelligent identification resource scheduling apparatus provided in the embodiment of the present disclosure may refer to the content in the resource scheduling method, and is not described herein again.

In an aspect of the present disclosure, a resource scheduling system may include a first proxy cluster, where the first proxy cluster may include a first proxy service node, the first proxy service node may correspond to a first service processing resource, a second service processing resource, and a third service processing resource, the first service processing resource may correspond to a first service generation device, the second service processing resource may correspond to a second service generation device, and the third service processing resource may correspond to a third service generation device.

The first proxy service node may be configured to obtain, from the first service generation device, a current service load generated at a current time; when the first service processing resource is in a full working load state at the current moment, acquiring first current working load information of the second service processing resource and second current working load information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; obtaining first historical service data of the second service processing resource and obtaining second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; and distributing the current service load to the target service processing resource. The target traffic processing resource may be operable to receive and process the current traffic load from the first proxy service node.

In an exemplary embodiment, the first proxy cluster may further include a second proxy service node, and the second proxy service node may be connected with the third traffic processing resource, the third traffic generating device, and the first proxy service node, respectively.

In an exemplary embodiment, the system may further include a second proxy cluster, which may include at least one proxy service node, and a control node may exist in the proxy service node of the first proxy cluster.

The control node of the first proxy cluster may be configured to send the current traffic load to the second proxy cluster when the traffic processing resources corresponding to each proxy service node in the first proxy cluster are all in the full workload state, so as to determine the target traffic processing resource from the traffic processing resources corresponding to at least one proxy service node of the second proxy cluster.

It should be noted that although in the above detailed description several units of the device for action execution are mentioned, this division is not mandatory. Indeed, the features and functions of two or more units described above may be embodied in one unit, in accordance with embodiments of the present disclosure. Conversely, the features and functions of one unit described above may be further divided into embodiments by a plurality of units.

Reference is now made to fig. 15, which illustrates a schematic diagram of an electronic device suitable for use in implementing embodiments of the present application. The electronic device shown in fig. 15 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present application.

Referring to fig. 15, an electronic device provided in an embodiment of the present disclosure may include: a processor 1501, a communication interface 1502, memory 1503, and a communication bus 1504.

Wherein the processor 1501, the communication interface 1502 and the memory 1503 communicate with each other via a communication bus 1504.

Alternatively, the communication interface 1502 may be an interface of a communication module, such as an interface of a GSM (Global System for Mobile communications) module. Processor 1501 is used to execute programs. The memory 1503 is used for storing programs. The program may comprise a computer program comprising computer operating instructions.

The processor 1501 may be a central processing unit CPU, or an application Specific Integrated circuit asic, or one or more Integrated circuits configured to implement embodiments of the present disclosure.

The memory 1503 may include a Random Access Memory (RAM) memory, and may further include a non-volatile memory (e.g., at least one disk memory).

Among them, the procedure can be specifically used for: acquiring a current service load generated by the first service generation equipment at the current moment; if the first service processing resource is in a full workload state at the current moment, acquiring first current workload information of the second service processing resource and second current workload information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; obtaining first historical service data of the second service processing resource and obtaining second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

Alternatively, the program may be specifically for: acquiring a current service load generated by the first service generation equipment at the current moment; if the first service processing resource is in a first working load state at the current moment, obtaining first current working load information of the second service processing resource and second current working load information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information and the second network connection information; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

Alternatively, the program may be specifically for: acquiring a current service load generated by the first service generation equipment at the current moment; if the first service processing resource is in a first working load state at the current moment, acquiring first historical service data of the second service processing resource and acquiring second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first historical service data and the second historical service data; and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

Alternatively, the program may be specifically for: acquiring a security inspection image acquired by the first security inspection equipment at the current moment; generating a current service load according to the security check image; if the first single intelligent identification device is in a full workload state at the current moment, acquiring first current workload information of the second single intelligent identification device and second current workload information of the third service processing resource; acquiring first network connection information between the second single intelligent identification device and the first proxy service node, and acquiring second network connection information between the third single intelligent identification device and the second proxy service node; acquiring first historical service data of the second single intelligent identification device, and acquiring second historical service data of the third single intelligent identification device; determining target single intelligent identification equipment from the second single intelligent identification equipment and the third single intelligent identification equipment according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; and distributing the current service load to the target single intelligent identification equipment so as to enable the target single intelligent identification equipment to identify the suspected object in the current service load.

In one aspect of the disclosed embodiments, a computer-readable storage medium is provided, which stores computer-executable instructions for performing the methods provided in the various alternative implementations of the embodiments described above.

According to an aspect of the application, a computer program product or computer program is provided, comprising computer instructions, the computer instructions being stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the method provided in the various alternative implementations of the embodiments described above.

It is to be understood that any number of elements in the drawings of the present disclosure are by way of example and not by way of limitation, and any nomenclature is used for differentiation only and not by way of limitation.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A resource scheduling method is characterized in that the method is applied to a resource scheduling system, the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device; wherein the method comprises the following steps:

acquiring a current service load generated by the first service generation equipment at the current moment;

if the first service processing resource is in a full workload state at the current moment, acquiring first current workload information of the second service processing resource and second current workload information of the third service processing resource;

acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node;

obtaining first historical service data of the second service processing resource and obtaining second historical service data of the third service processing resource;

determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data;

and distributing the current service load to the target service processing resource so that the target service processing resource processes the current service load.

2. The method of claim 1, wherein obtaining first current workload information for the second traffic processing resource and second current workload information for the third traffic processing resource comprises:

acquiring a first historical service load processed by the second service processing resource and first historical processing time information thereof, and a second historical service load processed by the third service processing resource and second historical processing time information thereof;

acquiring first historical service processing duration information of the second service processing resource according to the first historical service load and the first historical processing time information thereof, and acquiring second historical service processing duration information of the third service processing resource according to the second historical service load and the second historical processing time information thereof;

obtaining first current to-be-processed service load data of the second service processing resource and second current to-be-processed service load data of the third service processing resource;

determining the first current working load information according to the first historical service processing duration information and the first current service load data to be processed;

and determining the second current working load information according to the second historical service processing duration information and the second current service load data to be processed.

3. The method of claim 2, wherein the first network connection information comprises a first network distance and the second network connection information comprises a second network distance; wherein obtaining first network connection information between the second service processing resource and the first proxy service node, and obtaining second network connection information between the third service processing resource and the second proxy service node includes:

respectively sending target network requests to the second service processing resource and the third service processing resource;

receiving target network responses generated by the second service processing resource and the third service processing resource responding to the target network request respectively;

obtaining a first network distance between the second business processing resource and the first proxy service node according to the target network request and the target network response of the second business processing resource;

and obtaining a second network distance between the third business processing resource and the first proxy service node according to the target network request and the target network response of the third business processing resource.

4. The method of claim 3, wherein obtaining the first network distance between the second traffic processing resource and the first proxy service node based on the target network request and the target network response of the second traffic processing resource comprises:

acquiring first network time consumption, first network delay and first network survival time between the second service processing resource and the first proxy service node according to the target network request and the target network response of the second service processing resource;

obtaining the first network survival time mapping to obtain a first routing hop count;

acquiring a first weight of the first routing hop count, a second weight of the first network time consumption and a third weight of the first network time delay;

and obtaining the first network distance according to the first routing hop count and the first weight thereof, the first network time consumption and the second weight thereof, and the first network delay and the third weight thereof.

5. The method of claim 3, wherein the first historical traffic data comprises first historical traffic load generation data of a second traffic generation device corresponding to the second traffic processing resource and first historical traffic load allocation data of the second traffic processing resource, and the second historical traffic data comprises second historical traffic load generation data of a third traffic generation device corresponding to the third traffic processing resource and second historical traffic load allocation data of the third traffic processing resource; wherein the method further comprises:

according to the first historical service load generation data, obtaining a first generated service load processing probability of a second service generation device corresponding to the second service processing resource in a first historical time period before the current time and a first generated service load probability of the second service generation device corresponding to the second service processing resource in a first future time period after the current time;

obtaining a probability of a traffic load occurring in a first future period of the second traffic processing resource according to the first historical traffic load distribution data, the first generated traffic load processing probability, and the first to-be-generated traffic load probability;

according to the second historical service load generation data, obtaining a second generated service load processing probability of a third service generation device corresponding to the third service processing resource in the first historical time period and a second generated service load probability of the third service generation device corresponding to the third service processing resource in the first future time period;

and obtaining the probability of the service load occurring in the second future period of the third service processing resource according to the second historical service load distribution data, the second generated service load processing probability and the second to-be-generated service load probability.

6. The method of claim 5, wherein determining a target traffic processing resource from the second traffic processing resource and the third traffic processing resource based on the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical traffic data and the second historical traffic data comprises:

comparing the first to-be-generated traffic load probability with the second to-be-generated traffic load probability if the first current workload information is superior to the second current workload information, or if the first current workload information is equal to the second current workload information and the first network distance is less than or equal to the second network distance;

and if the first probability of generating the service load is greater than the second probability of generating the service load, determining the third service processing resource as the target service processing resource.

7. The method of claim 1, wherein the resource scheduling system further comprises a second proxy cluster, the second proxy cluster comprising at least one proxy service node; wherein the method further comprises:

if the service processing resources in the first agent cluster are all in the full working load state, the current service load is sent to the second agent cluster, so that the target service processing resource is determined from the service processing resources corresponding to at least one agent service node of the second agent cluster.

8. A resource scheduling method is characterized in that the method is applied to a resource scheduling system, the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device; wherein the method comprises the following steps:

if the first service processing resource is in a first working load state at the current moment, obtaining first current working load information of the second service processing resource and second current working load information of the third service processing resource;

determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information and the second network connection information;

9. The method of claim 8, wherein determining a target traffic processing resource from the second traffic processing resource and the third traffic processing resource based on the first current workload information, the second current workload information, the first network connection information, and the second network connection information comprises:

if the first current workload information is superior to the second current workload information, determining the second service processing resource as the target service processing resource;

comparing the first network connection information with the second network connection information if the first current workload information is equal to the second current workload information;

and if the first network connection information is smaller than the second network connection information, determining the second service processing resource as the target service processing resource.

10. A resource scheduling method is characterized in that the method is applied to a resource scheduling system, the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device; wherein the method comprises the following steps:

if the first service processing resource is in a first working load state at the current moment, acquiring first historical service data of the second service processing resource and acquiring second historical service data of the third service processing resource;

determining a target service processing resource from the second service processing resource and the third service processing resource according to the first historical service data and the second historical service data;

11. The method of claim 10, wherein the first historical traffic data comprises first historical traffic load generation data of a second traffic generation device corresponding to the second traffic processing resource and first historical traffic load allocation data of the second traffic processing resource, and the second historical traffic data comprises second historical traffic load generation data of a third traffic generation device corresponding to the third traffic processing resource and second historical traffic load allocation data of the third traffic processing resource; wherein the method further comprises:

12. The method of claim 11, wherein determining a target traffic processing resource from the second traffic processing resource and the third traffic processing resource based on the first historical traffic data and the second historical traffic data comprises:

13. An intelligent identification resource scheduling method is applied to an intelligent identification resource scheduling system, the intelligent identification resource scheduling system comprises a first agent cluster, the first agent cluster comprises a first single intelligent identification device agent service node, the first single intelligent identification device agent service node corresponds to a first single intelligent identification device, a second single intelligent identification device and a third single intelligent identification device, the first single intelligent identification device corresponds to a first security inspection device, the second single intelligent identification device corresponds to a second security inspection device, and the third single intelligent identification device corresponds to a third security inspection device; wherein the method comprises the following steps:

acquiring a security inspection image acquired by the first security inspection equipment at the current moment;

generating a current service load according to the security check image;

if the first single intelligent identification device is in a full workload state at the current moment, acquiring first current workload information of the second single intelligent identification device and second current workload information of the third service processing resource;

acquiring first network connection information between the second single intelligent identification device and the first proxy service node, and acquiring second network connection information between the third single intelligent identification device and the second proxy service node;

acquiring first historical service data of the second single intelligent identification device, and acquiring second historical service data of the third single intelligent identification device;

determining target single intelligent identification equipment from the second single intelligent identification equipment and the third single intelligent identification equipment according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data;

and distributing the current service load to the target single intelligent identification equipment so as to enable the target single intelligent identification equipment to identify the suspected object in the current service load.

14. A resource scheduling device is applied to a resource scheduling system, the resource scheduling system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device; wherein the apparatus comprises:

a current service load obtaining unit, configured to obtain a current service load generated by the first service generation device at the current time;

a workload information obtaining unit, configured to obtain first current workload information of the second service processing resource and second current workload information of the third service processing resource if the first service processing resource is in a full workload state at the current time;

a network connection information obtaining unit, configured to obtain first network connection information between the second service processing resource and the first proxy service node, and obtain second network connection information between the third service processing resource and the second proxy service node;

a historical service data obtaining unit, configured to obtain first historical service data of the second service processing resource, and obtain second historical service data of the third service processing resource;

a target processing resource determining unit, configured to determine a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, and the first historical service data and the second historical service data;

a current service load allocation unit, configured to allocate the current service load to the target service processing resource, so that the target service processing resource processes the current service load.

15. A resource scheduling system is characterized in that the system comprises a first proxy cluster, the first proxy cluster comprises a first proxy service node, the first proxy service node corresponds to a first service processing resource, a second service processing resource and a third service processing resource, the first service processing resource corresponds to a first service generation device, the second service processing resource corresponds to a second service generation device, and the third service processing resource corresponds to a third service generation device; wherein the content of the first and second substances,

the first proxy service node is used for acquiring the current service load generated at the current moment from the first service generation equipment; when the first service processing resource is in a full working load state at the current moment, acquiring first current working load information of the second service processing resource and second current working load information of the third service processing resource; acquiring first network connection information between the second service processing resource and the first proxy service node, and acquiring second network connection information between the third service processing resource and the second proxy service node; obtaining first historical service data of the second service processing resource and obtaining second historical service data of the third service processing resource; determining a target service processing resource from the second service processing resource and the third service processing resource according to the first current workload information, the second current workload information, the first network connection information, the second network connection information, the first historical service data and the second historical service data; distributing the current service load to the target service processing resource;

the target traffic processing resource is configured to receive and process the current traffic load from the first proxy service node.

16. The system of claim 15, wherein the first proxy cluster further comprises a second proxy service node, the second proxy service node being connected to the third traffic processing resource, the third traffic generating device, and the first proxy service node, respectively.

17. The system of claim 15, further comprising a second proxy cluster, the second proxy cluster comprising at least one proxy service node, a control node being present in the proxy service nodes of the first proxy cluster; wherein the content of the first and second substances,

the control node of the first proxy cluster is configured to send the current traffic load to the second proxy cluster when the traffic processing resources corresponding to each proxy service node in the first proxy cluster are all in the full workload state, so as to determine the target traffic processing resource from the traffic processing resources corresponding to at least one proxy service node of the second proxy cluster.

18. A computer-readable storage medium having stored thereon computer-executable instructions for performing the method of any one of claims 1 to 13.

19. An electronic device, comprising:

at least one processor;

a storage device configured to store at least one program that, when executed by the at least one processor, causes the at least one processor to implement the method of any one of claims 1 to 13.