HK40089547A - Method, apparatus, device, and storage medium for determining number of copies - Google Patents

Description

A method, apparatus, device, and storage medium for determining the number of copies.

Technical Field

This application relates to the field of data processing technology, and in particular to a method, apparatus, device, and storage medium for determining the number of copies.

Background Technology

With the rapid development of technologies such as cloud computing and distributed storage, the research on Kubernetes cluster architecture systems has become a hot topic in the field of Internet communication technology due to their low cost, high efficiency, and lightweight testing characteristics. Kubernetes container cluster management tools are mainly used to manage computationally intensive large-scale task applications in cloud computing environments. However, in order to meet the computing resources required by container clusters, it is usually necessary to add host nodes to the cluster. Since there is an upper limit to the number of computing nodes that a single cluster can manage, it is necessary to continuously add clusters to ensure that the cluster can be quickly switched to a normal cluster in case of cluster failure, so as to make the container cluster highly available and meet the resource requirements of the container cluster.

However, currently, cluster deployment is usually chosen by the application itself. Before the multi-cluster scheduler schedules the clusters, the application does not know the specific resource usage of the clusters. In reality, some clusters may have sufficient remaining resources, while others may have scarce resources. If the application chooses to create too many replicas of a cluster with scarce resources, it will lead to a situation where the cluster with scarce resources does not have enough resources to create replicas and cannot produce, thus failing to meet the resource requirements of the container cluster and resulting in an unbalanced cluster resource load.

Summary of the Invention

This application provides a method, apparatus, device, and storage medium for determining the number of replicas. It is used to obtain the total number of first target replicas of a first cluster by selecting the number of first target replicas of each first target node in the first cluster. It can accurately perceive the total number of first target replicas that the first cluster can support creation before the multi-cluster scheduler schedules, and can avoid the multi-cluster scheduler from allocating more than the total number of first target replicas to the first cluster for production, thereby achieving cluster resource load balancing.

One embodiment of this application provides a method for determining the number of copies, including:

Obtain the replica node configuration information and replica resource configuration information of the first cluster, which includes several nodes;

Based on the replica node configuration information, a filtering operation is performed on several nodes in the first cluster to obtain at least one first target node;

Obtain the used resources and total resources of the smallest cluster scheduling unit in each first target node;

Calculate the remaining resources for each first target node based on the amount of resources used and the total amount of resources;

Based on the quantity calculation strategy, the remaining resource quantity and the resource quantity in the replica resource configuration information are calculated to obtain the number of first target replicas for each first target node;

The total number of first target replicas for each first target node is summed to obtain the total number of first target replicas for the first cluster.

This application, in another aspect, provides an apparatus for determining the number of copies, comprising:

The acquisition unit is used to acquire the replica node configuration information and replica resource configuration information of the first cluster, which includes several nodes;

The processing unit is used to perform a filtering operation on several nodes of the first cluster based on the replica node configuration information to obtain at least one first target node.

The acquisition unit is also used to acquire the amount of resources used and the total amount of resources of the smallest cluster scheduling unit in each first target node;

The processing unit is also used to calculate the remaining resources of each first target node based on the amount of resources used and the total amount of resources.

The processing unit is also used to calculate the number of first target replicas for each first target node based on the quantity calculation strategy, the remaining resource quantity and the resource quantity in the replica resource configuration information.

The processing unit is also used to summarize the number of first target replicas for each first target node to obtain the total number of first target replicas for the first cluster.

In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

Based on the selector configuration data, the first candidate node is selected from several nodes in the first cluster;

Based on affinity configuration data, select second candidate nodes from several first candidate nodes;

Based on the taint and tolerance configuration data, a third candidate node is selected from several second candidate nodes;

The node that meets the scheduling conditions among several third candidate nodes is selected as the first target node.

In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

Obtain the node status information of several third candidate nodes;

If the node status information is in a runnable and schedulable state, then the third candidate node will be selected as the first target node.

In one possible design, in another implementation of the embodiments of this application,

The acquisition unit is also used to acquire the number of running and total number of the smallest cluster scheduling units in each first target node;

Specifically, the processing unit can be used to: calculate the number of first target replicas for each first target node based on the quantity calculation strategy, the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of already run replicas, and the total number of replicas.

In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

Calculate the ratio between the remaining resource quantity and the resource quantity in the replica resource configuration information to obtain the number of at least one first replica;

Calculate the ratio between the number of already run replicas and the total number of replicas to obtain the number of second replicas;

Select the smallest value from at least one first replica quantity and a second replica quantity as the first target replica quantity.

In one possible design, in another implementation of the embodiments of this application,

The acquisition unit is also used to acquire the total number of second target replicas in the second cluster out of at least two clusters;

The processing unit is also used to calculate the ratio between the total number of the first target replicas and the total number of the second target replicas, so as to obtain the replica allocation weights for at least two clusters.

In one possible design, in another implementation of the embodiments of this application,

The receiving unit is used to receive cluster scheduling requests, wherein the cluster scheduling request carries the total number of replicas to be produced;

The processing unit is also used to divide the total number of replicas to be produced into the first number of replicas to be produced in the first cluster and the second number of replicas to be produced in the second cluster according to the replica allocation weight.

The processing unit is also used to allocate the first number of replicas to be produced and the second number of replicas to be produced to the first cluster and the second cluster respectively for replica production.

Another aspect of this application provides a computer device, including: a memory, a transceiver, a processor, and a bus system;

The memory is used to store programs;

The processor implements the methods described above when executing a program in memory;

Bus systems are used to connect memory and processor to enable communication between them.

Another aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described above.

As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

By obtaining the replica node configuration information and replica resource configuration information of the first cluster, at least one first target node can be selected from several nodes in the first cluster based on the replica node configuration information. The used resources and total resources of the smallest cluster scheduling unit in each first target node can be obtained. Based on the used resources and total resources, the remaining resources of each first target node can be calculated. Then, according to the quantity calculation strategy, the remaining resources and each resource quantity in the replica resource configuration information can be calculated to obtain the number of first target replicas for each first target node. The number of first target replicas for each first target node is then summed to obtain the total number of first target replicas in the first cluster. Through this method, at least one first target node can be selected from the first cluster using the obtained replica node configuration information, and the number of first target replicas for each first target node can be calculated using the replica resource configuration information to obtain the total number of first target replicas in the first cluster. This allows for accurate perception of the total number of first target replicas that the first cluster can support creation before multi-cluster scheduler scheduling, preventing the multi-cluster scheduler from allocating more replicas than the total number of first target replicas to the first cluster for production, thereby achieving cluster resource load balancing.

Attached Figure Description

Figure 1 is a schematic diagram of the architecture of a copy data control system in an embodiment of this application;

Figure 2 is a flowchart of an embodiment of the method for determining the number of copies in this application;

Figure 3 is a flowchart of another embodiment of the method for determining the number of copies in this application;

Figure 4 is a flowchart of another embodiment of the method for determining the number of copies in this application;

Figure 5 is a flowchart of another embodiment of the method for determining the number of copies in this application;

Figure 6 is a flowchart of another embodiment of the method for determining the number of copies in this application;

Figure 7 is a flowchart of another embodiment of the method for determining the number of copies in this application;

Figure 8 is a flowchart of another embodiment of the method for determining the number of copies in this application;

Figure 9 is a flowchart of a principle embodiment of the method for determining the number of copies in this application;

Figure 10 is a schematic diagram of an embodiment of the device for determining the number of copies in this application;

Figure 11 is a schematic diagram of one embodiment of the computer device in this application.

Detailed Implementation

This application provides a method, apparatus, device, and storage medium for determining the number of replicas. It is used to obtain the total number of first target replicas of a first cluster by selecting the number of first target replicas of each first target node in the first cluster. It can accurately perceive the total number of first target replicas that the first cluster can support creation before the multi-cluster scheduler schedules, and can avoid the multi-cluster scheduler from allocating more than the total number of first target replicas to the first cluster for production, thereby achieving cluster resource load balancing.

The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

With the rapid development of information technology, cloud technology is gradually permeating all aspects of people's lives. Cloud technology is a general term encompassing network technology, information technology, integration technology, management platform technology, and application technology based on the cloud computing business model. It can form resource pools, providing on-demand, flexible, and convenient access. Cloud computing technology will become a crucial support. Backend services of technical network systems require substantial computing and storage resources, such as video websites, image websites, and many portal websites. With the rapid development and application of the internet industry, every item may have its own identification mark in the future, requiring data to be transmitted to backend systems for logical processing. Data at different levels will be processed separately, and various industry data will require robust system support, which can only be achieved through cloud computing.

Cloud security refers to the collective term for security software, hardware, users, organizations, and security cloud platforms based on cloud computing business models. Cloud security integrates emerging technologies and concepts such as parallel processing, grid computing, and unknown virus behavior detection. It uses a large network of clients to monitor abnormal software behavior on the network, obtain the latest information on Trojans and malware on the internet, and send it to the server for automatic analysis and processing. Finally, solutions for viruses and Trojans are distributed to each client. The method for determining the number of copies provided in this application can be implemented using cloud computing technology and cloud security technology.

The following describes some of the concepts involved in the embodiments of this application.

1. Workload

A workload is a type of application that can contain multiple replica instances.

2. Copy

The instance unit of a workload, each replica instance is an independent container.

3. Kubernetes

A production-grade, large-scale container orchestration and scheduling system for the automated deployment, scaling, and operation and maintenance of container clusters.

4. Container

Containers generally refer to Docker containers. Through their isolation features, they decouple from the host machine, allowing services to run independently of it and without interference. Docker containers are very lightweight. Kubernetes, on the other hand, manages all Docker containers within the service, creating, running, restarting, and deleting them.

5. Pod

A Pod is a collection of containers and is the smallest unit of cluster scheduling. A Pod represents a copy of a workload.

It should be understood that the method for determining the number of replicas provided in this application can be applied to fields such as cloud technology, artificial intelligence, and intelligent transportation, for scenarios such as maintaining cluster load balancing based on replica scheduling. For example, determining the number of replicas for each cluster can be used to schedule the workload of the cluster. As another example, determining the number of replicas for each cluster can help the cluster scheduler perceive the maximum number of replicas that each cluster can produce. As yet another example, determining the number of replicas for each cluster can be used to allocate corresponding replica production tasks to each cluster.

To address the aforementioned issues, this application proposes a method for determining the number of replicas. This method is applied to the replica data control system shown in Figure 1. Figure 1 is a schematic diagram of the architecture of the replica data control system in an embodiment of this application. As shown in Figure 1, the server obtains the replica node configuration information and replica resource configuration information of the first cluster. Based on the replica node configuration information, at least one first target node can be selected from several nodes in the first cluster. The server can obtain the used resources and total resources of the smallest cluster scheduling unit in each first target node, and calculate the remaining resources of each first target node based on the used resources and total resources. Then, based on the quantity calculation strategy, the remaining resources and each resource in the replica resource configuration information are calculated to obtain the number of first target replicas for each first target node. Finally, the number of first target replicas for each first target node is summarized to obtain the total number of first target replicas of the first cluster. Using the above method, at least one first target node can be selected from the first cluster by obtaining the replica node configuration information, and the number of first target replicas of each first target node can be calculated by using the replica resource configuration information to obtain the total number of first target replicas of the first cluster. This allows for accurate perception of the total number of first target replicas that the first cluster can support before scheduling by the multi-cluster scheduler, which can prevent the multi-cluster scheduler from allocating more replicas than the total number of first target replicas to the first cluster for production, thereby achieving cluster resource load balancing.

It is understandable that Figure 1 only shows one type of terminal device. In real-world scenarios, many more types of terminal devices can participate in the data processing. These terminal devices include, but are not limited to, mobile phones, computers, smart voice interaction devices, smart home appliances, and in-vehicle terminals. The specific number and types depend on the actual scenario and are not limited here. Additionally, while Figure 1 shows one server, multiple servers can participate in real-world scenarios, especially in multi-model training and interaction scenarios. The number of servers depends on the actual scenario and is not limited here.

It should be noted that in this embodiment, the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Terminal devices and servers can be directly or indirectly connected via wired or wireless communication, and terminal devices and servers can be connected to form a blockchain network; this application does not impose any limitations on this.

To address the aforementioned issues, this application proposes a method for determining the number of copies. This method is generally executed by a server or terminal device, and correspondingly, the device for determining the number of copies is generally located in the server or terminal device.

It is understood that, as disclosed in this application, the method, apparatus, device, and storage medium for determining the number of copies can form a blockchain, where multiple servers or terminal devices can constitute nodes on the blockchain. In practical applications, data sharing between nodes may be required within the blockchain, and each node can store copy data, node data, etc.

The method for determining the number of copies in this application will be described below. Please refer to Figure 2. One embodiment of the method for determining the number of copies in this application includes:

In step S101, the replica node configuration information and replica resource configuration information of the first cluster are obtained. The first cluster includes several nodes.

In this embodiment, when it is necessary to schedule the cluster to produce replicas, the replica node configuration information and replica resource configuration information of the pre-scheduled cluster, i.e. the first cluster, can be obtained first, so that subsequent resource scheduling can be performed based on the replica node configuration information and replica resource configuration information to maintain the load balance of the cluster.

The first cluster refers to a cluster containing several nodes, which can specifically be a cluster running within a container cluster management tool. This tool can be Kubernetes, used for automated deployment, scaling, and management of container clusters; other cluster management tools can also be used, without specific limitations. Replica node configuration information refers to the scheduling requirements for nodes in the first cluster, specifically including selector requirements, affinity requirements, taint and tolerance requirements, etc., or other node requirements, without specific limitations. Replica resource configuration information refers to the resource requirements for replicas produced by the nodes in the first cluster, specifically including CPU core count requirements, memory size requirements, and expansion resource requirements, or other resource requirements, without specific limitations.

Specifically, as shown in Figure 9, the target object can pre-upload the cluster it pre-schedules to use (i.e., the first cluster), the replica node requirements (i.e., replica node configuration information), and the replica resource requirements (i.e., replica resource configuration information) for the node to produce replicas to the platform database via the terminal device. When the target object needs to schedule the cluster for replica production, it can retrieve the pre-scheduled cluster, the corresponding replica node configuration information, and the replica resource configuration information from the database based on the target object's object identifier. The object identifier (identity ID) indicates the target object and can be represented as an integer (int) string or a string, etc. It is understood that in this embodiment and subsequent embodiments, the target object refers to the user of the terminal device used; or Alternatively, when the target object needs to schedule a cluster for replica production, the cluster scheduling operation can be performed through a client installed on the terminal device. The client can receive the target object's selection operation for the target cluster to be scheduled, and generate an input interface for the cluster replica estimation instruction based on the target object's selection operation, so as to obtain the cluster replica estimation instruction input by the target object, as shown in Figure 9. This allows the server to index to the target cluster, i.e., the first cluster, based on the cluster identifier or cluster name in the received cluster replica estimation instruction. Specifically, this can be done by traversing several clusters that are consistent with the cluster identifier or cluster name, or by other methods, which are not specifically limited here. The server can also read the replica node configuration information and replica resource configuration information corresponding to the cluster identifier or cluster name from the database, or obtain the replica node configuration information and replica resource configuration information by other methods, which are not specifically limited here.

Understandably, when the server receives the replica node configuration information and replica resource configuration information uploaded by the target object, it can perform format conversion processing on the received replica node configuration information and replica resource configuration information according to the preset data storage format, so that the computer can quickly read the replica node configuration information and replica resource configuration information.

In step S102, a filtering operation is performed on several nodes of the first cluster according to the replica node configuration information to obtain at least one first target node;

In this embodiment, after obtaining the replica node configuration information of the first cluster, at least one first target node required by several nodes of the first cluster can be selected from several nodes of the first cluster according to the replica node configuration information, so that the first target node can be scheduled to produce replicas in the future.

Specifically, as shown in Figure 9, after obtaining the replica node configuration information of the first cluster, several nodes in the first cluster can be pre-selected based on the replica node configuration information. Specifically, the selector requirements, affinity requirements, taint and tolerance requirements, etc., contained in the replica node configuration information can be matched with the node labels of each node. If a node that meets the selector requirements, affinity requirements, taint and tolerance requirements is matched, it can be understood as a node in the first cluster that can be scheduled and used for replica production, i.e., the first target node. If a node that does not fully meet the selector requirements, affinity requirements, taint and tolerance requirements is matched, it can be understood as a node in the first cluster that can be scheduled and used for replica production, i.e., the first target node. Nodes in the cluster that cannot be scheduled for replica production can be filtered or ignored, and other operations can be performed. There are no specific restrictions here. Node labels are used to represent the attributes of Kubernetes objects that are meaningful to the target object, or to describe metadata, or to filter during command queries. Node labels can be key-value pairs attached to Kubernetes objects, for example, `kubectl label nodes<node-name><label-key>＝<label-value>`, or other label formats. There are no specific restrictions here.

In step S103, the used resources and total resources of the smallest cluster scheduling unit in each first target node are obtained;

In this embodiment, after obtaining at least one first target node in the first cluster, the used resources and total resources of the smallest cluster scheduling unit in each first target node can be obtained, so that the maximum number of replicas that each first target node can produce can be estimated based on the used resources and total resources of the smallest cluster scheduling unit.

Each first target node can contain at least one minimum cluster scheduling unit. A minimum cluster scheduling unit refers to a collection of containers called Pods; one Pod can represent a replica of a workload. The used resources of a minimum cluster scheduling unit refer to the resource requirements of each Pod currently running on each first target node. The total resources refer to the pre-configured total resource specifications of each first target node.

Understandably, Pods can serve as carriers for vertical application integration, supporting local collaboration and management of applications. For example, Pods can be used for content management systems, file and data loading, and local caching; or for log and checkpoint backup, compression, looping, and snapshots; or for data exchange monitoring, log tracing, logging and monitoring adapters, and event publishing; or as proxies, bridges, or adapters; or for controlling, managing, configuring, or updating applications, and may have other uses. No specific restrictions are made here. In general, a standalone Pod will not load multiple instances of the same application.

Specifically, as shown in Figure 9, after obtaining at least one first target node in the first cluster, the thread corresponding to each first target node can be called to first obtain each Pod of the smallest cluster scheduling unit corresponding to each first target node, and find the Pods already running on the first target node from several Pods, denoted as p1, p2, ..., pn. Then, the resource requirements of each running Pod, that is, the amount of resources used by each smallest cluster scheduling unit, and the resource specifications of the first target node, that is, the total amount of resources of the first target node, can be obtained respectively.

For example, suppose that the total resources of a first target node A in the first cluster are 8 CPU cores, 16GB of memory, 8 extended resources (such as GPUs), and a maximum number of supported Pods of 4.

For example, suppose that the minimum cluster scheduling unit Pod A1 of the first target node A in the first cluster has 3 CPU cores and 5GB of memory, and the minimum cluster scheduling unit Pod A2 of the first target node A has 1 CPU core, 6GB of memory, and 4 extended resources (such as GPUs).

In step S104, the remaining resources of each first target node are calculated based on the amount of resources used and the total amount of resources.

In this embodiment, after obtaining the used resources and total resources of the smallest cluster scheduling unit in each first target node, the remaining resources of each first target node can be calculated based on the used resources and total resources, so that the number of first target replicas of each first target node can be estimated based on the remaining resources of each first target node.

Specifically, as shown in Figure 9, after obtaining the used resources and total resources of the smallest cluster scheduling unit in each first target node, the thread corresponding to each first target node can be called to calculate the difference between the used resources and total resources of the smallest cluster scheduling unit in each first target node. Specifically, the following formula (1) can be used to subtract the total resources of each first target node from the resource requirements of each Pod, i.e., the used resources of the smallest cluster scheduling unit, to obtain the remaining resources of each first target node:

Where r_{_s} represents the total resources of each first target node, r _{_pn} represents the used resources of each smallest cluster scheduling unit, r_rl represents the remaining resources of each first target node, and n is the number of pods in each first target node.

For example, suppose that the total resources of a first target node A in the first cluster are 8 CPU cores, 16GB memory, 8 extended resources (such as GPUs) and a maximum number of supported Pods of 4. The used resources of a minimum cluster scheduling unit Pod A1 of a first target node A in the first cluster are 3 CPU cores and 5GB memory, and the used resources of a minimum cluster scheduling unit Pod A2 of a first target node A are 1 CPU core, 6GB memory and 4 extended resources (such as GPUs), etc. Then, the remaining resources of the first target node A can be calculated by formula (1) as 4 CPU cores, 5GB memory, 4 extended resources (such as GPUs) and a maximum number of supported Pods of 2.

In step S105, the remaining resource quantity and the resource quantity in the replica resource configuration information are calculated according to the quantity calculation strategy to obtain the first target replica quantity of each first target node.

In this embodiment, after obtaining the remaining resources of each first target node, the remaining resources and each resource in the replica resource configuration information can be calculated according to the quantity calculation strategy to obtain the number of first target replicas of each first target node.

The quantity calculation strategy can specifically represent quantity calculation rules, constraints, function expressions, or quantity calculation formulas, and can also be represented by other calculation strategies, without specific restrictions here.

Specifically, as shown in Figure 9, after obtaining the replica resource configuration information and remaining resource quantity of each first target node, the ratio between the remaining resource quantity and the resource quantity in the replica resource configuration information can be calculated according to the following formula (2) to obtain the maximum number of replicas that the remaining resource quantity of the first target node can support, i.e., the number of first target replicas:

Where _a1 represents the number of first target replicas for each first target node, _rd represents the resource quantity in the replica resource configuration information of each first target node, and rl represents the remaining resource quantity of each first target node.

For example, suppose the input workload R has the following resource requirements, i.e., the replica resource configuration information is 2 CPU cores and 3GB memory. Suppose the remaining resources of the first target node A are 4 CPU cores, 5GB memory, 4 extended resources (such as GPUs), and a maximum number of supported Pods of 2. According to formula (2), the ratio between the number of CPU cores in the remaining resources and the number of CPU cores in the replica resource configuration information is 2. The ratio between the memory size in the remaining resources and the memory size in the replica resource configuration information is taken as the smaller integer value of approximately 1. Then, the minimum value among the ratios between the remaining resources and the resource sizes in the replica resource configuration information can be taken as the number of the first target replicas, which is 1.

In step S106, the number of first target replicas for each first target node is summed to obtain the total number of first target replicas for the first cluster.

In this embodiment, after obtaining the number of first target replicas for each first target node, the number of first target replicas for each first target node can be summarized to obtain the total number of first target replicas for the first cluster. This allows for accurate perception of the total number of first target replicas that the first cluster can support creation before scheduling by the multi-cluster scheduler, thereby maintaining the load balance of cluster resources.

Specifically, as shown in Figure 9, after obtaining the number of first target replicas for each first target node, the number of first target replicas for each first target node can be summarized. Specifically, the obtained number of first target replicas for each first target node can be summed, or other summation methods such as weighted summation can be used. No specific restrictions are imposed here, in order to obtain the total number of first target replicas for the first cluster.

For example, assuming that the number of first target replicas of the first target node A is 1, the number of first target replicas of the first target node B is 2, and the number of first target replicas of the first target node C is 1, the number of first target replicas of each first target node can be summed to obtain the total number of first target replicas of the first cluster as 4.

In this embodiment of the application, a method for determining the number of replicas is provided. By means of the above method, multiple first target nodes can be selected from the first cluster by obtaining the replica node configuration information, and the number of first target replicas of each first target node can be calculated by means of the replica resource configuration information, so as to obtain the total number of first target replicas of the first cluster. This method can accurately perceive the total number of first target replicas that the first cluster can support creation before the multi-cluster scheduler schedules, and can avoid the multi-cluster scheduler from allocating replicas exceeding the total number of first target replicas to the first cluster for production, thereby achieving cluster resource load balancing.

Optionally, based on the embodiment corresponding to Figure 2 above, in another optional embodiment of the method for determining the number of replicas provided in this application, as shown in Figure 3, the replica node configuration information includes selector configuration data, affinity configuration data, and taint and tolerance configuration data;

Based on the replica node configuration information, a filtering operation is performed on several nodes in the first cluster to obtain at least one primary target node, including:

In step S301, a first candidate node is selected from several nodes in the first cluster according to the selector configuration data;

In step S302, a second candidate node is selected from several first candidate nodes based on the affinity configuration data;

In step S303, a third candidate node is selected from several second candidate nodes based on the taint and tolerance configuration data;

In step S304, the node that meets the scheduling conditions among the several third candidate nodes is selected as the first target node.

In this embodiment, after obtaining the replica node configuration information of the first cluster, a first candidate node can be selected from several nodes of the first cluster based on the selector configuration data in the replica node configuration information. Then, a second candidate node can be selected from several first candidate nodes based on the affinity configuration data in the replica node configuration information. Furthermore, a third candidate node can be selected from several second candidate nodes based on the taint and tolerance configuration data in the replica node configuration information. Finally, the third candidate node that meets the scheduling conditions can be used as the first target node so that the first target node can be scheduled for replica production in the future.

The selector configuration data is used to select labels using node selectors to obtain node labels that satisfy the expressions of the node selectors. Label selection can be performed using equality selection, set selection, `matchLabels`, or `matchExpressions`, or other methods, without specific restrictions. Equality selection can use expressions such as "disktype=ssd&&disksize=big", where equality can be represented by =, ==, or !=. It's understood that = and == are indistinguishable, and the && operator can be used for multiple requirements (such as multiple labels). Set selection supports operators such as in, notin, and exists. `matchLabels` is a mapping composed of {key, value} pairs. A single {key, value} in a `matchLabels` mapping is equivalent to an element in a `matchExpressions`, where the key field is "key", the operator is "In", and the values array only contains "value". matchExpressions can be understood as a list of operators that a Pod needs to select. Valid operators include In, NotIn, Exists, and DoesNotExist. However, when the operator is In or NotIn, the value set must usually be non-empty.

Affinity configuration data is similar to selector configuration data. It can constrain which nodes a pod can be scheduled to based on the labels on the nodes, and can also be used for node filtering. Affinity can be set through the -Affinity field, for example, nodeAffinity. Anti-affinity is set using the -AntiAffinity field, for example, nodeAntiAffinity. Then, affinity set expressions can be configured using operators supported by the node affinity syntax, such as In, NotIn, Exists, DoesNotExist, Gt, and Lt, to filter node labels that match the affinity set.

The taint and tolerance configuration data includes node taints and tolerances. Node taints exclude a specific type of pod, while tolerances indicate whether a pod can tolerate a taint on that pod. Think of it this way: when a taint is added to a node, unless a pod declares it can tolerate that taint, the pod will not be scheduled to that node. Typically, the system tries to avoid scheduling pods to nodes with taints it cannot tolerate, but this is not mandatory. In essence, Kubernetes handles multiple taints and tolerances like a filter: starting from all taints on a node, it iterates through them, filtering out pods with taints that don't match the tolerance.

Specifically, as shown in Figure 9, after obtaining the replica node configuration information of the first cluster, several nodes of the first cluster can be filtered or screened layer by layer according to the replica node configuration information to obtain at least one first target node. Specifically, firstly, according to the replica's node selector requirements, i.e., selector configuration data, nodes whose node labels satisfy the selector are matched from all nodes in the first cluster as first candidate nodes. Then, according to the replica's node affinity requirements, i.e. affinity configuration data, nodes whose node labels satisfy the affinity set expression are matched from several first candidate nodes as second candidate nodes. Then, according to the replica taint and tolerance requirements, i.e. taint and tolerance configuration data, several second candidate nodes are traversed to select nodes whose taint matches the replica tolerance as third candidate nodes. Finally, nodes among several third candidate nodes that meet the scheduling conditions, such as being runnable and schedulable, can be selected as the first target node.

Optionally, based on the embodiment corresponding to Figure 3 above, in another optional embodiment of the method for determining the number of replicas provided in this application, as shown in Figure 4, the node that meets the scheduling conditions among a plurality of third candidate nodes is taken as the first target node, including:

In step S401, the node status information of several third candidate nodes is obtained;

In step S402, if the node status information is in a runnable and schedulable state, then the third candidate node is selected as the first target node.

In this embodiment, after obtaining several third candidate nodes of the first cluster, the node status information of each of the several third candidate nodes can be obtained. If the node status information is in a runnable and schedulable state, the third candidate node can be used as the first target node. The first target node that can meet the subsequent scheduling production replica requirements can be quickly selected from several third candidate nodes through the node status information, thereby improving the cluster resource scheduling efficiency to a certain extent.

Among them, node status information refers to the current running, scheduling and resource usage status of each node in the first cluster. Node status information can be represented as node status labels or node status lists, or other node status markers, without specific restrictions here.

Specifically, after obtaining several third candidate nodes of the first cluster, the node status information of these third candidate nodes is obtained. This can be done by calling a monitoring interface to obtain the real-time monitoring node status label of each of the several third candidate nodes, or by reading the node status data of each of the several third candidate nodes from the node status list corresponding to the first cluster, or by using other methods. No specific restrictions are imposed here. Then, based on the scheduling conditions of being runnable and schedulable, nodes whose current node status information is runnable and schedulable are selected from the several third candidate nodes as the first target nodes. This is to avoid the failure of subsequent replica production due to the third candidate nodes being inoperable or unschedulable, thereby maintaining the cluster resource load balance to a certain extent.

Optionally, based on the embodiment corresponding to Figure 2 above, in another optional embodiment of the method for determining the number of replicas provided in this application embodiment, as shown in Figure 5, before calculating the remaining resource quantity and each resource quantity in the replica resource configuration information according to the quantity calculation strategy to obtain the first target replica quantity of each first target node, the method further includes: step S501, and step S105 includes: step S502.

In step S501, the number of running and the total number of the smallest cluster scheduling units in each first target node are obtained;

In step S502, based on the quantity calculation strategy, the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of already run and the total number are calculated to obtain the number of first target replicas for each first target node.

In this embodiment, after obtaining at least one first target node in the first cluster, the number of running and total number of the smallest cluster scheduling units in each first target node can be obtained. Then, according to the quantity calculation strategy, the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of running and total number can be calculated to obtain the number of first target replicas for each first target node. The maximum number of production replicas that each first target node can support can be further estimated by the number of running and total number of the smallest cluster scheduling units in each first target node, so as to obtain the total number of first target replicas in the first cluster more accurately, thereby improving the resource load balancing of the cluster to a certain extent.

The number of running Pods in the smallest cluster scheduling unit of each first target node refers to the number of Pods already running on the first target node, such as n for p1, p2, ..., pn. The total number of smallest cluster scheduling units in each first target node refers to the number of running Pods and the number of non-running Pods on the first target node.

Specifically, after obtaining at least one first target node in the first cluster, the number of running and the total number of the smallest cluster scheduling unit in each first target node can be obtained. Then, the ratio between the remaining resource quantity and the resource quantity in the replica resource configuration information can be calculated according to the quantity calculation strategy as described in the above formula (2), and the number of running and the total number can be calculated according to the quantity calculation strategy such as difference or ratio, or other calculation strategies, without specific restrictions here, to obtain the corresponding calculation results. Thus, the smallest calculation result value can be selected from multiple calculation results as the number of first target replicas for each first target node.

Optionally, based on the embodiment corresponding to Figure 5 above, in another optional embodiment of the method for determining the number of replicas provided in this application, as shown in Figure 6, according to the quantity calculation strategy, the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of already running replicas, and the total number of replicas are calculated to obtain the first target replica quantity for each first target node, including:

In step S601, the ratio between the remaining resource quantity and each resource quantity in the replica resource configuration information is calculated to obtain at least one first replica quantity;

In step S602, the ratio between the number of copies already run and the total number is calculated to obtain the second number of copies;

In step S603, the smallest value among at least one first replica quantity and a second replica quantity is selected as the first target replica quantity.

In this embodiment, after obtaining the number of running and total number of the smallest cluster scheduling units in each first target node, the ratio between the remaining resource quantity and the resource quantity in the replica resource configuration information can be calculated according to the quantity calculation strategy to obtain at least one first replica quantity. Similarly, the ratio between the number of running and the total number can be calculated to obtain the second replica quantity. Then, the smallest value among the at least one first replica quantity and the second replica quantity is selected as the first target replica quantity. By comparing multiple dimensions such as the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of running and the total number, the maximum number of replicas that each first target node can support for production, i.e., the first target quantity, can be calculated more accurately. This allows for a more accurate acquisition of the total number of first target replicas in the first cluster. Thus, in the multi-cluster scheduling process, by pre-calculating the maximum number of available replicas for the workload of each cluster, i.e., the total number of first target replicas, it is possible to prevent the cluster from being unable to run due to too many replicas of the workload allocated to each cluster. This improves the accuracy of scheduling and can improve the resource load balancing of the cluster to a certain extent.

Specifically, after obtaining the remaining resources, replica resource configuration information, number of running units and total number of units in each first target node, the remaining resources, replica resource configuration information, number of running units and total number can be calculated according to the quantity calculation strategy. Specifically, the above formula (2) can be used to calculate the ratio between the remaining resources and the resource quantities in the replica resource configuration information to obtain at least one first replica quantity. This will not be elaborated here. Similarly, according to the quantity calculation strategy, the following formula (3) can be used to subtract the maximum number of Pods supported by the first target node (i.e., the total number) from the number of Pods already running on the first target node (i.e., the number of running units) to obtain the remaining number of Pods that the first target node can produce as the second replica quantity.

a ₂ = mn (3);

Where _a2 represents the number of second replicas for each first target node, m represents the total number of minimum cluster scheduling units for each first target node, and n represents the number of running minimum cluster scheduling units for each first target node.

Furthermore, after obtaining at least one first replica quantity and a second replica quantity, the at least one first replica quantity and the second replica quantity can be compared pairwise according to the following formula (4) to obtain the one with the smallest value as the first target replica quantity:

a ₃ =min(a ₁ , a ₂ ) (4);

Where _a3 represents the number of first target replicas for each first target node.

Optionally, based on the embodiment corresponding to Figure 2 above, in another optional embodiment of the method for determining the number of replicas provided in this application embodiment, as shown in Figure 7, when there are at least two clusters;

After summing the number of first target replicas for each first target node to obtain the total number of first target replicas for the first cluster, the method further includes:

In step S701, the total number of second target replicas in the second cluster of at least two clusters is obtained;

In step S702, the ratio between the total number of the first target replicas and the total number of the second target replicas is calculated to obtain the replica allocation weights for at least two clusters.

In this embodiment, if there are at least two clusters, after obtaining the first target total number of replicas for the first cluster, the second target total number of replicas for other clusters, such as the second cluster, can be obtained. Then, the replica allocation weights corresponding to at least two clusters can be obtained by calculating the ratio between the first target total number of replicas and the second target total number of replicas. This allows the replica allocation weights corresponding to at least two clusters to be obtained by pre-calculating the target total number of replicas for the workload of each cluster during subsequent multi-cluster scheduling. This enables more accurate and reasonable distribution of replicas to different clusters for replica production, thereby maintaining load balancing across multiple clusters to a certain extent.

Specifically, if there are at least two clusters, that is, there can be multiple pre-scheduled clusters, after obtaining the total number of the first target replicas of the first cluster, the total number of the second target replicas of other clusters, such as the second cluster, can be obtained in the same way as obtaining the total number of the first target replicas of the first cluster in steps S101 to S106. It will not be elaborated here. It can be understood that the second cluster can be used to refer to the third cluster, the fourth cluster, and the nth cluster, etc. Similarly, the total number of the third target replicas of the third cluster or the total number of the fourth target replicas of the fourth cluster can also be obtained, etc. There are no specific restrictions here.

Furthermore, after obtaining the total number of second target replicas in the second cluster of at least two clusters, the replica allocation weights corresponding to at least two clusters can be obtained by calculating the ratio between the total number of first target replicas and the total number of second target replicas, and these replica allocation weights corresponding to at least two clusters can be stored in the database. For example, assuming that the total number of first target replicas in the first cluster is 4 and the total number of second target replicas in the second cluster is 3, the replica allocation weights corresponding to at least two clusters can be obtained as 4:3 by calculating the ratio between the total number of first target replicas and the total number of second target replicas.

Optionally, based on the embodiment corresponding to Figure 7 above, in another optional embodiment of the method for determining the number of replicas provided in this application, as shown in Figure 8, after calculating the ratio between the first target total number of replicas and the second target total number of replicas to obtain the replica allocation weights corresponding to at least two clusters, the method further includes:

In step S801, a cluster scheduling request is received, wherein the cluster scheduling request carries the total number of replicas to be produced;

In step S802, the total number of replicas to be produced is divided into the first number of replicas to be produced in the first cluster and the second number of replicas to be produced in the second cluster according to the replica allocation weight.

In step S803, the first number of replicas to be produced and the second number of replicas to be produced are allocated to the first cluster and the second cluster respectively for replica production.

In this embodiment, after obtaining the replica allocation weights corresponding to at least two clusters, when a cluster scheduling request carrying the total number of replicas to be produced and the pre-scheduled cluster identifier is received, the total number of replicas to be produced can be divided into the first number of replicas to be produced in the first cluster and the second number of replicas to be produced in the second cluster according to the replica allocation weights. Then, the first number of replicas to be produced and the second number of replicas to be produced can be allocated to the first cluster and the second cluster respectively for replica production. This can more accurately and reasonably distribute the number of replicas to different clusters for replica production, thereby maintaining the load balance of multiple clusters to a certain extent.

Specifically, after obtaining the replica allocation weights corresponding to at least two clusters, when a cluster scheduling request carrying the total number of replicas to be produced and the pre-scheduled cluster identifier is received, the replica allocation weights corresponding to at least two clusters corresponding to the pre-scheduled cluster identifier can be read from the database according to the pre-scheduled cluster identifier. Then, the total number of replicas to be produced can be divided according to the replica allocation weights, specifically dividing the first number of replicas to be produced in the first cluster and the second number of replicas to be produced in the second cluster. Then, the first number of replicas to be produced and the second number of replicas to be produced can be allocated to the first cluster and the second cluster respectively for replica production, thereby achieving load balancing of multiple clusters.

The following is a detailed description of the device for determining the number of copies in this application. Please refer to Figure 10, which is a schematic diagram of an embodiment of the device for determining the number of copies in this application. The device 20 for determining the number of copies includes:

The acquisition unit 201 is used to acquire the replica node configuration information and replica resource configuration information of the first cluster, which includes several nodes;

Processing unit 202 is used to perform a filtering operation on several nodes of the first cluster according to the replica node configuration information to obtain at least one first target node;

The acquisition unit 201 is also used to acquire the amount of resources used and the total amount of resources of the smallest cluster scheduling unit in each first target node;

The processing unit 202 is also used to calculate the remaining resources of each first target node based on the amount of resources used and the total amount of resources;

Processing unit 202 is also used to calculate the number of first target replicas for each first target node based on the remaining resource quantity and the replica resource configuration information according to the quantity calculation strategy.

The processing unit 202 is also used to summarize the number of first target replicas of each first target node to obtain the total number of first target replicas of the first cluster.

Optionally, based on the embodiment corresponding to FIG10 above, in another embodiment of the device for determining the number of copies provided in this application, the processing unit 202 may specifically be used for:

Based on the selector configuration data, the first candidate node is selected from several nodes in the first cluster;

Based on affinity configuration data, select second candidate nodes from several first candidate nodes;

Based on the taint and tolerance configuration data, a third candidate node is selected from several second candidate nodes;

The node that meets the scheduling conditions among several third candidate nodes is selected as the first target node.

Optionally, based on the embodiment corresponding to FIG10 above, in another embodiment of the device for determining the number of copies provided in this application, the processing unit 202 may specifically be used for:

Obtain the node status information of several third candidate nodes;

If the node status information is in a runnable and schedulable state, then the third candidate node will be selected as the first target node.

Optionally, based on the embodiment corresponding to FIG10 above, in another embodiment of the device for determining the number of copies provided in this application,

The acquisition unit 201 is also used to acquire the number of running and the total number of the smallest cluster scheduling units in each first target node;

The processing unit 202 can be specifically used to: calculate the number of first target replicas for each first target node based on the quantity calculation strategy, the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of already run replicas and the total number of replicas.

Optionally, based on the embodiment corresponding to FIG10 above, in another embodiment of the device for determining the number of copies provided in this application, the processing unit 202 may specifically be used for:

Calculate the ratio between the remaining resource quantity and the resource quantity in the replica resource configuration information to obtain the number of at least one first replica;

Calculate the ratio between the number of already run replicas and the total number of replicas to obtain the number of second replicas;

Select the smallest value from at least one first replica quantity and a second replica quantity as the first target replica quantity.

Optionally, based on the embodiment corresponding to FIG10 above, in another embodiment of the device for determining the number of copies provided in this application,

The acquisition unit 201 is also used to acquire the total number of second target replicas of the second cluster in at least two clusters;

The processing unit 202 is also used to calculate the ratio between the total number of the first target replicas and the total number of the second target replicas, so as to obtain the replica allocation weights for at least two clusters.

Optionally, based on the embodiment corresponding to FIG10 above, in another embodiment of the device for determining the number of copies provided in this application,

The receiving unit 203 is used to receive a cluster scheduling request, wherein the cluster scheduling request carries the total number of replicas to be produced;

Processing unit 202 is also used to divide the total number of replicas to be produced into the first number of replicas to be produced in the first cluster and the second number of replicas to be produced in the second cluster according to the replica allocation weight.

The processing unit 202 is also used to allocate the first number of replicas to be produced and the second number of replicas to be produced to the first cluster and the second cluster respectively for replica production.

This application also provides another schematic diagram of a computer device, as shown in FIG11. FIG11 is a schematic diagram of a computer device structure provided by an embodiment of this application. The computer device 300 may vary considerably due to different configurations or performance, and may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 331 or data 332. The memory 320 and storage media 330 may be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the figure), each module may include a series of instruction operations on the computer device 300. Furthermore, the CPU 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the computer device 300.

The computer device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input/output interfaces 360, and/or one or more operating systems 333, such as Windows Server ^™ , Mac OS X ^™ , Unix ^™ , Linux ^™ , FreeBSD ^™ , etc.

The computer device 300 described above is also used to perform the steps in the embodiments corresponding to Figures 2 to 8.

Another aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the steps of the method described in the embodiments shown in Figures 2 to 8.

Another aspect of this application provides a computer program product containing instructions that, when run on a computer or processor, causes the computer or processor to perform the steps of the method described in the embodiments shown in Figures 2 to 8.

Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims (11)

1. A method for determining the number of copies, characterized in that it includes: Obtain the replica node configuration information and replica resource configuration information of the first cluster, which includes several nodes; Based on the replica node configuration information, a filtering operation is performed on several nodes of the first cluster to obtain at least one first target node; Obtain the used resources and total resources of the smallest cluster scheduling unit in each of the first target nodes; Calculate the remaining resources for each of the first target nodes based on the amount of resources used and the total amount of resources; Based on the quantity calculation strategy, the remaining resource quantity and the resource quantity of each replica in the replica resource configuration information are calculated to obtain the first target replica quantity of each first target node; The number of first target replicas for each first target node is summed to obtain the total number of first target replicas for the first cluster. 2. The method according to claim 1, wherein the replica node configuration information includes selector configuration data, affinity configuration data, and taint and tolerance configuration data; The step of filtering several nodes in the first cluster based on the replica node configuration information to obtain at least one first target node includes: Based on the selector configuration data, a first candidate node is selected from several nodes in the first cluster; Based on the affinity configuration data, a second candidate node is selected from a plurality of first candidate nodes; Based on the taint and tolerance configuration data, a third candidate node is selected from several second candidate nodes; The node that meets the scheduling conditions among the several third candidate nodes is selected as the first target node. 3. The method according to claim 2, wherein selecting the node that satisfies the scheduling conditions from among the plurality of third candidate nodes as the first target node includes: Obtain the node status information of several of the third candidate nodes; If the node status information indicates that it is in a runnable and schedulable state, then the third candidate node will be selected as the first target node. 4. The method according to claim 1, characterized in that, before calculating the remaining resource quantity and the resource quantity in the replica resource configuration information according to the quantity calculation strategy to obtain the first target replica quantity for each first target node, the method further includes: Obtain the number of running and total number of the smallest cluster scheduling units in each of the first target nodes; The step of calculating the number of first target replicas for each first target node based on the quantity calculation strategy includes: Based on the quantity calculation strategy, the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of already running replicas, and the total number are calculated to obtain the first target replica quantity for each first target node. 5. The method according to claim 4, characterized in that, the step of calculating the number of first target replicas for each first target node based on the remaining resource quantity, the resource quantity in the replica resource configuration information, the number of already run replicas, and the total number of replicas according to the quantity calculation strategy includes: Calculate the ratio between the remaining resource quantity and each resource quantity in the replica resource configuration information to obtain at least one first replica quantity; Calculate the ratio between the number of already run replicas and the total number of replicas to obtain the second number of replicas; The smallest value among the at least one first replica quantity and the second replica quantity is selected as the first target replica quantity. 6. The method according to claim 1, characterized in that, when at least two clusters exist; After summing the number of first target replicas for each first target node to obtain the total number of first target replicas for the first cluster, the method further includes: Obtain the total number of the second target replicas in the second cluster of the at least two clusters; Calculate the ratio between the total number of the first target replicas and the total number of the second target replicas to obtain the replica allocation weights for the at least two clusters. 7. The method according to claim 6, characterized in that, after calculating the ratio between the first target total number of replicas and the second target total number of replicas to obtain the replica allocation weights corresponding to the at least two clusters, the method further includes: Receive a cluster scheduling request, wherein the cluster scheduling request carries the total number of replicas to be produced; Based on the replica allocation weight, the total number of replicas to be produced is divided into the first number of replicas to be produced in the first cluster and the second number of replicas to be produced in the second cluster; The first number of replicas to be produced and the second number of replicas to be produced are respectively allocated to the first cluster and the second cluster for replica production. 8. A device for determining the number of copies, characterized in that it comprises: The acquisition unit is used to acquire the replica node configuration information and replica resource configuration information of the first cluster, which includes several nodes; The processing unit is used to perform a filtering operation on several nodes of the first cluster according to the replica node configuration information to obtain at least one first target node. The acquisition unit is also used to acquire the amount of resources used and the total amount of resources of the smallest cluster scheduling unit in each of the first target nodes; The processing unit is further configured to calculate the remaining resources of each first target node based on the amount of resources used and the total amount of resources. The processing unit is further configured to calculate the number of first target replicas for each first target node based on the remaining resource quantity and the replica resource configuration information according to the quantity calculation strategy. The processing unit is further configured to summarize the number of first target replicas for each first target node to obtain the total number of first target replicas for the first cluster. 9. A computer device, characterized in that it comprises: a memory, a transceiver, a processor, and a bus system; The memory is used to store programs; When the processor is used to execute a program in the memory, it implements the method as described in any one of claims 1 to 7; The bus system is used to connect the memory and the processor to enable communication between the memory and the processor. 10. A computer-readable storage medium comprising instructions, when executed on a computer, causing the computer to perform the method as claimed in any one of claims 1 to 7. 11. A computer program product comprising a computer program/instructions, characterized in that, when executed by a processor, the computer program/instructions implement the steps of the method according to any one of claims 1 to 7.