KR20230024744A - Method, apparatus, system and computer program for container distribution in multi-cluster environment - Google Patents

Abstract

The present invention relates to a method, device, system, and computer program for distributing a container in a multi-cluster environment and, more specifically, to a method, device, system, and computer program for distributing a container in a multi-cluster environment capable of effectively preventing performance degradation and abnormal termination of the container by considering a computing resource of each cluster and distributing the container in the multi-cluster environment. The method for distributing a container in a multi-cluster environment having a plurality of clusters according to the present invention comprises: a candidate cluster filtering step of allowing a container distributing server to filter a candidate cluster which is distributed among a plurality of clusters by considering a computing resource condition required in a first container to be distributed; a distribution cluster selecting step of selecting a first cluster to which the first container is distributed among the candidate cluster by considering a computing resource requiring amount of the first container and a computing resource amount of each machine included in the candidate cluster; and a machine selecting step of selecting a machine in which the first container is operated from machines included in the first cluster.

Description

Method, apparatus, system and computer program for container distribution in multi-cluster environment}

The present invention relates to a container distribution method, apparatus, system, and computer program in a multi-cluster environment, and more particularly, to a container distribution method in consideration of computing resources of each cluster in a multi-cluster environment, thereby reducing container performance and abnormal termination. It relates to a container deployment method, device, system, and computer program in a multi-cluster environment that can effectively prevent

Recently, a container technique for modularizing and running applications in a cloud environment has been widely used. The container technique is a type of virtualization achieved by packaging a user space instance at the application level, separated from the kernel, and has advantages such as lighter weight than a virtual machine, portability, and various operating environments.

However, the number of applications driven by the increase of various services and systems is continuously increasing, and accordingly, container integration, scheduling, and performance management remain difficult tasks. In particular, due to the operation of multiple devices such as IoT and the explosive increase in data throughput, there is an increasing technical demand for managing containers stably.

In this regard, as shown in FIG. 1, a technology for efficiently distributing and managing containerized applications is called container orchestration, and open platforms for this purpose (eg, Kubernetes, etc.) have been attempted, but various containers are still being tried. It is insufficient to say that it has reached a level that can efficiently and stably handle utilization services.

In particular, in the case of the above conventional platform, in a multi-cluster environment with multiple clusters, QoS (Quality of Service) of individual containers is not considered, and containers are deployed by simply focusing on work load balancing, while some Due to the limitations of computing resources such as CPU (Central Processing Unit), memory, storage, and network in containers, performance may deteriorate or, in some cases, a specific container may abnormally terminate. could

Accordingly, it is possible to stably and efficiently process various container-using services, and in particular, there is a need for a method to prevent performance degradation or abnormal termination of containers due to limitations of computing resources even in a multi-cluster environment. No suitable solution has been presented.

Republic of Korea Patent Publication No. 10-2018-0138123 (published on December 28, 2018)

An object of the present invention is to provide a container distribution method, apparatus, system, and computer program that can stably and efficiently process various container utilization services.

In addition, an object of the present invention is to provide a container distribution method, device, system, and computer program capable of selecting and distributing an optimal cluster to deploy a container by monitoring the state of computing resources of each cluster in a multi-cluster environment.

In particular, it is an object of the present invention to provide a method, apparatus, system, and computer program for distributing containers that can prevent performance degradation or abnormal termination of containers due to limitations in computational resources even in a multi-cluster environment.

Furthermore, an object of the present invention is to provide a container distribution method, apparatus, system, and computer program that can improve the problem that QoS (Quality of Service) of each container is not properly guaranteed in a multi-cluster environment.

Other detailed objects of the present invention will be clearly identified and understood by experts or researchers in the art through the specific details described below.

A method for deploying containers in a multi-cluster environment according to an aspect of the present invention for solving the above problems is a method for distributing containers in a multi-cluster environment in which a plurality of clusters are provided, wherein a container deployment server is configured to deploy a first container to be deployed. a candidate cluster filtering step of filtering distributable candidate clusters among a plurality of clusters in consideration of computing resource conditions required by the container; a distribution cluster selecting step of selecting a first cluster to which the first container is to be distributed from among the candidate clusters in consideration of a computational resource requirement of the first container and a computational resource quantity of each machine included in the candidate cluster; and a machine selection step of selecting a machine on which the first container will run from among the machines included in the first cluster.

At this time, in the candidate cluster filtering step, a CPU (Central Processing Unit), memory, storage space, and network required by the first container among the machines included in the plurality of clusters are included. A cluster including a machine that can satisfy all of the computing resource conditions for one or more of the computing resources to be selected may be selected as a candidate cluster.

The filtering of candidate clusters may further include selecting, as a candidate cluster, a cluster including a machine capable of allocating one or more ports required by the first container among machines included in the plurality of clusters. can

The filtering of candidate clusters may further include selecting, as a candidate cluster, a cluster including a machine capable of allocating one or more storage devices required by the first container among the machines included in the plurality of clusters. can

Furthermore, in the step of selecting the distribution cluster, the priority of one or more of the priority of required resources versus available resources for the candidate cluster, priority of core resources, priority of used/requested resources versus total resources, and priority of network resources is considered. You can select a deployment cluster.

At this time, in the step of selecting the distribution cluster, for each machine included in the candidate cluster, the available resource versus demand is determined from the sum of the ratios of the computational resource requirements in the first container to the available computational resources in each machine. A resource priority may be calculated, and the distribution cluster may be selected in consideration of the size of the priority of the required resource compared to the available resource.

In addition, in the distribution cluster selection step, for each machine included in the candidate cluster, a computational resource having the largest first ratio of computational resource usage to the total computational resource quantity in the machine is selected as a core resource, and the core resource The distribution cluster may be selected in consideration of a first ratio of resources (=core resource priority).

In addition, in the step of selecting the deployment cluster, for each machine included in the candidate cluster, the sum of the computational resource usage of already deployed containers and the computational resource requirement of the first container with respect to the total computational resource quantity in the machine The distribution cluster may be selected by calculating the priority of resources used/requested against the total resources from the ratio, and considering the priority of the resources used/requested against the total resources.

In addition, in the step of selecting the distribution cluster, for each machine included in the candidate cluster, a network resource priority is calculated in consideration of real-time network usage in the machine, and the first container is calculated in consideration of the network resource priority. You can select a distribution cluster to deploy.

In addition, in the distribution cluster selection step, the distribution cluster may be selected by calculating the sum of the priorities of each machine included in the candidate cluster and considering the size of the sum of the priorities of each of the candidate clusters. .

In addition, in the distribution cluster selection step, the distribution cluster may be selected by considering the highest priority among the priorities of each machine included in the candidate cluster.

Furthermore, the first cluster is provided with a single cluster monitoring unit to collect computing resource data for each machine included in the first cluster to generate computing resource data for the first cluster, and to the container distribution server. A multi-cluster monitoring unit is provided that collects computational resource data for a plurality of clusters including a first cluster, and collects computational resource data for each cluster from each single cluster monitoring unit provided in the plurality of clusters to obtain the plurality of clusters. Computational resource data can be created for the cluster.

In addition, the container distribution server determines one or more of the number of times the first container is distributed but not normally run or the number of times the first container is abnormally terminated after running, and calculates a quality of service (QoS) in consideration of this A service quality index management step to manage; may be further included.

In addition, a computer program according to another aspect of the present invention is characterized in that it is a computer program stored in a computer readable medium for executing each step of the above-described container distribution method in a multi-cluster environment on a computer.

In addition, a container distribution system according to another aspect of the present invention is a container distribution system for distributing containers in a multi-cluster environment having a plurality of clusters, in consideration of computing resource conditions required for a first container to be deployed. filtering candidate clusters that can be deployed among clusters of , and considering the amount of computational resources required by the first container and the amount of computational resources of each machine included in the candidate cluster, a first cluster to which the first container is to be distributed is selected from among the candidate clusters. Container distribution server that selects and distributes; and one or more cluster servers having a plurality of clusters and constituting multiple clusters.

Accordingly, in the container distribution method, apparatus, system, and computer program according to an embodiment of the present invention, various container utilization services can be stably and efficiently processed.

In addition, in the container distribution method, apparatus, system, and computer program according to an embodiment of the present invention, it is possible to select and distribute an optimal cluster to distribute a container by monitoring the state of computing resources of each cluster in a multi-cluster environment. .

In particular, in the container distribution method, apparatus, system, and computer program according to an embodiment of the present invention, it is possible to prevent a problem in which performance of a container is degraded or abnormally terminated due to limitations of computing resources even in a multi-cluster environment.

In addition, in the container distribution method, apparatus, system, and computer program according to an embodiment of the present invention, QoS (Quality of Service) of each container is guaranteed in a multi-cluster environment, and furthermore, the workload of a system manager related thereto is effectively reduced. can give

In addition, the container deployment method, device, system, and computer program according to an embodiment of the present invention can be applied to various systems such as cloud systems, SaaS, DaaS, and data centers, and through this, the stability and efficiency of the system can be improved. there will be

The accompanying drawings, which are included as part of the detailed description to aid understanding of the present invention, provide examples of the present invention and explain the technical idea of the present invention together with the detailed description.
1 is a diagram illustrating a structure in which containers are deployed to nodes of a cluster in Kubernetes according to the prior art.
2 is a configuration diagram of a container distribution system according to an embodiment of the present invention.
3 is a flowchart of a container deployment method according to an embodiment of the present invention.
4 is a diagram illustrating a specific configuration and operation of a container distribution system according to an embodiment of the present invention.
5 is a flowchart of a container deployment method according to another embodiment of the present invention.
6 is a flowchart illustrating specific operations of a container deployment method according to another embodiment of the present invention.
7 is a diagram illustrating the configuration and operation of a container distribution system according to another embodiment of the present invention.
8 is a configuration diagram of a container distribution server according to an embodiment of the present invention.

The present invention can apply various transformations and can have various embodiments. Hereinafter, specific embodiments will be described in detail based on the accompanying drawings.

The following examples are provided to facilitate a comprehensive understanding of the methods, apparatus and/or systems described herein. However, this is only an example and the present invention is not limited thereto.

In describing the embodiments of the present invention, if it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the subject matter of the present invention, the detailed description will be omitted. In addition, terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to the intention or custom of a user or operator. Therefore, the definition should be made based on the contents throughout this specification. Terms used in the detailed description are only for describing the embodiments of the present invention, and should not be limiting. Unless expressly used otherwise, singular forms of expression include plural forms. In this description, expressions such as "comprising" or "comprising" are intended to indicate any characteristic, number, step, operation, element, portion or combination thereof, one or more other than those described. It should not be construed to exclude the existence or possibility of any other feature, number, step, operation, element, part or combination thereof.

In addition, terms such as first and second may be used to describe various components, but the components are not limited by the terms, and the terms are used for the purpose of distinguishing one component from another. used only as

Hereinafter, exemplary embodiments of a container distribution method, apparatus, system, and computer program in a multi-cluster environment according to an embodiment of the present invention will be sequentially described with reference to the accompanying drawings.

First, in FIG. 2, a configuration diagram of the container distribution system 100 according to an embodiment of the present invention is illustrated. As can be seen in FIG. 2, the container distribution system 100 according to an embodiment of the present invention may include a container distribution server 110 and one or more cluster servers 120, and the container distribution server 110 and the cluster server 120 may operate in conjunction with each other while transmitting and receiving data through the communication network 140 .

At this time, the container distribution system 100 according to an embodiment of the present invention is a container distribution system 100 that distributes containers in a multi-cluster environment in which a plurality of clusters are provided, and computational resources required for the first container to be distributed. Distributable candidate clusters among a plurality of clusters are filtered in consideration of conditions, and the first container is distributed among the candidate clusters in consideration of the amount of computing resources required by the first container and the amount of computing resources of each machine included in the candidate cluster. It may include a container distribution server 110 for selecting and distributing a first cluster to be selected and one or more cluster servers 120 having a plurality of clusters and constituting multiple clusters.

In this case, the computing resource condition means a type of one or more computing resources required for driving the first container to be distributed and a standard required for the computing resource.

In the present invention, a CPU (Central Processing Unit), a memory, a storage space, a network, and the like may be examples of computing resources.

Here, the container distribution server 110 receives a request from the terminal 130 such as a user's PC, laptop, smartphone, tablet, PDA, mobile phone, etc., packages an application into a container, and distributes the application to the cluster 120. there is.

For a more specific example, the first container packaged in the application to be distributed on the user's terminal 130 may require CPU, memory, and storage device resources. When 1 GB of memory resources and 2 GB of storage resources are required, distribution of the first container may be requested by specifying the type of computing resource and its required standard in the form shown in the example of [Table 1] below, and furthermore, the distribution of the first container may be requested. It is also possible to specify limits for each computing resource together.

resource:
requests:
cpu: "0.5"
memory: "1Gi"
store: "2Gi"
limits:
cpu: "1"
memory: "2Gi"
store: "4Gi"

However, the present invention is not necessarily limited to this, and the container distribution server 110 receives requests from other servers (not shown), receives requests from various types of devices such as IoT (Internet of Things) devices, and other various Containers can also be deployed by receiving requests from different types of devices.

At this time, the container distribution server 110 may be composed of one or more servers, computer devices such as PCs, or may be implemented in various forms such as a virtual server running in a cloud system or the like.

In addition, the container distribution server 110 may operate while transmitting and receiving data to and from the cluster server 120 through the communication network 140 .

More specifically, the communication network 140 may include a wired network and a wireless network, and more specific examples include a local area network (LAN), a metropolitan area network (MAN), and a wide area network ( It may include various communication networks such as a wide area network (WAN). In addition, the network 140 may include the well-known World Wide Web (WWW) or the like. However, the communication network 140 according to the present invention is not limited to the above-listed networks, and may include various types of wired/wireless data networks, such as a known telephone network or a known wired/wireless television network.

In addition, although the cluster server 120 is shown in FIG. 2 as being configured to include a plurality of cluster servers 120-1, 120-2, ??, 120-m, the present invention is not necessarily limited thereto. No, in the present invention, the cluster server 120 can be configured with a single server device, or configured with a plurality of server devices as shown in FIG. 2 . Furthermore, in the present invention, the cluster server 120 can be implemented in various forms such as a virtual server running in the cloud system by configuring a cloud system with a plurality of server devices.

Accordingly, in the container deployment system 100 according to an embodiment of the present invention, various container utilization services can be stably and efficiently processed, and even in a multi-cluster environment, performance of containers may deteriorate due to limitations in computing resources or It is possible to prevent the problem of abnormal termination.

Hereinafter, a container distribution method, apparatus, system, and computer program in a multi-cluster environment according to an embodiment of the present invention will be described in more detail with reference to the drawings.

3 illustrates a flow chart of a container distribution method in a multi-cluster environment according to an embodiment of the present invention.

As can be seen in FIG. 3, the container deployment method in a multi-cluster environment according to an embodiment of the present invention deploys containers in a multi-cluster environment in which a plurality of clusters 120a, 120b, ??, 120m are provided. In the method, a candidate cluster filtering step (S110) of filtering deployable candidate clusters among a plurality of clusters in consideration of computational resource conditions required for a first container to be deployed by the container distribution server 110, A distribution cluster selection step (S120) of selecting a first cluster to which the first container is to be distributed among the candidate clusters in consideration of the computational resource requirements and the computational resource amount of each machine included in the candidate cluster; A machine selection step ( S130 ) of selecting a machine in which the first container is to be driven from among the machines may be included.

In addition, FIG. 4 shows a diagram for explaining the specific configuration and operation of the container distribution system 100 according to an embodiment of the present invention.

As can be seen in FIG. 4 , the container distribution system 100 according to an embodiment of the present invention may include a container distribution server 110 and a plurality of clusters 120a, ??, and 120m.

At this time, the scheduler 111 is provided in the container distribution server 110 to select and distribute a first cluster among a plurality of clusters 120a, ??, and 120m where the first container is to be distributed.

Accordingly, the first container may be driven on a first machine selected from among one or more machines included in the first cluster.

At this time, the first machine on which the first container runs may be selected from the container distribution server 110, but the present invention is not necessarily limited thereto, and the first machine is selected by itself in the first cluster or by an external server. It is also possible to select the first machine in a separate server or the like.

Hereinafter, with reference to FIGS. 3 and 4 , a container distribution method, apparatus, and system 100 in a multi-cluster environment according to an embodiment of the present invention will be described in more detail by dividing each component.

First, in the candidate cluster filtering step (S110), the container distribution server 110 filters and selects distributable candidate clusters among a plurality of clusters in consideration of the computing resource condition required for the first container to be distributed. will do

More specifically, in the candidate cluster filtering step (S110), CPU (Central Processing Unit), memory, storage space, and network required by the first container among the machines included in the plurality of clusters A cluster including a machine that can satisfy all computing resource conditions for one or more of computing resources including a network may be selected as a candidate cluster.

For example, when CPU resources, memory resources, and storage resources are required as computational resources, in the candidate cluster filtering step (S110), filtering may be performed according to Equation 1 below. .

[Equation 1]

For a more specific example, when the CPU resource required by the first container requested by the user terminal 130 is 5, the memory resource resource is 3, and the storage space resource is 7, the container distribution server ( 110), among the machines 121a, ??, 121n, 12ma, ??, 12mn included in the plurality of clusters 120a, ??, 120m, all of the conditions required for each of the computing resources are satisfied. Clusters including machines are selected as candidate clusters.

At this time, as can be seen in FIG. 4 , the machine may be a physical machine (PM), but may also be a virtual machine (VM).

In addition, as can be seen in FIG. 4, each cluster 120a, ??, 120m may include a real machine (PM) and a virtual machine (VM), but also each of the clusters 120a, ??, 120 m) may include a mixture of the real machine (PM) and the virtual machine (VM).

Accordingly, if PM #1 (121a) of cluster #1 (120a) in FIG. 4 can satisfy all computing resource conditions required by the first container, cluster #1 (120a) can be included in the candidate cluster. there will be

In addition, in the present invention, the candidate cluster filtering step (S110) is capable of allocating one or more ports required by the first container among machines included in the plurality of clusters 120a, ??, and 120m. A step (not shown) of selecting a cluster including the machine as a candidate cluster may be further included (eg, when both a port for TCP communication and a port for UDP communication are required).

In the present invention, the filtering of candidate clusters (S110) includes a machine capable of allocating one or more storage devices required by the first container among the machines included in the plurality of clusters 120a, ??, and 120m. A step (not shown) of selecting a cluster to be selected as a candidate cluster may be further included (eg, when accessing a specific database to save or read data).

Accordingly, in the step of filtering candidate clusters (S110) in the present invention, a machine capable of satisfying all computing resource conditions required for computing resources such as CPU, memory, storage space, and network is included. Candidate clusters may be calculated by filtering out clusters that require a desired port or by filtering out clusters including machines capable of allocating a required port or storage device.

Subsequently, in the distribution cluster selection step (S120), the container distribution server 110 considers the computational resource requirements of the first container and the computational resources of each machine included in the candidate cluster, and selects the first one of the candidate clusters. The first cluster where the container will be deployed is selected.

More specifically, in the step of selecting the distribution cluster (S120) in the present invention, the priority of the available resources versus the required resources for the candidate cluster, the priority of the core resources, the priority of the used/requested resources versus the total resources, and the priority of the network resources The distribution cluster may be selected by considering one or more priorities.

In this case, the priority of the required resource to the available resource may be calculated from the sum of the ratios of the computational resource requirement in the first container to the available computational resource capacity in each machine included in the candidate cluster.

For a more specific example, when there are N computational resources in the machine, the priority of the required resource versus the available resource may be calculated using Equation 2 below.

[Equation 2]

That is, when the computational resource demand for the computational resource availability is large, it means that the computational resources of the machine and the node composed of the corresponding machine can be maximally consumed, and accordingly, when containers are distributed to the machine or node In the step of filtering candidate clusters for next container deployment (S110), since the machines or nodes can be filtered out, more efficient management is possible by removing them from the management list for additional container deployment.

In addition, with respect to the priority of the core resource, first, a computing resource having the largest first ratio of the amount of computing resource usage to the total amount of computing resources in the machine is selected as a core resource, and then a first ratio (= core resource) of the core resource is selected. The distribution cluster may be selected in consideration of resource priority).

For a more specific example, when there are N computational resources in the machine, the core resource can be calculated using [Equation 3] below for each of the N computational resources.

[Equation 3]

As a more specific example, if cluster A's CPU resource usage/total CPU resource amount is 8/10, memory resource usage/total memory resource amount is 7/10, and storage device usage/total storage device resource amount is 6/10, The core resource of cluster A is CPU and the core resource priority can be 0.8.

On the other hand, if cluster B's CPU resource usage/total CPU resource amount is 5/10, memory resource usage/total memory resource amount is 4/10, and storage device usage/total storage device resource amount is 6/10, cluster B's core The resource is a storage device and the core resource priority can be 0.6.

At this time, since the core resource priority (CPU, 0.8) of cluster A is higher than the core resource priority (storage device, 0.6) of cluster B, when a container is first placed in cluster A, a candidate for deploying the next container In the cluster filtering step (S110), the cluster A can be filtered and filtered so that management can be performed more efficiently.

In addition, the priority of the resource used/required relative to the total resource is the sum of the computational resource usage of already deployed containers and the computational resource requirement of the first container with respect to the total computational resource amount in each machine included in the candidate cluster. can be calculated from the ratio.

For a more specific example, when there are N computational resources in the machine, the priority of the available resources versus the requested resources may be calculated using Equation 4 below.

[Equation 4]

Accordingly, as an embodiment of the present invention, a cluster having a small ratio of the sum of the computational resource usage of already deployed containers and the computational resource requirement of the first container to the total computational resource quantity in each machine is selected. It is also possible to select a machine or node to deploy the container to.

In the present invention, in the distribution cluster selection step (S120), network resource priorities are calculated in consideration of real-time network usage in each machine included in the candidate cluster, and the network resource priorities are considered. 1 You can also select a deployment cluster to deploy containers.

Furthermore, in the container distribution method, apparatus, and system in a multi-cluster environment according to an embodiment of the present invention, the sum of the priorities for each machine included in the candidate cluster is calculated in the distribution cluster selection step (S120). It is also possible to select a distribution cluster in consideration of the size of the sum of the priorities for each of the candidate clusters.

In this case, the distribution cluster may be selected by summing up the priority scores of each machine or node in the same cluster and selecting a cluster having the highest average acceptability.

In the present invention, in the distribution cluster selection step (S120), the distribution cluster may be selected by considering the highest priority among the priorities of each machine included in the candidate cluster.

In this case, a machine or node with the highest priority within the same cluster represents the cluster, and it is possible to guarantee that the machine or node with the corresponding priority accommodates the container.

Accordingly, in the deployment cluster selection step (S120), the first cluster to which the first container is to be distributed among the candidate clusters in consideration of the computational resource requirement of the first container and the computational resource quantity of each machine included in the candidate cluster. will select

Subsequently, in the machine selection step (S130), a machine on which the first container is to be driven is selected from among the machines included in the first cluster.

At this time, as described above, it is also possible to select the machine in consideration of the calculated core resource and core resource priority for the machine included in the first cluster, but the present invention is not necessarily limited thereto. With respect to the distribution cluster selected in the distribution cluster selection step (S120), it is possible to select a machine on which the first container is to be run using various algorithms.

Furthermore, as can be seen in FIG. 5, in the container distribution method in a multi-cluster environment according to an embodiment of the present invention, the container distribution server 110 distributes a first container but fails to operate normally; A QoS management step (S140) of recognizing one or more of the number of times the first container is abnormally terminated after being driven, calculating and managing a QoS in consideration of this may be further included.

Accordingly, in the present invention, the present invention does not stop at simply selecting and deploying a machine on which to deploy containers, but detects abnormal operation or abnormal termination occurring in the deployed container, and then adjusts the distribution of the container using this, so that the distribution of the deployed container You will be able to manage and improve your Quality of Service (QoS).

In this regard, FIG. 6 illustrates a flowchart illustrating specific operations according to a container distribution method in a multi-cluster environment according to an embodiment of the present invention.

As can be seen in FIG. 6 , when a user requests distribution of a first container to the container distribution server 110 through the terminal 130 (S210), the container distribution server 110 sends a request to the container distribution server 110 for all of the plurality of clusters. Filtering is performed to calculate candidate clusters (S220).

Subsequently, the container distribution server 110 sets a priority score of one or more of the priority of available resources versus requested resources, priority of core resources, priority of used/required resources versus total resources, and network resource priority for the candidate cluster. is calculated (Scoring) (S230).

Accordingly, the container distribution server 110 selects a cluster having the highest priority score among the candidate clusters as a distribution cluster and distributes the first container (S240).

Subsequently, after the distribution of the first container is completed, the computational resource quantity of machines or nodes in the distribution cluster where the first container is distributed is updated, and the computational resource quantity data at the machine or node stage is updated by the single cluster monitoring unit 122a. is extracted and transmitted (S250).

Accordingly, the multi-cluster monitoring unit 112 of the container distribution server 110 collects data on the amount of computational resources in each cluster, extracts them as real-time computational resources, and updates and uses them (S260).

Accordingly, in the method, apparatus, and system for deploying containers in a multi-cluster environment according to an embodiment of the present invention, the amount of computational resources of each cluster is updated in real time and reflected in the distribution of containers, thereby processing container distribution more efficiently. You will be able to do it.

In addition, in FIG. 7, one or more single cluster monitoring units 122a, ??, 122m and multiple cluster monitoring units 112 are added to the container deployment system 100 shown in FIG. 4 to configure the container deployment system 100. A case is exemplified.

More specifically, as shown in FIG. 7 , in the container deployment method, apparatus, and system in a multi-cluster environment according to an embodiment of the present invention, a single cluster is monitored in the first cluster 120a where the container is deployed. A unit 122a is provided to collect computational resource data for each machine or node included in the first cluster 120a to generate computational resource data for the first cluster, and then to the container distribution server. 110 is provided with a multi-cluster monitoring unit 112 that collects computing resource data for a plurality of clusters including the first cluster, and each single cluster provided in the plurality of clusters 120a, ??, 120m Computing resource data for each cluster 120a, ??, 120m is collected from the cluster monitoring unit 122a, ??, 122m, and computational resource data for the plurality of clusters 120a, ??, 120m is generated. can

Furthermore, the multi-cluster monitoring unit 112 uses the monitoring data for each cluster 120a, ??, 120m collected from the single cluster monitoring unit 122a, ??, 122m to monitor each cluster 120a. , ??, 120m), container distribution and quality of service (QoS) can be managed by identifying abnormal operation or abnormal termination of containers.

8 illustrates a configuration diagram of a container distribution server 110 in a container distribution method, apparatus, and system in a multi-cluster environment according to an embodiment of the present invention.

As can be seen in FIG. 8 , the container distribution server 110 according to the present invention may include a candidate cluster filtering unit 115 , a distribution cluster selection unit 116 and a machine selection unit 117 .

The container distribution server 110 according to the present invention has been described in detail in relation to the container distribution method and system 100 according to an embodiment of the present invention with reference to FIGS. 1 to 7, and the detailed description is repeated here. We omit what to do and only briefly look at the core configuration.

First, in the candidate cluster filtering unit 115, distributable candidate clusters among a plurality of clusters are filtered in consideration of computational resource conditions required for a first container to be distributed.

In addition, the distribution cluster selection unit 116 selects a first cluster to which the first container is to be distributed from among the candidate clusters in consideration of the computational resource requirement of the first container and the computational resource quantity of each machine included in the candidate cluster. will be selected

Subsequently, the machine selector 117 selects a machine on which the first container is to be driven from among the machines included in the first cluster.

Each component of the container distribution server 110 may be configured as a separate server or individual hardware module, but the present invention is not necessarily limited thereto, and each component of the container distribution server 110 is a software module. It is also possible to implement in various forms, such as configuring and running on one or more servers.

In addition, the computer program according to another aspect of the present invention is characterized in that it is a computer program stored in a computer readable medium in order to execute each step of the salpin container distribution method on a computer. The computer program may be a computer program including machine code generated by a compiler, as well as a computer program including high-level language code that can be executed on a computer using an interpreter or the like. At this time, the computer is not limited to a personal computer (PC) or a notebook computer, etc., and has a central processing unit (CPU) such as a server, smart phone, tablet PC, PDA, mobile phone, etc. to execute a computer program. All information processing include the device

In addition, the computer-readable medium may continuously store a computer-executable program or temporarily store it for execution or download. In addition, the medium may be various recording means or storage means in the form of a single or combined hardware, but is not limited to a medium directly connected to a certain computer system, and may be distributed on a network. Examples of the medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROM and DVD, magneto-optical media such as floptical disks, and ROM, RAM, flash memory, etc. configured to store program instructions. In addition, examples of other media include recording media or storage media managed by an app store that distributes applications, a site that supplies or distributes various other software, and a server.

Accordingly, in the container distribution method, apparatus, system, and computer program according to an embodiment of the present invention, various container utilization services can be stably and efficiently processed, and furthermore, even in a multi-cluster environment, containers can be It is possible to prevent performance deterioration or abnormal termination.

The above description is merely an example of the technical idea of the present invention, and various modifications and variations can be made to those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments described in the present invention are not intended to limit the technical spirit of the present invention, but to explain, and are not limited to these embodiments. The protection scope of the present invention should be construed by the following claims, and all technical ideas within the equivalent range should be construed as being included in the scope of the present invention.

100: container deployment system
110: container deployment server
111: scheduler
112: multi-cluster monitoring unit
115: candidate cluster filtering unit
116: distribution cluster selection unit
117: machine selection unit
120, 120-1, 120-2,120-m: cluster server
120a, 120b, 120m: cluster
121a, 121n, 12ma, 12mn: machine
122a, 122m: single cluster monitoring unit
130: Terminal
140: communication network

Claims (15)

A method for distributing a container in a multi-cluster environment in which a plurality of clusters are provided,
a candidate cluster filtering step of filtering, by a container distribution server, distributable candidate clusters among a plurality of clusters in consideration of computational resource conditions required for a first container to be distributed;
a distribution cluster selecting step of selecting a first cluster to which the first container is to be distributed from among the candidate clusters in consideration of a computational resource requirement of the first container and a computational resource quantity of each machine included in the candidate cluster; and
a machine selection step of selecting a machine on which the first container will run from among machines included in the first cluster;
Container deployment method in a multi-cluster environment comprising a. According to claim 1,
In the candidate cluster filtering step,
Computation for one or more of computing resources including a central processing unit (CPU), memory, storage, and network required by the first container among the machines included in the plurality of clusters. A container deployment method in a multi-cluster environment, characterized in that a cluster including a machine that can satisfy all resource conditions is selected as a candidate cluster. According to claim 2,
The candidate cluster filtering step,
Selecting, as a candidate cluster, a cluster including a machine capable of allocating one or more ports required by the first container among the machines included in the plurality of clusters; Container deployment method. According to claim 2,
The candidate cluster filtering step,
selecting a cluster including a machine capable of allocating one or more storage devices required by the first container among the machines included in the plurality of clusters as a candidate cluster; Container deployment method. According to claim 1,
In the step of selecting the distribution cluster,
Selecting the distribution cluster by considering the priority of one or more of the priority of required resources versus available resources, priority of core resources, priority of used/required resources versus total resources, and network resource priority for the candidate cluster. How to deploy containers in a multi-cluster environment. According to claim 5,
In the step of selecting the distribution cluster,
For each machine included in the candidate cluster,
calculating the priority of the available resources versus the required resources from the sum of the ratios of the computational resource requirements in the first container to the available computational resources in the machine;
Container distribution method in a multi-cluster environment, characterized in that the distribution cluster is selected in consideration of the size of the priority of the available resources versus the required resources. According to claim 5,
In the step of selecting the distribution cluster,
For each machine included in the candidate cluster,
Selecting a computing resource having the largest first ratio of computing resource usage to the total computing resource amount in the machine as a core resource;
Container distribution method in a multi-cluster environment, characterized in that the distribution cluster is selected in consideration of the first ratio (=key resource priority) of the core resource. According to claim 5,
In the deployment cluster selection step,
For each machine included in the candidate cluster,
Calculate priority of used/required resources relative to total resources from the ratio of the sum of the computational resource usage of already deployed containers and the computational resource requirements of the first container to the total computational resources in the machine;
A container deployment method in a multi-cluster environment, characterized in that the deployment cluster is selected in consideration of the priority of used/required resources compared to the total resources. According to claim 5,
In the step of selecting the distribution cluster,
For each machine included in the candidate cluster,
Calculate network resource priority in consideration of real-time network usage in the machine;
Container distribution method in a multi-cluster environment, characterized in that for selecting a distribution cluster to distribute the first container in consideration of the network resource priority. According to claim 5,
In the step of selecting the distribution cluster,
calculating a sum of the priorities for each machine included in the candidate cluster;
A container distribution method in a multi-cluster environment, characterized in that a distribution cluster is selected in consideration of the size of the sum of the priorities for each of the candidate clusters. According to claim 5,
In the step of selecting the distribution cluster,
Container distribution method in a multi-cluster environment, characterized in that the distribution cluster is selected in consideration of the highest priority among the priorities of each machine included in the candidate cluster. According to claim 1,
The first cluster is provided with a single cluster monitoring unit,
Computing resource data for each machine included in the first cluster is collected to generate computational resource data for the first cluster;
The container distribution server is provided with a multi-cluster monitoring unit for collecting computing resource data for a plurality of clusters including the first cluster,
A method for distributing containers in a multi-cluster environment, characterized in that generating computational resource data for the plurality of clusters by collecting computational resource data for each cluster from each single cluster monitoring unit provided in the plurality of clusters. According to claim 1,
The container distribution server determines one or more of the number of times the first container is distributed but not normally run or the number of times the first container is abnormally terminated after running;
A container deployment method in a multi-cluster environment, characterized in that it further comprises; a service quality index management step of calculating and managing a quality of service (QoS) in consideration of this. A computer program stored in a computer readable medium for executing each step of the container distribution method in a multi-cluster environment according to claims 1 to 13 on a computer. In a container distribution server for distributing containers in a multi-cluster environment in which a plurality of clusters are provided,
a candidate cluster filtering unit filtering distributable candidate clusters among a plurality of clusters in consideration of computational resource conditions required for a first container to be distributed;
a container distribution unit selecting and distributing a first cluster to which the first container is to be distributed among the candidate clusters in consideration of the computational resource requirements of the first container and the computational resource quantity of each machine included in the candidate cluster; and
a machine selection unit that selects a machine to be driven by the first cluster among machines included in the first cluster;
A container deployment server comprising a.

Images