Method, system, device, storage medium and program product for scheduling computing resources

Abstract

The application discloses a scheduling method, system and device of computing network resources, a storage medium and a program product, relates to the field of information technology, and is applied to the scheduling system of the computing network resources including a computing network operation layer. The method comprises the following steps: acquiring computing network resource scheduling information input by the computing network operation layer, wherein the computing network resource scheduling information comprises resource scheduling requirements and resource service topology information; constructing a resource service knowledge graph according to the resource service topology information, and determining a computing network scheduling strategy according to the resource service knowledge graph and the resource scheduling requirements; and scheduling the computing network resources according to the computing network scheduling strategy. The application solves the technical problem that the scheduling accuracy of the computing network resources is not high.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, in particular to a method and system for scheduling of computing network resources, a device, a storage medium and a program product. BACKGROUND

[0002] With the continuous development of IT (Information Technology) technology, the use of computing network is becoming more and more frequent, and therefore higher requirements are put forward for the scheduling of computing network resources.

[0003] At present, when scheduling computing network resources, the optimal solution of computing network resource scheduling is usually sought for the local resources that need to be scheduled, and then the resources are called based on the optimal solution of computing network resource scheduling of local resources. However, this scheduling method of computing network resources will cause the problem that the overall business operation is not the optimal computing network resource scheduling due to the optimal solution of resource scheduling for local resources only. Therefore, this scheduling method of computing network resources will cause the problem of low accuracy of computing network resource scheduling due to the overall business operation not being the optimal computing network resource scheduling.

[0004] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0005] The main purpose of the present application is to provide a method and system for scheduling of computing network resources, a device, a storage medium and a program product, aiming at solving the technical problem of low accuracy of computing network resource scheduling.

[0006] To achieve the above purpose, the present application provides a method for scheduling of computing network resources, which is applied to a computing network resource scheduling system including a computing network operation layer. The method comprises the following steps:

[0007] obtaining computing network resource scheduling information input by the computing network operation layer, wherein the computing network resource scheduling information includes resource scheduling requirements and resource business topology information;

[0008] constructing a resource business knowledge graph according to the resource business topology information, and determining a computing network scheduling strategy according to the resource business knowledge graph and the resource scheduling requirements;

[0009] scheduling computing network resources according to the computing network scheduling strategy.

[0010] To achieve the above purpose, the present application also provides a computing network resource scheduling system, which includes a computing network operation layer and a controller connected with the computing network operation layer. The controller comprises:

[0011] An information obtaining module is configured to obtain algorithm network operation layer input algorithm network resource scheduling information, wherein the algorithm network resource scheduling information comprises resource scheduling requirements and resource service topology information.

[0012] A strategy determining module is configured to construct a resource service knowledge graph based on the resource service topology information, and determine an algorithm network scheduling strategy based on the resource service knowledge graph and the resource scheduling requirements.

[0013] A resource scheduling module is configured to perform algorithm network resource scheduling based on the algorithm network scheduling strategy.

[0014] The application further provides an algorithm network resource scheduling device, which comprises at least one processor and a memory connected to the at least one processor in communication, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the algorithm network resource scheduling method.

[0015] The application further provides a storage medium, wherein the storage medium stores a program for implementing the algorithm network resource scheduling method, and the program is executed by a processor to implement the steps of the algorithm network resource scheduling method.

[0016] The application further provides a program product, which comprises an algorithm network resource scheduling program, and the algorithm network resource scheduling program is executed by a processor to implement the steps of the algorithm network resource scheduling method.

[0017] The algorithm network resource scheduling method of the application is applied to an algorithm network resource scheduling system, which comprises an algorithm network operation layer. The algorithm network resource scheduling information input by the algorithm network operation layer is obtained, the algorithm network resource scheduling information comprises resource scheduling requirements and resource service topology information. A resource service knowledge graph is constructed based on the resource service topology information, and an algorithm network scheduling strategy is determined based on the resource service knowledge graph and the resource scheduling requirements. Algorithm network resource scheduling is performed based on the algorithm network scheduling strategy.

[0018] By acquiring resource scheduling requirements and resource service topology information, and constructing a resource service knowledge graph according to the resource service topology information, and determining an algorithm network scheduling strategy based on the resource service knowledge graph and the resource scheduling requirements, and finally scheduling the algorithm network resources according to the algorithm network scheduling strategy. Thus, the problem that the existing technology only optimizes the local resources for resource scheduling, and causes the overall business operation to be not the optimal algorithm network resource scheduling, is avoided. This algorithm network resource scheduling method constructs a resource service knowledge graph related to the resource service topology information, and then determines an algorithm network scheduling strategy based on the resource scheduling requirements in the resource service knowledge graph, so as to schedule the algorithm network resources from the aspect of considering the overall business, and thus avoids the defect of local resource optimization for resource scheduling, and improves the scheduling accuracy of the algorithm network resources. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings, which are incorporated herein and constitute part of the specification, illustrate embodiments consistent with the application and, together with the description, serve to explain the principles of the application.

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative labor.

[0021] Figure 1 A flowchart of the algorithm network resource scheduling method provided by Embodiment One of the present application;

[0022] Figure 2 A flowchart of the algorithm network resource scheduling method;

[0023] Figure 3 A scene diagram of the algorithm network resource scheduling method;

[0024] Figure 4 A flowchart of the algorithm network resource scheduling method of the present application;

[0025] Figure 5 A knowledge graph diagram of the algorithm network resource scheduling method of the present application;

[0026] Figure 6 A structural diagram of the algorithm network resource scheduling system provided by Embodiment Four of the present application;

[0027] Figure 7 A structural diagram of the electronic device provided by Embodiment Five of the present application.

[0028] The purpose of the present application, functional features and advantages will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0029] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0030] Embodiment one

[0031] With the development of computing power network, customers have higher requirements for computing network scheduling, such as efficient use of computing network resources and rapid completion of tasks. Among them, cloud cost optimization is one of the indicators that most customers care about. The cost of cloud resources = the unit price of cloud resources x resource consumption, where the unit price belongs to the resource operation level and is a variable attribute managed by the computing network operation layer. The existing computing network resource scheduling often manages different resources according to the attributes of resource type, specification, region, unit price, etc., and cannot provide the overall cost of the business. Therefore, in the resource arrangement process, the computing network scheduling layer obtains the pricing information of individual resources from the computing network operation layer to find the "cloud cost optimization" scheduling strategy of individual resources, but lacks overall business information. That is, the scheduling strategy can only be the optimal solution of local resources and the optimal solution of local time, and such local optimization scheme may increase the overall cost when facing large-scale and complex business demands. For example, the scheduling strategy of a certain resource may be optimal in local time and local resources, but it may not be optimal in the overall operation of the business, and even increase the cost of cross-province transmission.

[0032] Therefore, based on the above defects of computing network resource scheduling, a new computing network resource scheduling method is proposed, which schedules computing network resources by constructing a resource business knowledge graph and considering the overall business, thereby avoiding the defects of local resource optimal solution resource scheduling and improving the accuracy of computing network resource scheduling.

[0033] The embodiment of the present application provides a computing network resource scheduling method. In embodiment one of the computing network resource scheduling method of the present application, referring to Figure 1 , Figure 1 The flowchart of the computing network resource scheduling method provided by the embodiment one of the present application is provided. The computing network resource scheduling method is applied to a computing network resource scheduling system. The computing network resource scheduling system includes a computing network operation layer. The computing network resource scheduling method includes the following steps.

[0034] Step S10, obtaining computing network resource scheduling information input by the computing network operation layer, the computing network resource scheduling information including resource scheduling demand and resource business topology information;

[0035] By way of example, referring to Figure 2 , Figure 2 is a flowchart of a scheduling method for a common computing network resource. The functional module diagram of the common computing network resource scheduling is shown in the figure. The infrastructure layer is responsible for the management of computing network physical resources, including computing power resources and network resources. When the infrastructure resources change (such as adding or offline computing power nodes, network connectivity state changes, customer location changes, etc.), the change information will be perceived as the computing network state and synchronized to the computing network operation layer (updating the resource state of the computing network operation layer), and then affect the resource selling of the computing network operation layer. At the same time, feedback is given to the computing network scheduling layer to affect the resource scheduling decision. When external customers have computing power needs, they will provide resource subscription requests to the computing network operation layer. The request will continue to be sent to the computing network scheduling layer for resource scheduling. The computing power arrangement center in the computing network scheduling layer obtains a candidate resource list according to the resource type, performance, and other parameters input by the customer, and then obtains the pricing information of the computing network resources from the computing network operation layer, that is, queries the "cloud cost" information in the computing network scheduling layer, and selects the resource set with the lowest cost in the candidate resource list as the final solution and sends it to the computing power scheduling center. At this time, the computing power scheduling center sends a physical resource scheduling command to the infrastructure layer to open the resource, and at the same time forwards the computing network request to the computing network operation layer to feedback to the customer. That is, the infrastructure layer collects real-time data of the computing network resources, including the type, connectivity, service quality, etc. of the resources, and transmits the collected information to the computing network perception layer; the computing network perception layer returns the updated resource state to the computing network operation layer to display the resources that can be used and are being used; the computing network perception layer provides real-time resource information to the computing network scheduling layer for use in later scheduling; the computing network operation layer forwards the resource appeal of the customer to the computing network scheduling layer for resource application. Further, refer to Figure 3 , Figure 3A scene diagram of a commonly used scheduling method of computing network resources. In the diagram, a simplified architecture diagram of a typical content service (such as Douyin) is shown. Video and other content data are stored in a database and provided to customers for viewing through an online service. The operation logs of the customers when viewing are fed back to a big data task to calculate customer portraits. The customer portrait data are used to support the online service. The construction of the online service, the content database, the customer portrait database, and the big data task may be initiated by different teams at different times. At this time, after the superposition of different batches of scheduling strategies, the overall business cost may not be optimal. For example, the online service and the content database may be constructed in the first batch. At this time, in order to reduce the cost caused by network transmission, they may be deployed in a Shanghai computer room close to the customers. However, when the second batch calculates the customer portraits, a computer room in Inner Mongolia with relatively low computing and storage costs may be selected for deployment. However, the customer portraits need to be provided to the online service. In this scheme, the online service needs to transmit across provinces when requesting the customer portrait database. As the business volume increases, the cost of cross-province transmission will continue to increase. It is finally found that if the second batch applies for a suboptimal computer room in Nanjing, the overall cost may be lower. Based on the above scheduling method of computing network resources, the scheduling method of computing network resources in the present embodiment is proposed.

[0036] In the present embodiment, priority is given to Figure 4 , Figure 4 A flowchart of the scheduling method of computing network resources of the present application. The scheduling method of computing network resources is applied to a computing network resource scheduling system. The computing network resource scheduling system includes a computing network operation layer and a controller of the present embodiment. The controller can be a FinOps (Financial Operations) module or a controller composed of other control modules. In this case, the FinOps module is taken as an example for illustration. The FinOps module is directly connected to the computing network operation layer to obtain the synchronous business expansion and request global cost information of the computing network operation layer. The global scheduling strategy is determined in the FinOps module. Finally, the global scheduling strategy is implemented in the computing network scheduling layer to schedule the computing network resources. The FinOps module includes a business topology management function: a resource-business knowledge graph is constructed based on the connectivity relationship between resources and the ownership relationship between resources and businesses; a cost management function: based on the time dimension (such as day or hour), the historical cost information of all resources is recorded, including the latest cost. A cost correlation analysis function: based on the resource-business knowledge graph and the cost information of the resource time dimension, the time-based cost information of the business dimension is calculated and applied to Figure 3All services can be calculated by calculating the cost of each resource to know the cost of online services, the cost of customer portraits. Instead of the cost of a certain server or the cost of a certain disk, the cost can have business meaning to make the cost more valuable to customer decision-making. Cost governance function: according to the business cost information, the scheduling strategy of the resource is optimized and adjusted at any time (customer by demand or cost change), and the resource preference parameters are issued to the algorithm network scheduling layer through the algorithm network operation layer to perform final scheduling. Among them, the cost governance function will abstract the matching strategy of the resource into a subgraph matching algorithm in the cost map, that is, find a subgraph in a large graph with weight, whose topology structure meets the customer demand graph, and whose weight is the smallest. The weight can be cost, real-time, region and security, etc. For example, the best subgraph is determined by considering cost, real-time, region and security as the global scheduling strategy to ensure the accuracy of the scheduling of the algorithm network resource.

[0037] In an embodiment, when the FinOps module based on Figure 4 schedules the algorithm network resource, after obtaining the algorithm network resource scheduling information input by the algorithm network operation layer, the algorithm network resource scheduling information includes resource scheduling demand and resource business topology information. The resource scheduling demand refers to the scheduling demand of the target customer, such as low cost, general real-time, and information about what kind of algorithm network resource needs to be scheduled. The resource business topology information refers to the information of the target customer's resource and business, and then the resource business knowledge graph of the customer can be constructed based on the information of the resource and business, and the calling based on the resource business knowledge graph is performed to ensure the accuracy of the scheduling of the algorithm network resource. It is worth noting that the resource scheduling demand can not be initiated by the customer. In one case, such as customer A1 is normally using algorithm network resource A, and the controller finds that the cost of algorithm network resource A has changed, the algorithm network resource called by customer A1 will be analyzed again, that is, the resource business knowledge graph of customer A1 is reconstructed, and then it is determined in the resource business knowledge graph whether algorithm network resource A needs to be continued to be used or better algorithm network resource B needs to be used to meet the latest input resource scheduling demand of customer A1, so as to improve the intelligence of the scheduling of the algorithm network resource, and intelligently select the best algorithm network resource to improve the customer experience (the customer can set whether to adapt to the change).

[0038] Step S20, constructing a resource business knowledge graph according to the resource business topology information, and determining an algorithm network scheduling strategy according to the resource business knowledge graph and the resource scheduling demand;

[0039] In the embodiment, after obtaining the resource scheduling requirement and the resource service topology information, a resource service knowledge graph is constructed based on the resource service topology information, that is, the resource service knowledge graph of the entire target customer required is constructed by determining the resources and services in the resource service topology information, wherein the resource service knowledge graph refers to a topology network formed by resources, services, and the association relationship therebetween when the target customer satisfies the service appeal by applying for resources. In the topology network, the type, id, cost, and other attribute information of the resources and services are attached, thereby forming the knowledge graph of the entire service. Based on the determined resource service knowledge graph, the best network calculation scheduling strategy of the resource scheduling requirement is determined. The network calculation scheduling strategy refers to the best network calculation scheduling manner determined based on the resources and services, such as the commonly used manner that the A calculation network resource is scheduled best for the S1 scheduling of the target customer A1, but the overall cost is not considered, such as most of the scheduling resources of the target customer A1 are the calculation network resources in the province, and from the overall service, the scheduling of the province outside in the S1 scheduling is not the best for the whole, and the A calculation network resource is scheduled best for the S1 scheduling in the province. The scheduling based on the whole improves the accuracy of the calculation network resource scheduling, and further satisfies the use experience of the customer. It is worth noting that the resource scheduling requirement can be the cost, real-time performance, region, and security requirement, such as the calculation network resource with the highest real-time performance is preferentially selected for scheduling, the calculation network resource with the highest real-time performance in the resource service knowledge graph is preferentially considered, the calculation network resource in the province is preferentially selected for scheduling, the calculation network resource in the province in the resource service knowledge graph is preferentially considered, and when multiple requirements exist, the selection is based on the weight, such as the cost is preferentially selected, the real-time performance is secondly selected, and then the calculation network resource is determined based on the weight (the priority of the cost and the real-time performance) of the calculation network resource.

[0040] In step S30, the calculation network resource is scheduled according to the network calculation scheduling strategy.

[0041] In the embodiment, after the network calculation scheduling strategy is determined, the calculation network resource is scheduled based on the network calculation scheduling strategy. The main process is that the network calculation scheduling strategy is sent to the network calculation scheduling layer, the calculation network resource is scheduled in the infrastructure layer by controlling the network calculation scheduling layer, and then the entire calculation network resource scheduling process based on the global service is completed, so as to ensure the scheduling accuracy of the calculation network resource.

[0042] In an embodiment, the scheduling process of the entire algorithm network resource is that the FinOps module queries all the business FinOps data associated with the current resource demand through the association and traversal algorithm of the knowledge graph, instead of the previous isolated query method of each resource, and finally transmits the queried FinOps data to the algorithm network scheduling layer for resource scheduling through the algorithm network operation layer, synchronizes the state information to the algorithm network operation layer after the resource scheduling is successful, and synchronizes the topology (including connectivity and cost) information of the business to the FinOps module through the algorithm network operation layer. The entire algorithm network resource scheduling process queries all the data of the subgraph in batches through the association and traversal algorithm of the business graph, which is more efficient than the point query method of initiating a query process for each resource node. At the same time, by constructing a more comprehensive business knowledge graph, the cost at the business level can be more accurately evaluated, so that a more optimal scheduling strategy can be developed. It is worth noting that the FinOps module can further analyze and predict resource price fluctuations and demand peaks (such as cost prediction according to cost trends) by collecting time series cost information of resources and businesses, so as to take measures in advance to reduce the overall algorithm network resource scheduling cost.

[0043] As an example, the algorithm network resource scheduling method comprises steps S10 to S30: an algorithm network resource scheduling system is applied to the algorithm network resource scheduling system, the algorithm network resource scheduling system comprises an algorithm network operation layer, algorithm network resource scheduling information is obtained by inputting the algorithm network operation layer, the algorithm network resource scheduling information comprises resource scheduling demand and resource business topology information; a resource business knowledge graph is constructed according to the resource business topology information, and an algorithm network scheduling strategy is determined according to the resource business knowledge graph and the resource scheduling demand; and the algorithm network resource is scheduled according to the algorithm network scheduling strategy. By obtaining the resource scheduling demand and the resource business topology information, constructing the resource business knowledge graph according to the resource business topology information, and determining the algorithm network scheduling strategy based on the resource business knowledge graph and the resource scheduling demand, the algorithm network resource is finally scheduled according to the algorithm network scheduling strategy. Thus, the problem that the overall business operation is not the optimal algorithm network resource scheduling caused by the optimal solution of the local resource in the prior art is avoided. This algorithm network resource scheduling method constructs a resource business knowledge graph related to the resource business topology information, and then determines an algorithm network scheduling strategy based on the resource scheduling demand in the resource business knowledge graph, so as to schedule the algorithm network resource from the aspect of considering the overall business, thereby avoiding the defect of the optimal solution of the local resource scheduling and improving the accuracy of the algorithm network resource scheduling.

[0044] Embodiment two

[0045] Further, in another embodiment of the present application, the same or similar content as the above embodiment one can be referred to the above introduction, and the subsequent will not be described. On this basis, the resource service topology information includes resource topology information and service topology information, the resource topology information includes initial costs of a plurality of resource nodes and edge costs of communication of each resource node, and the step of constructing the resource service knowledge graph according to the resource service topology information includes:

[0046] Step S21, for each resource node, determining the sum value between the initial cost and the edge cost as the resource point cost;

[0047] Step S22, constructing the resource service knowledge graph according to the resource point cost of each resource node and the service topology information.

[0048] In the embodiment, the resource service topology information includes resource topology information and service topology information, the resource topology information includes initial costs of a plurality of resource nodes and edge costs of communication of each resource node, the resource topology information refers to the related information of the algorithm network resource point, the service topology information is the related information of the business based on the algorithm network resource, the resource node refers to the node on the algorithm network resource, the initial cost refers to the initial cost of the resource node, which can be defined in advance based on the customer, and the edge cost refers to the cost of communication with the resource node. Further, all resource nodes of the resource layer and the "access" relationship therebetween are filtered out. The resource node includes all software and hardware resources that will directly generate costs, such as a bare metal machine, a hosted database, etc., and these resources can be queried in the billing details. The "access" relationship reflects the cost allocation and attribution relationship. For example, the use relationship between a bare metal machine and a microservice because the bare metal machine is deployed with the microservice, and the attribution relationship of a hosted database to a business because the hosted database stores the data of the business, i.e., the edge cost of the resource node. For each resource node, the sum value between the initial cost and the edge cost is determined as the resource point cost, and then the resource service knowledge graph is constructed according to the resource point cost of the resource node and the service topology information. The calculation formula of the resource point cost is as follows:

[0049]

[0050] Wherein, Node old is the initial cost of the node, Node new is the updated cost (resource point cost) of the node, Edge kis the edge cost of the node to the kth target node, such as the initial cost of the resource node F is F11, the edge cost to the target node F1 is F12, and the edge cost to the target node F2 is F13, that is, the resource point cost of the resource node F is F11+F12+F13. Finally, the resource point cost corresponding to each resource node can be determined, and then the resource service knowledge graph is constructed based on the resource point cost corresponding to each resource node and the service topology information, to provide a basis for subsequent network resource scheduling. That is, the resource nodes and the nodes of the service layer after updating the cost jointly constitute a new DAG graph (Directed Acyclic Graph, directed acyclic graph), and the nodes in the DAG graph are connected to each other through the “belonging” relationship, and then the resource service knowledge graph is constructed through the DAG graph.

[0051] Further, the service topology information includes the cost proportion of the downstream cost of each resource node, and the step of constructing the resource service knowledge graph according to the resource point cost of each resource node and the service topology information includes:

[0052] Step S221, for the belonging relationship of each resource node, determining the product between the resource point cost and the cost proportion as the downstream node cost;

[0053] Step S222, constructing the resource service knowledge graph based on the downstream node cost, the resource point cost and the belonging relationship of the resource node.

[0054] In this embodiment, the service topology information includes the cost proportion of the downstream cost of each resource node, wherein a candidate node set of the cost to be apportioned can be found first, and the characteristics of such nodes are that all upstream nodes have apportioned the cost to it, but it has not apportioned the cost to the downstream node (the first execution must be the resource node obtained in the above step); then these candidate nodes will apportion their own cost to the next level node according to the proportion along the “belonging” relationship. The calculation formula is:

[0055]

[0056] wherein, Node next is the cost of the downstream node, Node k is the cost of the kth upstream candidate node of the downstream node, and Ratio kis the proportion of the kth candidate node allocated to the node. As a downstream node exists an upstream node, the cost of the upstream node is F11, and the cost proportion is a, then the cost of the downstream node is aF11, and the downstream node cost of all downstream nodes can be determined, and the cost calculation of each downstream node is continued until the cost of all nodes in the DAG graph is calculated. Finally, the resource business knowledge graph can be constructed based on the downstream node cost, the resource point cost and the attribution relationship of the resource node, that is, the cost of each business is determined based on the resource node in the DAG graph, such as a business in the DAG graph, the scheduling cost of the algorithm network resource of the business is determined, and if there are multiple businesses, the scheduling cost of the algorithm network resource of the multiple businesses is determined. Further, the resource business knowledge graph can be constructed based on the resource business knowledge graph and the resource scheduling demand. Figure 5 , Figure 5 is a knowledge graph diagram of the algorithm network resource scheduling method of the present application. The entire knowledge graph diagram can be divided into two parts: a business layer and a resource layer. The resource layer: described in the first / second machine room data center, describes the connection relationship of physical resources, including the "deployment" relationship between resources (such as EC2) and data centers, and the "access" relationship between resources (such as between gateways and k8s nodes). The "access" relationship can also be divided into types such as intranet access, cross-network dedicated line access, and public network access according to the data center to which it belongs. The resource layer nodes and relationships of the graph contain cost attribute information, such as the cost of the EC2 node, the dedicated line cost of the cross-province dedicated line access, and the traffic cost of the public network access. The business layer: describes the business "attribution" relationship, including which sub-businesses the resources "belong to" and which parent businesses the sub-businesses "belong to". The nodes and relationships of the business layer also contain cost attribute information. The cost on the node represents the overall cloud cost of the business; the cost on the relationship represents how much cloud cost the downstream "attributes" to the upstream business through this path. Based on the resource scheduling, the algorithm network resource scheduling accuracy is ensured from the overall business.

[0057] In an embodiment, the step of determining the algorithm network scheduling strategy according to the resource business knowledge graph and the resource scheduling demand comprises:

[0058] Step S23: determining the scheduling basis in the resource scheduling demand, and selecting a matching schedulable subgraph in the resource business knowledge graph based on the scheduling basis;

[0059] Step S24: when there is one schedulable subgraph, taking the scheduling strategy corresponding to the schedulable subgraph as the algorithm network scheduling strategy;

[0060] Step S25: when there are multiple schedulable subgraphs, determining a target schedulable subgraph with the smallest weight in the schedulable subgraphs, and taking the scheduling strategy corresponding to the target schedulable subgraph as the algorithm network scheduling strategy.

[0061] In the embodiment, when determining the algorithm network scheduling strategy based on the resource scheduling demand in the resource service knowledge graph, the scheduling basis in the resource scheduling demand is determined, and then a matching schedulable subgraph is selected in the resource service knowledge graph based on the scheduling basis. The scheduling basis refers to the scheduling basis of the customer, such as the need to schedule storage algorithm network resources, and the requirement for the storage algorithm network resources is low cost, etc. Then a matching schedulable subgraph can be selected in the resource service knowledge graph. The subgraph generally meets the storage algorithm network resource, that is, the subgraph is determined based on the primary demand of the customer, wherein the primary demand refers to the basic scheduling demand of the customer, that is, the demand of the customer scheduling the algorithm network resource. At this time, a schedulable subgraph that meets the primary demand of the customer can be selected in the resource service knowledge graph, that is, a business subgraph composed of algorithm network resources that meet the primary demand of the customer. At this time, when there is one schedulable subgraph, the scheduling strategy corresponding to the schedulable subgraph is taken as the algorithm network scheduling strategy, because there is no optional case at this time, the scheduling strategy corresponding to the schedulable subgraph is directly taken as the algorithm network scheduling strategy. The scheduling strategy corresponding to the schedulable subgraph refers to the scheduling strategy of scheduling each algorithm network resource in the schedulable subgraph. Conversely, when there are multiple schedulable subgraphs, the target schedulable subgraph with the minimum weight in the schedulable subgraph is determined, and the scheduling strategy corresponding to the target schedulable subgraph is taken as the algorithm network scheduling strategy. The minimum weight is actually based on the secondary demand of the scheduling basis (that is, what advantage needs to be met this time, such as real-time performance, cost, etc.), and then the schedulable subgraph with the minimum comprehensive weight of real-time performance and cost (that is, the best real-time performance and cost, such as selecting a schedulable subgraph with preferred real-time performance and low cost as the target schedulable subgraph, assuming that the user weight is 80% real-time performance and 20% cost, the real-time performance is preferred, and the cost is appropriately considered) can be determined as the target schedulable subgraph. The scheduling strategy corresponding to the target schedulable subgraph refers to the scheduling strategy of scheduling each algorithm network resource in the target schedulable subgraph, and then the algorithm network resources in the schedulable subgraph can be scheduled based on the best scheduling mode to improve the scheduling accuracy of the algorithm network resources. It is worth noting that the entire process of determining the schedulable subgraph is the process of determining the idle resources of the algorithm network resources, that is, the idle state of the algorithm network resources needs to be determined first, then the judgment of the primary demand and the judgment of the secondary demand are executed, and then the accuracy of the algorithm network resource scheduling can be ensured. Figure 1

[0062] Embodiment three

[0063] ​Further, in another embodiment of the present application, the same or similar content as the above embodiment one can be referred to the above introduction, and the subsequent will not be described. On this basis, the scheduling system of the algorithm network resource includes an algorithm network scheduling layer connected with an algorithm network operation layer, and an infrastructure layer connected with the algorithm network scheduling layer, and the steps of scheduling the algorithm network resource according to the algorithm network scheduling strategy include:

[0064] In step S31, the target algorithm network resource to be scheduled in the algorithm network scheduling strategy is determined, and the algorithm network scheduling layer is controlled to call the target algorithm network resource in the infrastructure layer.

[0065] In this embodiment, the scheduling system of the algorithm network resource includes an algorithm network scheduling layer connected with an algorithm network operation layer, and an infrastructure layer connected with the algorithm network scheduling layer. After determining the algorithm network scheduling strategy, the target algorithm network resource to be scheduled in the algorithm network scheduling strategy is determined, and then the algorithm network scheduling layer is controlled to call the target algorithm network resource in the infrastructure layer to complete the algorithm network resource scheduling. By using the knowledge graph of FinOps calculation and recording algorithm network business data (the knowledge graph organizes the topology of algorithm network system resources, the topology of business, and the correlation between them), and based on this graph information, the scheduling strategy parameters required by the downstream algorithm network scheduling layer are output. Compared with the existing proposal which can only rely on individual resource independent scheduling, the conflict problem between local optimal and global optimal solutions can be well solved, and the use cost of the algorithm network is reduced.

[0066] In an embodiment, the step of obtaining the algorithm network resource scheduling information input by the algorithm network operation layer includes:

[0067] In step S11, the algorithm network resource information input by the algorithm network operation layer is obtained, wherein the algorithm network resource information includes resource cost state and resource scheduling state.

[0068] In step S12, when the resource cost state is a preset cost change state, and / or the resource scheduling state is a preset demand resource state, the resource scheduling demand corresponding to the target scheduling object and the resource business topology information are taken as the algorithm network resource scheduling information, wherein the target scheduling object includes a scheduling object whose resource cost state is a preset cost change state, and / or whose resource scheduling state is a preset demand resource state.

[0069] In this embodiment, when communicating with the algorithm network operation layer in real time, it is determined whether the algorithm network resource scheduling is needed based on the communication information between the algorithm network operation layer. The algorithm network resource information input by the algorithm network operation layer is obtained in real time, wherein the algorithm network resource information includes resource cost state and resource scheduling state. The resource cost state refers to the cost state of the resource points of the entire algorithm network, which is generally changed or unchanged. Whether the algorithm network resource scheduling is needed can be determined based on whether the cost is changed. The resource scheduling state refers to the demand input by the customer, such as the demand of customer A for calling the algorithm network resource. The demand is input to the FinOps module as the resource scheduling state. Further, when the resource cost state is a preset cost change state and / or the resource scheduling state is a preset demand resource state, the resource scheduling demand corresponding to the target scheduling object and the resource business topology information are taken as the algorithm network resource scheduling information, wherein the target scheduling object includes a scheduling object whose resource cost state is a preset cost change state and / or whose resource scheduling state is a preset demand resource state. The preset demand resource state refers to that there can be an algorithm network resource scheduling demand. The preset cost change state refers to that the algorithm network resource cost changes. The target scheduling object can be a customer who has a scheduling demand or a customer whose resource cost changes and the changed resource is used by the customer (under the premise that the customer defines the self-adaptive adjustment of the algorithm network resource). The algorithm network resource of the customer is scheduled based on the customer demand and the real-time cost change, so as to improve the intelligence of the algorithm network resource scheduling.

[0070] In an embodiment, the input of the FinOps module includes two parts: a change event Eventtopo of the business topology (resource cost state) and newly applied resource parameters Argsresource (resource scheduling state). The structure of Eventtopo is {type: type1, id: id1, args: args1}. Type indicates what type of change event, the candidate set includes: adding a node, modifying a node, deleting a node, adding an edge, modifying an edge, and deleting an edge; id is the unique ID of the modified object; args indicates the parameters of the modified node or edge, mainly attribute information, which is a kv structure organized in json format, and the possible attributes include cost, updatetime, operator, etc. The structure of Argsresource is consistent with the parameters for requesting the algorithm network scheduling layer in the old algorithm, mainly including resource type, resource quantity, security preference, cost preference, etc. For example Figure 3In the scenario shown, the resource type is disk storage; the resource quantity is 1TB; the security preference is that data should only flow within the province and not be stored in other provinces; the cost preference is to obtain the lowest possible price; and the performance requirement is hourly response time. Furthermore, the output structure of the FinOps module is: {nodes:[{id:id1,type:type1,cluster:cluster1...},{id:id2,type:type2,cluster:cluster2...}...],edges:[{id:id1,type:type1,from:from1,to:to1}...]}. Here, `nodes` records information about all computing nodes, mainly including `id` (unique ID of the computing node), `type` (node ​​type), and `cluster` (cloud and cluster information where the node resides); `edges` records information about all networks, mainly including `id` (unique ID of the network), `type` (network type), `from` (starting computing node information for the network), and `to` (target computing node information for the network). In addition, it can also contain other attribute information used for calculation or querying. For example... Figure 3 In this scenario, the output nodes are: id1 represents the unique ID of the disk resource; type is SATAHDD, a low-cost, low-performance disk type; cluster is the Beijing cluster. The edges information is: id-edge1 represents the unique ID of the network path; type is a leased line; from is the ID of a microservice (idx); to is id1, which is the disk ID mentioned above. There are multiple sets of this information. After this information is passed to the network scheduling layer, it attempts to schedule resources. If the scheduling is successful, it returns the result to the client; if it fails, it feeds back the failure information to the FinOps module to recalculate candidate resources. The above is just one example of the input / output structure of the FinOps module; other structures are also possible and are not limited here.

[0071] Example 4

[0072] This application also provides a scheduling system for computing network resources, referring to... Figure 6 , Figure 6 This is a schematic diagram of the structure of the computing network resource scheduling system provided in Embodiment 5 of this application. The computing network resource scheduling system includes a computing network operation layer and a controller connected to the computing network operation layer. The controller includes:

[0073] The information acquisition module A01 is used to acquire computing network resource scheduling information input by the computing network operation layer. The computing network resource scheduling information includes resource scheduling requirements and resource service topology information.

[0074] A policy determination module A02 is configured to construct a resource service knowledge graph based on the resource service topology information, and determine an algorithm network scheduling policy based on the resource service knowledge graph and resource scheduling requirements;

[0075] A resource scheduling module A03 is configured to schedule algorithm network resources based on the algorithm network scheduling policy.

[0076] Optionally, the policy determination module A02 is further configured to:

[0077] For each resource node, determine a sum value between an initial cost and an edge cost as a resource point cost;

[0078] Construct a resource service knowledge graph based on the resource point cost of each resource node and the service topology information.

[0079] Optionally, the policy determination module A02 is further configured to:

[0080] For each resource node, determine a product value between the resource point cost and a cost ratio as a downstream node cost;

[0081] Construct a resource service knowledge graph based on the downstream node cost, the resource point cost, and the ownership relationship of the resource node.

[0082] Optionally, the policy determination module A02 is further configured to:

[0083] Determine a scheduling basis in the resource scheduling requirements, and select a matching schedulable subgraph in the resource service knowledge graph based on the scheduling basis;

[0084] When there is one schedulable subgraph, take a scheduling policy corresponding to the schedulable subgraph as the algorithm network scheduling policy;

[0085] When there are multiple schedulable subgraphs, determine a target schedulable subgraph with the minimum weight in the schedulable subgraphs, and take a scheduling policy corresponding to the target schedulable subgraph as the algorithm network scheduling policy.

[0086] Optionally, the resource scheduling module A03 is further configured to:

[0087] Determine a target algorithm network resource that needs to be scheduled in the algorithm network scheduling policy, and control the algorithm network scheduling layer to call the target algorithm network resource in the infrastructure layer.

[0088] Optionally, the information acquisition module A01 is further configured to:

[0089] Acquire algorithm network resource information input by the algorithm network operation layer, wherein the algorithm network resource information includes resource cost states and resource scheduling states;

[0090] When the resource cost state is the preset cost change state and / or the resource scheduling state is the preset demand resource state, the resource scheduling demand corresponding to the target scheduling object and the resource service topology information are taken as the network resource scheduling information, wherein the target scheduling object includes a scheduling object whose resource cost state is the preset cost change state and / or whose resource scheduling state is the preset demand resource state.

[0091] The scheduling system for network resources provided by the present application adopts the scheduling method for network resources in the above embodiments, and solves the technical problem of low accuracy of scheduling for network resources. Compared with the prior art, the scheduling system for network resources provided by the embodiments has the same beneficial effects as the scheduling method for network resources provided by the above embodiments, and other technical features in the scheduling system for network resources are the same as the features disclosed in the above embodiments, which will not be repeated here.

[0092] Embodiment five

[0093] The embodiments of the present application provide an electronic device, which comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the scheduling method for network resources in the above embodiment one.

[0094] Reference will now be made to Figure 7 which shows a structural schematic diagram of an electronic device suitable for implementing the embodiments of the present application. The electronic device in the embodiments of the present application can include but is not limited to mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablets), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. Figure 7 The electronic device shown is only an example, and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0095] As shown in Figure 7 , the electronic device can include a processing system 1001 (such as a central processor, a graphics processor, etc.), which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or loaded from a storage system 1003 to a random access memory (RAM) 1004. In the RAM 1004, various programs and data required for the operation of the electronic device are also stored. The processing system 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus.

[0096] Generally, the following systems can be connected to the I / O interface 1006: an input system 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output system 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage system 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication system 1009. The communication system can allow the electronic device to communicate wirelessly or wiredly with other devices to exchange data. Although the electronic device is shown with various systems, it is understood that not all of the shown systems are required to be implemented or possessed. More or fewer systems can alternatively be implemented or possessed.

[0097] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present disclosure include a program product including a computer program carried on a computer readable medium, the computer program containing program code for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication system 1009, or installed from the storage system 1003, or installed from the ROM 1002. When the computer program is executed by the processing system 1001, the above-mentioned functions defined in the methods of the embodiments of the present disclosure are performed.

[0098] The electronic device provided by the present application adopts the algorithm network resource scheduling method in the above-mentioned embodiments, and solves the technical problem of low accuracy of algorithm network resource scheduling. Compared with the prior art, the electronic device provided by the embodiments of the present application has the same beneficial effects as the algorithm network resource scheduling method provided by the above-mentioned embodiments, and other technical features in the electronic device are the same as the features disclosed in the above-mentioned method, which will not be repeated here.

[0099] It should be understood that parts of the present disclosure can be realized by hardware, software, firmware or a combination thereof. In the description of the above-mentioned embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0100] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0101] Embodiment six

[0102] The embodiment provides a storage medium, which can be a computer readable storage medium, and has computer readable program instructions stored thereon, the computer readable program instructions being used for executing the scheduling method of the computing network resource in the above embodiment.

[0103] The computer readable storage medium provided by the embodiment of the application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the embodiment, the computer readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, system, or device. The program code contained on the computer readable storage medium can be transmitted by any suitable medium, including but not limited to an electric wire, an optical cable, an RF (radio frequency), and the like, or any suitable combination of the above.

[0104] The computer readable storage medium described above can be contained in an electronic device, or can exist separately and not be assembled into the electronic device.

[0105] The computer readable storage medium described above carries one or more programs, when the one or more programs are executed by the electronic device, the electronic device is caused to perform the following: obtaining computing network operation layer input computing network resource scheduling information, the computing network resource scheduling information including resource scheduling requirements and resource service topology information; constructing a resource service knowledge graph according to the resource service topology information, and determining a computing network scheduling strategy according to the resource service knowledge graph and the resource scheduling requirements; and performing scheduling of the computing network resource according to the computing network scheduling strategy.

[0106] Computer program code for carrying out operations of the present disclosure can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0107] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0108] The modules involved in the embodiments of the present disclosure can be implemented in the manner of software or hardware. Among them, the name of the module does not constitute a limitation to the unit itself in some cases.

[0109] The computer readable storage medium provided by the present application stores computer readable program instructions for executing the above-mentioned algorithm network resource scheduling method, and solves the technical problem that the algorithm network resource scheduling accuracy is not high. Compared with the prior art, the beneficial effects of the computer readable storage medium provided by the embodiments of the present application are the same as those of the algorithm network resource scheduling method provided by the above-mentioned embodiments, which will not be repeated here.

[0110] Embodiment seven

[0111] The application further provides a program product comprising a computer program which, when executed by a processor, implements the steps of the method for scheduling network resources as described above.

[0112] The program product provided by the application solves the technical problem of low accuracy of scheduling network resources. Compared with the prior art, the beneficial effects of the program product provided by the embodiment of the application are the same as those of the method for scheduling network resources provided by the above-described embodiment, and are not described herein.

[0113] The above is only the preferred embodiment of the application, and does not limit the patent scope of the application, and any equivalent structure or equivalent process transformation using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent processing scope of the application.

Claims

1. A method for scheduling of grid resources, characterized in that, The scheduling method for computing network resources is applied to a scheduling system for computing network resources. The scheduling system for computing network resources includes a computing network operation layer, and the scheduling method for computing network resources includes: Obtain the computing network resource scheduling information input by the computing network operation layer, wherein the computing network resource scheduling information includes resource scheduling requirements and resource service topology information; A resource service knowledge graph is constructed based on the resource service topology information, and a computing network scheduling strategy is determined based on the resource service knowledge graph and the resource scheduling requirements. The resource service topology information includes resource topology information and service topology information. The resource topology information includes the initial cost of multiple resource nodes and the edge cost of communication for each resource node. The service topology information includes the cost ratio of the downstream cost of each resource node. The computing network resources are scheduled according to the computing network scheduling strategy.

2. The scheduling method for computing network resources as described in claim 1, characterized in that, The step of constructing a resource service knowledge graph based on the resource service topology information includes: For each resource node, the sum of the initial cost and the edge cost is determined as the resource node cost; A resource business knowledge graph is constructed based on the resource point cost of each resource node and the business topology information.

3. The method for scheduling computing network resources as described in claim 2, characterized in that, The step of constructing a resource business knowledge graph based on the resource point cost of each resource node and the business topology information includes: For each resource node's ownership relationship, the product of the resource node's cost and the cost ratio is determined as the downstream node cost; A resource business knowledge graph is constructed based on the downstream node cost, the resource point cost, and the ownership relationship of the resource nodes.

4. The method for scheduling computing network resources as described in claim 1, characterized in that, The step of determining the computing network scheduling strategy based on the resource service knowledge graph and the resource scheduling requirements includes: Determine the scheduling criteria in the resource scheduling requirements, and select a matching schedulable subgraph in the resource business knowledge graph based on the scheduling criteria; When there is one schedulable subgraph, the scheduling strategy corresponding to the schedulable subgraph is used as the network scheduling strategy. When there are multiple schedulable subgraphs, the target schedulable subgraph with the smallest weight is determined, and the scheduling strategy corresponding to the target schedulable subgraph is used as the network scheduling strategy.

5. The method for scheduling computing network resources as described in any one of claims 1 to 4, characterized in that, The computing network resource scheduling system includes a computing network scheduling layer connected to the computing network operation layer, and an infrastructure layer connected to the computing network scheduling layer. The step of scheduling computing network resources according to the computing network scheduling strategy includes: The target computing network resources that need to be scheduled in the computing network scheduling strategy are determined, and the computing network scheduling layer is controlled to call the target computing network resources in the infrastructure layer.

6. The method for scheduling computing network resources as described in any one of claims 1 to 4, characterized in that, The step of obtaining the computing network resource scheduling information input by the computing network operation layer includes: Obtain the computing network resource information input by the computing network operation layer, wherein the computing network resource information includes resource cost status and resource scheduling status; When the resource cost status is a preset cost change status and / or the resource scheduling status is a preset demand resource status, the resource scheduling demand and resource service topology information corresponding to the target scheduling object are used as computing network resource scheduling information. The target scheduling object includes the scheduling object whose resource cost status is a preset cost change status and / or whose resource scheduling status is a preset demand resource status.

7. A scheduling system for computing network resources, characterized in that, The scheduling system for computing network resources includes a computing network operation layer and a controller connected to the computing network operation layer, the controller comprising: The information acquisition module is used to acquire the computing network resource scheduling information input by the computing network operation layer. The computing network resource scheduling information includes resource scheduling requirements and resource service topology information. The strategy determination module is used to construct a resource service knowledge graph based on the resource service topology information, and determine a computing network scheduling strategy based on the resource service knowledge graph and the resource scheduling requirements. The resource service topology information includes resource topology information and service topology information. The resource topology information includes the initial cost of multiple resource nodes and the edge cost of communication for each resource node. The service topology information includes the cost ratio of the downstream cost of each resource node. The resource scheduling module is used to schedule computing network resources according to the computing network scheduling strategy.

8. A scheduling device for computing network resources, characterized in that, The scheduling equipment for the computing network resources includes: At least one processor; A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the steps of the scheduling method for computing network resources according to any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a program that implements a method for scheduling computing network resources, and the program that implements the method for scheduling computing network resources is executed by a processor to implement the steps of the method for scheduling computing network resources as described in any one of claims 1 to 6.

10. A program product, characterized in that, The program product includes a scheduling program for computing network resources, which, when executed by a processor, implements the steps of the scheduling method for computing network resources as described in any one of claims 1 to 6.

Images