Target node selection method and device, equipment and medium

Abstract

The invention relates to the technical field of computers, and discloses a target node selection method and device, equipment and a medium, and the method comprises the steps: obtaining a container scheduling demand; traversing the plurality of nodes in the cluster according to a pre-selection strategy corresponding to the container scheduling demand to obtain a plurality of candidate nodes; respectively obtaining the total resource amount and the resource occupation condition of each resource on each candidate node; counting the number of containers deployed by all candidate nodes and the number of containers to be deployed; determining an evaluation index of the i-th node according to the number of containers deployed by the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each resource, the total resource amount of the d-th resource on the candidate node, the resource occupation condition of the d-th resource, the demand quantity of the d-th resource and the total number of all candidate nodes; and selecting a target node according to the selection index to complete container scheduling requirements. According to the method, resources are reasonably allocated, and the resource utilization rate is improved to the greatest extent.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a target node selection method, device, equipment and medium. Background Art

[0002] With the widespread application of microservice architecture and container technology in the field of cloud computing, how to manage containerized microservices in large clusters has become an issue that developers are eager to solve. Related technologies provide platform-as-a-service (PaaS) layer cluster management systems, but due to certain limitations of these systems, development has been slow. The emergence of Kubernetes has broken this situation, and its development momentum is getting stronger and stronger, and it has become the preferred container cluster management platform.

[0003] Kubernetes is built on Docker technology and provides a complete set of solutions for containerized applications, including resource scheduling, deployment and operation, and capacity expansion and contraction. Kubernetes is used for container management and scheduling, and its resource scheduling method is the shared state resource scheduling model. However, the default scheduling method of the Kubernetes container cloud platform only considers CPU and memory factors, resulting in low resource utilization and load imbalance, and it is also easy to cause overload of certain nodes in the container cloud cluster. Summary of the invention

[0004] In view of this, the present invention provides a target node selection method, device, equipment and medium to solve the above technical problems.

[0005] In a first aspect, the present invention provides a target node selection method, the method comprising:

[0006] Obtaining container scheduling requirements, where the container scheduling requirements include multiple resources and the required amount of each resource;

[0007] Traverse multiple nodes in the cluster according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes;

[0008] Respectively obtaining the total amount of resources and resource occupancy status of each of the multiple resources on each of the candidate nodes;

[0009] Count the number of containers that have been deployed and the number of containers to be deployed on all the candidate nodes;

[0010] Determine the selection index of the i-th node according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, wherein the i-th node is any one of the multiple candidate nodes, and i and d are both positive integers;

[0011] According to the selection index of each candidate node, a target node is selected from the plurality of candidate nodes to meet the container scheduling requirement.

[0012] The target node selection method provided by the present invention has the following advantages:

[0013] Obtain the container scheduling requirements, wherein the container scheduling requirements include multiple resources required for container scheduling and the demand for each resource. Then, perform a screening according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes. Then, obtain the total amount of resources and resource occupancy of each resource in the multiple resources on each candidate node respectively. In addition, count the number of containers deployed on the candidate nodes and the number of containers to be deployed. Through the above data, the resource occupancy of each candidate node and the number of containers that can be deployed can be known, so as to facilitate the subsequent measurement of which candidate node to allocate the container to be scheduled. Specifically, it is based on the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of resources of the d-th resource in the multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirements, and the total number of all candidate nodes, determine the selection index of the i-th node, wherein the weight value of each resource is a measure of the importance of the resource. After obtaining the selection index of each candidate node through the above data, the optimal node can be selected to complete the container scheduling requirements according to the selection index corresponding to all candidate nodes. Through the above method, resources can be reasonably allocated, resource utilization can be maximized, system load can be reduced, system response speed and performance can be improved, and system stability and reliability can be improved. In addition, user satisfaction can be improved and user waiting time can be reduced.

[0014] In an optional implementation, the selection index of the i-th node is determined according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight values ​​corresponding to each pre-acquired resource, the total amount of the d-th resource among the multiple resources on the i-th candidate node, the resource occupancy of the d-th resource on the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, specifically including:

[0015] Determine the container deployment idle rate of the i-th candidate node according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed;

[0016] Determine a resource idle rate of the dth resource corresponding to the ith candidate node according to the total amount of resources of the dth resource type on the ith candidate node, the resource occupancy of the dth resource on the ith candidate node, and the demand for the dth resource included in the container scheduling requirement;

[0017] The selection index of the i-th node is generated according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes.

[0018] Specifically, by calculating the container deployment idle rate and resource idle rate, we can optimize the use of resources, avoid resource waste, and ensure that there are enough resources to meet the needs of container deployment. The generated selection indicators provide an objective basis for selecting suitable candidate nodes, which helps to make more informed resource allocation decisions. In the above process, by evaluating the idle rate of the node, we can quickly find the node that is most suitable for deploying the container and reduce the deployment time. By effectively utilizing resources, we can reduce unnecessary resource acquisition and operating costs. By evaluating the resource idle rate, the system can automatically expand or adjust when resources are tight, thereby enhancing the elasticity of the system. Based on the resource idle rate, we can ensure that container scheduling is more efficient and reduce deployment failures caused by insufficient resources. The selection indicators can be used to dynamically adjust the resource allocation strategy to adapt to the changing workload. Sort according to the resource requirements of the application, so as to optimize the allocation and utilization of resources and maximize resource utilization. By reasonably scheduling resources, we can reduce the waiting time of tasks and improve the execution efficiency of tasks. Reasonable sorting of resource scheduling can reduce the load of the system and improve the response speed and performance of the system.

[0019] In an optional implementation, the selection index of the i-th node is generated according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes, specifically including:

[0020] Generate a resource idle index corresponding to the i-th candidate node according to the weight values ​​corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node;

[0021] Generate a feedback scoring indicator corresponding to the i-th candidate node according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes;

[0022] The selection index of the i-th node is generated according to the resource idle index and the feedback scoring index respectively corresponding to each of the candidate nodes.

[0023] Specifically, by considering the weight value of resource distribution, resource idle rate, container deployment idle rate and feedback score, a comprehensive and integrated evaluation can be provided to ensure that the evaluation results are more accurate and comprehensive. By generating resource idle indicators, the resource utilization of each candidate node can be intuitively understood, thereby optimizing the resource allocation strategy and avoiding resource waste. The generated feedback score indicators provide decision makers with quantitative data, which helps to make more scientific and reasonable decisions. Finally, by evaluating resource idle rates and feedback scores, it can help identify nodes with insufficient resource utilization, thereby improving overall resource utilization. By selecting nodes with high resource idle rates and reasonable container deployment idle rates, the system can be ensured to be in the best state in terms of performance and availability. Comprehensive consideration of multiple indicators can ensure that the system maintains stable operation when facing different workloads. Quantitative selection indicators simplify the complexity of resource management, allowing administrators to monitor and manage resources more easily.

[0024] In an optional implementation, determining the container deployment idle rate of the i-th candidate node according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed includes:

[0025] It is expressed by the following formula:

[0026]

[0027] Among them, c i is the container deployment idle rate on the i-th candidate node, p iis the number of containers deployed on the i-th candidate node, n is the number of containers to be deployed; c is the total number of containers, and the expression of c is as follows: Among them, m is the total number of candidate nodes.

[0028] In an optional implementation, the resource idle rate of the dth resource corresponding to the ith candidate node is determined according to the total amount of resources of the dth resource type on the ith candidate node, the resource occupancy of the dth resource on the ith candidate node, and the demand for the dth resource included in the container scheduling requirement, and is represented by the following expression:

[0029]

[0030] in, represents the resource idle rate after the demand for the dth resource is deployed for the i-th node, sd i It is represented as the total amount of resources of the dth resource type owned by the i-th candidate node, rd i It is represented as the resource occupancy of the ith candidate node for the dth resource type, p d It represents the demand for the dth resource included in the container scheduling requirement.

[0031] In an optional implementation, the resource idle index corresponding to the i-th candidate node is generated according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node, and is represented by the following expression:

[0032]

[0033] Among them, v i It is represented as the resource idle index corresponding to the i-th candidate node, a d It is represented as the weight value corresponding to the d-th resource, c i is the container deployment idle rate of the i-th candidate node, D is the total number of resource types, a p The weight of the load of the deployed container.

[0034] In an optional implementation, the feedback scoring index corresponding to the i-th candidate node is generated according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes, and is expressed by the following expression:

[0035]

[0036] Among them, b i is represented by the feedback score index corresponding to the i-th candidate node, D is represented by the total number of resource types, a d is represented by the weight value corresponding to the d-th resource, represents the resource idle rate after the demand for the d-th resource is deployed for the i-th node, It is expressed as the average idle rate of the dth resource on all candidate nodes, c i It is represented as the container deployment idle rate of the i-th candidate node, c v It is expressed as the average idle rate of container resources on all candidate nodes, a p The weight of the load of the deployed container.

[0037] In a second aspect, the present invention provides a target node selection device, the device comprising:

[0038] An acquisition module is used to acquire container scheduling requirements, where the container scheduling requirements include multiple resources and the required amount for each resource;

[0039] The traversal module is used to traverse multiple nodes in the cluster according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes;

[0040] The acquisition module is further used to respectively acquire the total amount of resources and resource occupancy status of each of the multiple resources on each of the candidate nodes;

[0041] A statistics module is used to count the number of containers that have been deployed and the number of containers to be deployed on all the candidate nodes;

[0042] A processing module, used to determine the selection index of the i-th node according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, wherein the i-th node is any one of the multiple candidate nodes, and i and d are both positive integers;

[0043] The selection module is used to select a target node from a plurality of the candidate nodes according to the selection index of each of the candidate nodes, so as to meet the container scheduling requirements.

[0044] A target node selection device provided by the present invention has the following advantages: obtaining container scheduling requirements, wherein the container scheduling requirements include multiple resources required for container scheduling, and the demand for each resource. Then, a screening is performed according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes. Then, the total amount of resources and the resource occupancy of each of the multiple resources on each candidate node are obtained respectively. In addition, the number of containers already deployed and the number of containers to be deployed on the candidate nodes are counted. Through the above data, the resource occupancy of each candidate node and the number of containers that can be deployed can be known, thereby facilitating the subsequent measurement of which candidate node the container to be scheduled should be allocated to. Specifically, the selection index of the i-th node is determined according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among the multiple resources on the i-th candidate node, the resource occupancy of the i-th candidate node for the d-th resource, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, where the weight value of each resource is a measure of the importance of the resource. After obtaining the selection index of each candidate node through the above data, the optimal node can be finally selected to complete the container scheduling requirement according to the selection index corresponding to all candidate nodes. Through the above method, resources can be reasonably allocated, resource utilization can be maximized, system load can be reduced, system response speed and performance can be improved, and the stability and reliability of the system can be improved. In addition, user satisfaction can be improved and user waiting time can be reduced.

[0045] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, computer instructions being stored in the memory, and the processor executing the target node selection method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the target node selection method of the first aspect or any corresponding embodiment thereof.

[0047] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions, wherein the computer instructions are used to enable a computer to execute the target node selection method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 is a flow chart of a target node selection method provided by an embodiment of the present invention;

[0050] Figure 2 is a flow chart of another target node selection method provided by an embodiment of the present invention;

[0051] Figure 3 is a flowchart of another target node selection method provided by an embodiment of the present invention;

[0052] Figure 4 is a specific example schematic diagram of a target node selection method provided by an embodiment of the present invention;

[0053] Figure 5 It is a structural block diagram of a target node selection device provided by an embodiment of the present invention;

[0054] Figure 6 It is a schematic diagram of the hardware structure of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0056] With the widespread application of microservice architecture and container technology in the field of cloud computing, how to manage containerized microservices in large clusters has become an issue that developers are eager to solve. The relevant technologies provide platform-as-a-service (PaaS) layer cluster management systems, but due to certain limitations of these systems, their development has been slow. The emergence of Kubernetes has broken this situation, and its development momentum is getting stronger and stronger, and it has become the preferred container cluster management platform. However, the default scheduling solution of Kubernetes can only perform simple scheduling management operations, which cannot meet the diversity and complexity of scheduling requirements in specific application scenarios.

[0057] Docker is a Linux container tool set designed for building, delivering and running distributed applications. Kubernetes is built on Docker technology and provides a complete set of solutions for containerized applications, such as resource scheduling, deployment and operation, and capacity expansion and contraction. Kubernetes is used for container management and scheduling, and its resource scheduling method is exactly the shared state resource scheduling model. However, the default scheduling method of the Kubernetes container cloud platform only considers the factors of CPU and memory, and does not schedule according to the characteristics of the currently deployed application and the number of application deployments on the current node, resulting in low resource utilization and load imbalance. Different container applications have different requirements for resources. Kubernetes only considers a few single-factor scheduling schemes, which cannot achieve multi-dimensional resource utilization load balancing, and easily causes some nodes in the container cloud cluster to be overloaded, resulting in resource waste and load imbalance. To solve the above problems, an embodiment of the present invention provides a target node selection embodiment. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system (computer device) including a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0058] In this embodiment, a target node selection method is provided, which can be used for the above-mentioned terminal devices, such as mobile phones, tablet computers, etc. Figure 1 is a flow chart of a target node selection method provided by an embodiment of the present invention, such as Figure 1 As shown, the process includes the following steps.

[0059] Before introducing the steps of the method of this application, some symbols in the following text are explained.

[0060] Specifically, the symbolic representation of multiple types of resource indicators is defined:

[0061] Assume that there are m computing nodes that meet the application resource requirements after pre-selection, and N = (n1, n2, ..., n m ). Each computing node has D resource indicators. In this embodiment, for example, there are five resources: CPU, GPU, memory, disk, and network bandwidth, so D = 5. The total resources of the nodes in the cluster are represented by S = (sd1, sd2, ..., sd m ), d∈D, where sd i The resource usage (occupancy) of a computing node can also be expressed in the above way as R = (rd1, rd2, ..., rd m ), d∈D, where rd irepresents the usage of the dth resource at node i. The resource demand of the current scheduling application can be defined as p d , indicating the demand of container application for d types of resources.

[0062] Step S101, obtaining container scheduling requirements.

[0063] Specifically, the container scheduling requirements include multiple resources and the required amount of each resource. In an optional example, the types of resources may include multiple resources such as CPU, GPU, memory, disk, and network bandwidth.

[0064] Step S102: traverse multiple nodes in the cluster according to a pre-selection strategy corresponding to the container scheduling requirement to obtain multiple candidate nodes.

[0065] Specifically, commonly used pre-selection strategies may include the following:

[0066] 1. CheckNodeCondition

[0067] Checks whether a Pod object can be scheduled on a node if the node reports that the disk or network is unavailable or not ready.

[0068] 2. General Predicates

[0069] HostName: If the Pod object has a spec.hostname property, check whether the node name string matches the property value.

[0070] 3. PodFitsHostPorts

[0071] If the Pod object defines the ports.hostPort property, check whether the port specified by the Pod is already occupied by other containers or services on the node.

[0072] 4. MatchNodeSelector

[0073] If the Pod object defines a spec.nodeSelector property, check whether the node label matches that property.

[0074] 5. PodFitsResources

[0075] Check whether the resource (CPU, memory) availability on the node meets the running requirements of the Pod object.

[0076] Multiple candidate nodes in the cluster are traversed through one or more of the above pre-selection strategies to obtain multiple candidate nodes.

[0077] Step S103, respectively obtaining the total amount of resources and resource occupancy status of each resource among the multiple resources on each candidate node.

[0078] Step S104: Count the number of containers that have been deployed on all candidate nodes and the number of containers to be deployed.

[0079] Step S105, determining the selection index of the i-th node according to the number of containers already deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling demand, and the total number of all candidate nodes.

[0080] Specifically, the load of the container on the i-th node can be determined based on the number of containers deployed on the i-th node and the number of containers to be deployed, and then the idleness of the container on the i-th node can be determined. The importance of resources can be measured according to the weight of each resource. Among them, the importance of various resources on the computing node often varies greatly. For example, CPU, GPU and memory are relatively scarce resources, so their importance will also be higher.

[0081] The specific method of obtaining the weight value corresponding to the resource can be directly determined based on big data statistics, and of course it can also be obtained through other methods.

[0082] For example, the resource utilization of D resources on m nodes in the target cluster is obtained to obtain a resource utilization set; the entropy value of each resource indicator in the D resource indicators is determined through the resource utilization set to obtain D entropy values, and the weight of each resource in the D resources is obtained according to the D entropy values.

[0083] Specifically, resource utilization is the ratio of occupied resources to total resources. The resource utilization set can be a D*M order matrix, for example, P = [p xy ] D*M .

[0084] The resource utilization set is normalized according to the type of each resource in the D types of resources to obtain the first target matrix; the proportion of each node under each resource in the D types of resources to the resource index is determined by the first target matrix to obtain the second target matrix; the entropy value of each resource in the D types of resources is determined by the second target matrix to obtain the D entropy value.

[0085] Specifically, the first target matrix Q = [b xy ] D*MIt is a matrix obtained by normalizing each data in the resource utilization set under each resource, and calculating the proportion of the normalized value in each resource under each resource, and obtaining the second target matrix F = [f xy ] D*M Afterwards, the entropy value ex of each resource indicator is calculated according to the second target matrix, where x=1, 2, ..., d, ..., D.

[0086] By normalizing the data in the resource utilization set, the data of different resource indicator dimensions can be unified to the same scale, avoiding the impact of the dimension and range differences of data of different dimensions on the results, thereby improving the comparability and interpretability of the data. At the same time, the entropy weight calculated after data normalization can more accurately reflect the importance of each resource indicator to the decision-making result, effectively avoiding the deviation and error caused by different dimensions of data, thereby improving the accuracy and reliability of decision-making.

[0087] As an optional implementation, after obtaining D entropy values, entropy redundancy processing is performed on each of the D entropy values ​​to obtain D entropy redundancies; and the weights of various resources in the D resources are determined by the D entropy redundancies.

[0088] Specifically, the weight of each resource of D resources can be obtained by the following formula:

[0089]

[0090] where w x is the weight of the xth resource indicator, d xi is the entropy redundancy of the x-th resource indicator of the ith node, It is the sum of the entropy redundancy of resource indicators on the i-th node.

[0091] Finally, the selection index of the i-th node is determined based on the idle status of the i-th candidate node, the weight of each resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirements, and the total number of all candidate nodes.

[0092] The target node selection method provided in this embodiment obtains the container scheduling requirements, wherein the container scheduling requirements include multiple resources required for container scheduling, and the demand for each resource. Then, a screening is performed according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes. Then, the total amount of resources and resource occupancy of each resource in the multiple resources on each candidate node are obtained respectively. In addition, the number of containers deployed on the candidate node and the number of containers to be deployed are counted. Through the above data, the resource occupancy of each candidate node and the number of containers that can be deployed can be known, so as to facilitate the subsequent measurement of which candidate node to allocate the container to be scheduled. Specifically, it is based on the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of resources of the d-th resource in the multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirements, and the total number of all candidate nodes, to determine the selection index of the i-th node, wherein the weight value of each resource is a measure of the importance of the resource. After obtaining the selection index of each candidate node through the above data, the optimal node can be selected to complete the container scheduling requirements according to the selection index corresponding to all candidate nodes. Through the above method, resources can be reasonably allocated, resource utilization can be maximized, system load can be reduced, system response speed and performance can be improved, and system stability and reliability can be improved. In addition, user satisfaction can be improved and user waiting time can be reduced.

[0093] In this embodiment, a target node selection method is provided, which can be used in the above-mentioned mobile terminal, such as a mobile phone, a tablet computer, etc. Figure 2 is a flow chart of a target node selection method provided by an embodiment of the present invention, such as Figure 2 As shown, on the basis of the foregoing embodiment, according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight values ​​corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource on the i-th candidate node, the demand for the d-th resource included in the container scheduling demand, and the total number of all candidate nodes, the selection index of the i-th node is determined, which can be achieved in the following way:

[0094] Step S201, determining the container deployment idle rate of the i-th candidate node according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed.

[0095] Specifically, as described above, the load of the container on the i-th candidate node can be determined based on the number of containers already deployed on the i-th candidate node and the number of containers to be deployed. For details, see the following expression:

[0096]

[0097] Among them, p i Indicates the number of Pods deployed on the i-th candidate node. For example, if the number of containers to be deployed this time is 1, C represents the load of the container on the i-th candidate node.

[0098] The expression for the container deployment idle rate of the i-th candidate node is as follows:

[0099]

[0100] Among them, c i Represents the container deployment idle rate of the i-th candidate node.

[0101] Step S202: Determine the resource idle rate of the dth resource corresponding to the i-th candidate node according to the total amount of resources of the dth resource type on the i-th candidate node, the resource occupancy of the dth resource on the i-th candidate node, and the demand for the dth resource included in the container scheduling requirement.

[0102] Specifically, the calculation of resource idle rate can be expressed by referring to the following formula:

[0103]

[0104] in, Indicates the deployment of the i-th candidate node p d The idle rate of the dth resource.

[0105] In an optional example, considering that in certain time periods, such as peak resource occupancy periods, if the resource idle rate is directly calculated in the above manner, and the target node is selected according to the calculated resource idle rate combined with other data to complete container scheduling, it may be difficult to cope with the peak resource occupancy or impossible to cope with the peak resource occupancy. Therefore, in an embodiment of the present application, when calculating the resource idle rate, it can also include: reserving buffer resources and then calculating the resource idle rate. Specific reserved buffer resources can be reserved according to a certain ratio.

[0106] The specific ratio determination method can be based on historical data, such as analyzing historical resource usage data to find out the peak hours and peak values ​​of resource usage, and determining the appropriate buffer ratio based on these data. Alternatively, the necessary buffer ratio can be determined based on business needs, such as the SLA (service level agreement) requirements of the service.

[0107] In an optional example, the buffer ratio is calculated as follows:

[0108] First, use monitoring tools (such as Prometheus, Grafana, etc.) to collect system resource usage data in real time, and record resource usage peaks regularly (such as every minute). Then, multiply the resource usage peak by the preconfigured buffer factor to get a preliminary buffer ratio. Compare this preliminary ratio with the set threshold, and take the larger value of the two as the actual buffer ratio. Finally, compare this actual buffer ratio with the maximum buffer limit, and take the smaller value of the two as the final buffer ratio. In short, it is to ensure that the buffer ratio is neither lower than the minimum value calculated by the peak value and the usage factor, nor exceeds the maximum allowed buffer amount.

[0109] Finally, the buffer resource margin is calculated based on the buffer ratio and the total resource amount of the d types of resources, and after being subtracted as the minuend in the above formula, the final resource idle rate is calculated.

[0110] Step S203, generating an evaluation index for the i-th node according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes.

[0111] In an optional example, the method steps can be implemented as follows, see Figure 3 As shown, including:

[0112] Step S301, generating a resource idle index corresponding to the i-th candidate node according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node.

[0113] Specifically, the expression of resource idle indicator is as follows:

[0114]

[0115] Among them, a d represents the weight value corresponding to the d-th resource, v i represents the resource idle index corresponding to the i-th candidate node, a p is the weight of the load of the deployed container, where a p The calculation method is the same as a d The calculation method of a p +a1+a2+…+a d +…=1, where a1+a2+…+a d +…, represents the sum of the weights of d types of resources.

[0116] Step S302, generating a feedback scoring index corresponding to the i-th candidate node according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes.

[0117] Specifically, the feedback scoring index is expressed as follows:

[0118]

[0119] in,

[0120]

[0121] b i Indicates the feedback rating indicator; represents the average idle rate of the dth resource on all m candidate nodes that meet the resource requirements; c v Represents the average idle rate of Pod resources. Therefore, It can be used to measure the imbalance of the idle rate of the dth resource of node i in the entire cluster. The larger the ratio, the more idle the dth resource of the node is, the more unbalanced the resource utilization is, and thus the larger the feedback score indicator is. Indicates the imbalance of Pod factors. The larger the value, the greater the feedback score indicator.

[0122] Step S303: Generate the selection index of the i-th node according to the resource idle index and feedback score index corresponding to each candidate node.

[0123] Specifically, the final selection index is shown in the following expression:

[0124] f i =v i +b i (Formula 9)

[0125] Selecting a target node from multiple candidate nodes based on the evaluation index of each candidate node actually means selecting the candidate node with the largest value of the final evaluation index as the target node to meet the container scheduling requirements.

[0126] For a specific example, see Figure 4As shown in the figure, for example, based on the application deployment and scheduling experiment on 9 servers using Kubernetes 1.22.0, Docker 26.1.0 and NVIDIA docker2 are installed on all 9 servers. One server is used as the Master node, and the other 8 servers are used as Node nodes. In order to make full use of the Master node resources, it is set to Pod schedulable, that is, Pod can also be allocated to the Master node and run. The overall process of Kubernetes resource scheduling is as follows:

[0127] Get the usage of various resources of Node nodes in the cluster;

[0128] Filter out inappropriate Node nodes in the pre-selection stage;

[0129] The weights of all Node nodes are calculated using the above method;

[0130] The final selection index is obtained by the aforementioned method;

[0131] Select the Node corresponding to the maximum selection index and schedule the pod to the node.

[0132] The above method flow has been introduced in detail in the previous article, so it will not be repeated here.

[0133] In the present embodiment, a target node selection device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0134] This embodiment provides a target node selection device, such as Figure 5 As shown, the device includes: an acquisition module 501, a traversal module 502, a statistical module 503, a processing module 504, and a selection module 505.

[0135] The acquisition module 501 is used to acquire the container scheduling requirements, which include multiple resources and the required amount of each resource;

[0136] A traversal module 502 is used to traverse multiple nodes in the cluster according to a pre-selection strategy corresponding to the container scheduling requirement to obtain multiple candidate nodes;

[0137] The acquisition module 501 is further used to respectively acquire the total amount of resources and resource occupancy status of each resource among the multiple resources on each candidate node;

[0138] A statistics module 503 is used to count the number of containers that have been deployed and the number of containers to be deployed on all candidate nodes;

[0139] The processing module 504 is used to determine the selection index of the i-th node according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling demand, and the total number of all candidate nodes, wherein the i-th node is any candidate node among multiple candidate nodes, and i and d are both positive integers;

[0140] The selection module 505 is used to select a target node from multiple candidate nodes according to the selection index of each candidate node to meet the container scheduling requirements.

[0141] In an optional example, the processing module 504 is specifically configured to determine the container deployment idle rate of the i-th candidate node according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed;

[0142] Determine the resource idle rate of the dth resource corresponding to the ith candidate node according to the total amount of resources of the dth resource type on the ith candidate node, the resource occupancy of the dth resource on the ith candidate node, and the demand for the dth resource included in the container scheduling requirement;

[0143] The selection index of the i-th node is generated according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes.

[0144] In an optional example, the processing module 504 is specifically used to generate a resource idle index corresponding to the i-th candidate node according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node;

[0145] Generate a feedback scoring indicator corresponding to the i-th candidate node based on the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes;

[0146] According to the resource idle index and feedback score index corresponding to each candidate node, the selection index of the i-th node is generated.

[0147] In an optional example, the container deployment idle rate of the i-th candidate node is determined according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed, which is expressed by the following formula:

[0148]

[0149] Among them, c i is the container deployment idle rate on the i-th candidate node, p i is the number of containers deployed on the i-th candidate node, n is the number of containers to be deployed; c is the total number of containers, and the expression of c is as follows: Among them, m is the total number of candidate nodes.

[0150] In an optional example, according to the total amount of resources of the dth resource type on the ith candidate node, the resource occupancy of the dth resource on the ith candidate node, and the demand for the dth resource included in the container scheduling requirement, the resource idle rate of the dth resource corresponding to the ith candidate node is determined, which is expressed by the following expression:

[0151]

[0152] in represents the resource idle rate after the demand for the dth resource is deployed for the ith node, sd i It is represented as the total amount of resources of the dth resource type owned by the i-th candidate node, rd i It is represented as the resource occupancy of the ith candidate node for the dth resource type, p d It represents the demand for the dth resource included in the container scheduling requirement.

[0153] In an optional example, according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node, a resource idle index corresponding to the i-th candidate node is generated, which is expressed by the following expression:

[0154]

[0155] Among them, v i It is represented as the resource idle index corresponding to the i-th candidate node, a d Expressed as the weight value corresponding to the d-th resource, c i is the container deployment idle rate of the i-th candidate node, D is the total number of resource types, and a p is the load weight of the deployed container.

[0156] In an optional example, according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes, a feedback scoring indicator corresponding to the i-th candidate node is generated, which is expressed by the following expression:

[0157]

[0158] Among them, b i is represented by the feedback score index corresponding to the i-th candidate node, D is represented by the total number of resource types, and a d It is expressed as the weight value corresponding to the d-th resource, It represents the resource idle rate after the demand for the d-th resource is deployed for the ith node, It is expressed as the average idle rate of the dth resource on all candidate nodes, c i It is represented as the container deployment idle rate of the i-th candidate node, c v It is expressed as the average idle rate of container resources on all candidate nodes, a p is the load weight of the deployed container.

[0159] The target node selection device in this embodiment is presented in the form of a functional module, where the module refers to an application specific integrated circuit (ASIC), a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0160] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0161] A target node selection device provided by an embodiment of the present invention obtains a container scheduling requirement, wherein the container scheduling requirement includes multiple resources required for container scheduling and the demand for each resource. Then, a screening is performed according to the pre-selection strategy corresponding to the container scheduling requirement to obtain multiple candidate nodes. Then, the total amount of resources and resource occupancy of each resource in the multiple resources on each candidate node are obtained respectively. In addition, the number of containers deployed on the candidate node and the number of containers to be deployed are counted. Through the above data, the resource occupancy of each candidate node and the number of containers that can be deployed can be known, so as to facilitate the subsequent measurement of which candidate node to allocate the container to be scheduled. Specifically, the selection index of the i-th node is determined according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of resources of the d-th resource in the multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, wherein the weight value of each resource is a measure of the importance of the resource. After obtaining the selection index of each candidate node through the above data, the optimal node can be selected to complete the container scheduling requirements according to the selection index corresponding to all candidate nodes. Through the above method, resources can be reasonably allocated, resource utilization can be maximized, system load can be reduced, system response speed and performance can be improved, and system stability and reliability can be improved. In addition, user satisfaction can be improved and user waiting time can be reduced.

[0162] The embodiment of the present invention also provides a computer device having the above Figure 5 The target node selection device is shown.

[0163] See also Figure 6 , Figure 6 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 6 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0164] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include an integrated circuit. The integrated circuit may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0165] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0166] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the use of a computer device based on the presentation of a small program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0167] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0168] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 6 The example of connecting through bus is taken in the following.

[0169] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0170] The embodiment of the present invention also provides a computer-readable storage medium. The method provided in the above embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or is implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium and downloaded through a network, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0171] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0172] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A target node selection method, characterized in that: The method comprises: Obtaining container scheduling requirements, where the container scheduling requirements include multiple resources and the required amount of each resource; Traverse multiple nodes in the cluster according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes; Respectively obtaining the total amount of resources and resource occupancy status of each of the multiple resources on each of the candidate nodes; Count the number of containers that have been deployed and the number of containers to be deployed on all the candidate nodes; Determine the selection index of the i-th node according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, wherein the i-th node is any one of the multiple candidate nodes, and i and d are both positive integers; According to the selection index of each candidate node, a target node is selected from the plurality of candidate nodes to meet the container scheduling requirement.

2. The method according to claim 1, characterized in that The selection index of the i-th node is determined according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight values ​​corresponding to each pre-acquired resource, the total amount of the d-th resource among the multiple resources on the i-th candidate node, the resource occupancy of the d-th resource on the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, specifically including: Determine the container deployment idle rate of the i-th candidate node according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed; Determine a resource idle rate of the dth resource corresponding to the ith candidate node according to the total amount of resources of the dth resource type on the ith candidate node, the resource occupancy of the dth resource on the ith candidate node, and the demand for the dth resource included in the container scheduling requirement; The selection index of the i-th node is generated according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes.

3. The method according to claim 2, characterized in that According to the weight values ​​corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes, the selection index of the i-th node is generated, which specifically includes: Generate a resource idle index corresponding to the i-th candidate node according to the weight values ​​corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node; Generate a feedback scoring indicator corresponding to the i-th candidate node according to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes; The selection index of the i-th node is generated according to the resource idle index and the feedback scoring index respectively corresponding to each of the candidate nodes.

4. The method according to claim 2 or 3, characterized in that: The container deployment idle rate of the i-th candidate node is determined according to the number of containers already deployed on the i-th candidate node and the number of containers to be deployed, which is expressed by the following formula: Among them, c i is the container deployment idle rate on the i-th candidate node, p i is the number of containers deployed on the i-th candidate node, n is the number of containers to be deployed; c is the total number of containers, and the expression of c is as follows: Among them, m is the total number of candidate nodes.

5. The method according to claim 4, characterized in that The resource idle rate of the dth resource corresponding to the ith candidate node is determined according to the total amount of resources of the dth resource type on the ith candidate node, the resource occupancy of the dth resource on the ith candidate node, and the demand for the dth resource included in the container scheduling requirement, and is expressed by the following expression: in, represents the resource idle rate after the demand for the dth resource is deployed for the i-th node, sd i It is represented as the total amount of resources of the dth resource type owned by the i-th candidate node, rd i It is represented as the resource occupancy of the ith candidate node for the dth resource type, p d It represents the demand for the dth resource included in the container scheduling requirement.

6. The method according to claim 5, characterized in that According to the weight values ​​corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, and the container deployment idle rate of the i-th candidate node, a resource idle index corresponding to the i-th candidate node is generated, which is represented by the following expression: Among them, v i It is represented as the resource idle index corresponding to the i-th candidate node, a d It is represented as the weight value corresponding to the d-th resource, c i is the container deployment idle rate of the i-th candidate node, D is the total number of resource types, a p The weight of the load of the deployed container.

7. The method according to claim 4, characterized in that According to the weight value corresponding to each resource, the resource idle rate of each resource corresponding to the i-th candidate node, the container deployment idle rate of the i-th candidate node, and the total number of all candidate nodes, a feedback scoring index corresponding to the i-th candidate node is generated, which is expressed by the following expression: Among them, b i is represented by the feedback score index corresponding to the i-th candidate node, D is represented by the total number of resource types, a d is represented by the weight value corresponding to the d-th resource, represents the resource idle rate after the demand for the d-th resource is deployed for the i-th node, It is expressed as the average idle rate of the dth resource on all candidate nodes, c i It is represented as the container deployment idle rate of the i-th candidate node, c v It is expressed as the average idle rate of container resources on all candidate nodes.

8. A target node selection device, characterized in that: The device comprises: An acquisition module is used to acquire container scheduling requirements, where the container scheduling requirements include multiple resources and the required amount for each resource; The traversal module is used to traverse multiple nodes in the cluster according to the pre-selection strategy corresponding to the container scheduling requirements to obtain multiple candidate nodes; The acquisition module is further used to respectively acquire the total amount of resources and resource occupancy status of each of the multiple resources on each of the candidate nodes; A statistics module is used to count the number of containers that have been deployed and the number of containers to be deployed on all the candidate nodes; A processing module, used to determine the selection index of the i-th node according to the number of containers deployed on the i-th candidate node, the number of containers to be deployed, the weight value corresponding to each pre-acquired resource, the total amount of the d-th resource among multiple resources on the i-th candidate node, the resource occupancy of the d-th resource by the i-th candidate node, the demand for the d-th resource included in the container scheduling requirement, and the total number of all candidate nodes, wherein the i-th node is any one of the multiple candidate nodes, and i and d are both positive integers; The selection module is used to select a target node from a plurality of the candidate nodes according to the selection index of each of the candidate nodes, so as to meet the container scheduling requirements.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the target node selection method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the target node selection method according to any one of claims 1 to 7.

Images