KR20220000592A - Method and device of dynamic resource allocation for container-based applications for iot edge computing infrastructure - Google Patents

Abstract

Disclosed are a method and device for allocating dynamic resource allocation based on a container for IoT edge computing environment, which provides a resource provisioning function in consideration of kubernetes resource allocation problem. According to an embodiment of the present invention, the method comprises: a step of counting the number of allocated application replicas for edge nodes included in a cluster, in association with the cluster constituting an intermediate layer of kubernetes-based edge computing; and a step of re-adjusting the number of application replicas required by each of the edge nodes in consideration of request traffic status from an IoT device.

Description

Container-based dynamic resource distribution method and apparatus for IoT edge computing environment

The present invention provides a container-based dynamic resource distribution method and apparatus for an IoT edge computing environment that distributes copies of application services in proportion to the amount of requests to each edge node in consideration of the client's request traffic status. it's about

In particular, in the present invention, it is possible to provide a technology for improving the overall system throughput by periodically monitoring the client's real-time status and dynamically adjusting the number of copies of the application service according to the requested resource for each edge node. .

In recent years, a large amount of data is being generated from numerous Internet of Things (IoT) devices, and the number of IoT devices is expected to increase from 20 billion units at present to about 30 billion units in 2023.

In addition, in a cloud computing environment, large-scale data generated by various IoT devices is transmitted and processed through the Internet to a centralized server.

This traffic pattern not only requires a high bandwidth of the network, but also accompanies a delay time according to the traffic processing process, which can reduce the quality of IoT application services such as augmented reality sensitive to real-time, intelligent transportation system, and smart city. .

Edge computing technology, also referred to as fog computing technology, is a new computing paradigm that can improve the problems caused by centralized transmission of cloud computing. By placing computational resources close to IoT devices, it accelerates analysis and processes application services It can provide several benefits, including lower latency.

In other words, edge computing technology can lower latency and reduce the amount of traffic across the network through edge computing resources instead of sending data to the cloud. In addition, edge computing technology can reduce the burden of cloud operator resources by utilizing the computing resources of edge computing.

In the edge computing environment, containers are used in consideration of dynamic distribution of application services and ease of resource management. The container is composed of a bundle package composed of application-related libraries instead of the entire operating system like a virtual machine (VM, Virtual Machine), and has the advantage of enabling efficient, light and fast deployment.

However, edge computing technology requires tools to efficiently deploy and manage resources between containers.

Kubernetes (K8S) is an open source-based container orchestration platform for this purpose, and can be mainly used to build an infrastructure for edge computing technology.

Kubernetes can provide a variety of features for container orchestration, including deployment of application services, resource management, resource monitoring, load balancing, and autoscaling.

1 is a diagram illustrating the structure of a general Kubernetes-based edge computing.

As shown in FIG. 1 , in the lower layer 130 of Kubernetes, several IoT devices that send requests to the upper layer 110 mainly through a wireless gateway directly connected to the edge node of the middle layer 120 are provided. Included.

The middle layer 120 of Kubernetes is an edge computing (fork computing) layer, and includes edge nodes that can analyze and calculate requests from IoT devices.

Also, the upper layer 110 of Kubernetes may be a cloud computing layer including several servers in a data center.

In Kubernetes, resource processing efficiency is improved by configuring a plurality of arbitrary edge nodes into a cluster and delivering and sharing client requests in the configured cluster unit.

However, the collaboration function of edge nodes in the cluster configured in Kubernetes also entails latency and message processing costs in the process of transferring requests and responses between edge nodes.

Accordingly, by distributing appropriate resources as needed in each edge node, minimizing the cost of collaboration between nodes may be a method of optimizing the system performance of a cluster including edge nodes.

However, the conventional Kubernetes platform has limitations in two aspects.

First, Kubernetes supports distributed deployment of application service replicas to edge nodes, but lacks the ability to consider the resource request status of each edge node when supporting distributed deployment.

For example, as shown in FIG. 1, despite the situation in which four clients are connected to the edge node Location A, one client is connected to the edge node Location B, and one client is connected to the edge node Location C, the application service is being requested. In the conventional Kubernetes, the same number of replicas are distributed to each edge node, 3 copies each, without considering the resource request status to each edge node.

If a larger number of requests than distributed replicas are input, the possibility of being processed by other edge nodes at a remote location instead of the edge node receiving the user's request message may increase. In other words, whenever Kubernetes receives a request from each application service, it selects edge nodes to process the request according to the given policy and delivers the request. If it is forwarded, there is a problem in that the request processing delay time inevitably increases.

Second, in Kubernetes, dynamic changes in the request state of IoT devices cannot be applied.

That is, due to the characteristics of IoT application service infrastructure, the amount of requests to access each edge node may vary depending on the concentration of users by time period, etc. That is, the problem of imbalance between the requested resource and the actual allocated resource as shown in FIG. 1 may continuously occur over time.

Therefore, in consideration of Kubernetes resource allocation, there is an urgent need to develop a technology that provides a resource provisioning function.

An embodiment of the present invention aims to provide a container-based dynamic resource distribution method and apparatus for an IoT edge computing environment, which provides a resource provisioning function in consideration of the Kubernetes resource allocation problem.

In addition, an embodiment of the present invention aims to distribute copies for each application service in proportion to the amount of requests to each edge node in consideration of the client's request traffic status.

In addition, an embodiment of the present invention aims to improve the overall throughput of the system by periodically monitoring the real-time status of the client and dynamically adjusting the number of service copies according to the requested resource for each edge node.

A container-based dynamic resource distribution method for an IoT edge computing environment according to an embodiment of the present invention relates to a cluster constituting the middle layer of a Kubernetes-based edge computing, with respect to an edge node included in the cluster. , counting the number of assigned application replicas; and re-adjusting the number of application copies required by each of the edge nodes in consideration of the request traffic status from the IoT device.

In addition, the container-based dynamic resource distribution apparatus for the IoT edge computing environment according to an embodiment of the present invention relates to a cluster constituting the middle layer of the Kubernetes-based edge computing, and provides an edge node included in the cluster. For, a counting unit for counting the number of copies of the application to be allocated; and a processing unit that readjusts the number of application copies required by each of the edge nodes in consideration of the request traffic status from the IoT device.

According to an embodiment of the present invention, in consideration of the Kubernetes resource allocation problem, it is possible to provide a container-based dynamic resource distribution method and apparatus for an IoT edge computing environment, which provides a resource provisioning function.

In addition, according to the present invention, it is possible to distribute copies for each application service in proportion to the amount of requests to each edge node in consideration of the client's request traffic status.

In addition, according to the present invention, by periodically monitoring the client's real-time status and dynamically adjusting the number of service copies according to the requested resource for each edge node, the overall system throughput can be improved.

1 is a diagram illustrating the structure of a general Kubernetes-based edge computing.
2 is a block diagram illustrating a configuration of a container-based dynamic resource distribution device for an IoT edge computing environment according to an embodiment of the present invention.
3 is a diagram illustrating setting of nodeAffinity for scheduling according to an embodiment of the present invention.
4 is a view for explaining the components and operation of the technology proposed in the present invention.
5 is a diagram for explaining an operation algorithm in ElasticFog according to an embodiment of the present invention.
6 is a flowchart illustrating a container-based dynamic resource distribution method for an IoT edge computing environment according to an embodiment of the present invention.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, since various changes may be made to the embodiments, the scope of the patent application is not limited or limited by these embodiments. It should be understood that all modifications, equivalents and substitutes for the embodiments are included in the scope of the rights.

The terms used in the examples are used for the purpose of description only, and should not be construed as limiting. The singular expression includes the plural expression unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that a feature, number, step, operation, component, part, or a combination thereof described in the specification exists, but one or more other features It should be understood that this does not preclude the existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiment belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

In addition, in the description with reference to the accompanying drawings, the same components are given the same reference numerals regardless of the reference numerals, and the overlapping description thereof will be omitted. In describing the embodiment, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the embodiment, the detailed description thereof will be omitted.

2 is a block diagram illustrating a configuration of a container-based dynamic resource distribution device for an IoT edge computing environment according to an embodiment of the present invention.

Referring to FIG. 2 , according to an embodiment of the present invention, a container-based dynamic resource distribution device for an IoT edge computing environment (hereinafter, abbreviated as 'dynamic resource distribution device') 200 includes a count unit 210 , and the processing unit 220 may be included. Also, according to an embodiment, the dynamic resource distribution apparatus 200 may be configured by selectively adding the monitoring unit 230 .

First, the count unit 210 counts the number of allocated application replicas for the edge nodes included in the cluster in association with the cluster constituting the middle layer of the Kubernetes-based edge computing. That is, the count unit 210 may serve to count the number of application service replicas currently allocated to the edge node.

1, an initially set number (eg, three) of application copies may be allocated to each of the edge nodes, and in this case, the count unit 210 for each edge node, the application to be allocated The number of replicas can be counted as '3'.

The processing unit 220 readjusts the number of application copies required by each of the edge nodes in consideration of the request traffic status from the IoT device. That is, the processing unit 220 may play a role in dynamically determining the number of application copies to be allocated to individual edge nodes in consideration of the degree of request input from the IoT device connected to the edge node.

In the above example in which the number of application replicas allocated to all edge nodes is counted as '3', the processing unit 220 is an edge node to which the four IoT devices are connected according to the amount of request traffic generated by the four IoT devices It is possible to determine the number of required application replicas in '6'. Thereafter, the processing unit 220 may readjust the number of application copies allocated to the corresponding edge node from three to six.

In determining the number of application replicas required by the edge node, the processing unit 220 may create an endpoint object corresponding to the number of request traffic from the IoT device connected to each edge node. The endpoint entity may be a means for temporarily storing request traffic input from the IoT device when processed by the edge node.

In Kubernetes, which is the basis of the present invention, the processing unit 220 can create the endpoint object in real time by being linked to the amount of request traffic input from the IoT device in the same period. For example, as each of the four IoT devices connected to a specific edge node generates request traffic during the same period, the processing unit 220 may create an endpoint entity for each of these four IoT devices.

Thereafter, the processing unit 220 determines that, for an active edge node in which the number of created endpoint entities exceeds a set threshold, the request traffic status of the IoT device is determined to be busy, and increases the number of application copies to allocate make a readjustment

That is, the processing unit 220 determines an edge node in which a large number of request traffic is input from the connected IoT device, and thus a number of endpoint objects exceed the limit value, as an activity edge node, and to the determined activity edge node. You can make adjustments to increase the number of application replicas than the current one.

Here, the threshold value may be a reference value that allows it to be determined whether an edge node is a node currently processing an actual job, based on the endpoint entity, and an arbitrary value ( For example, it can be set to '5').

In the above example in which the number of application replicas allocated to all edge nodes is counted as '3', the processing unit 220 determines if the endpoint object created in Location A is '6', which is more than the threshold value '5', the corresponding edge By determining a node as an active edge node, the number of allocated application replicas can be readjusted from three to six.

In another embodiment, the processing unit 220 determines that, for an inactive edge node in which the number of created endpoint entities is less than or equal to the threshold value or '0', the request traffic status from the IoT device is determined to be free, and the At least a portion of the application replica allocated to the active edge node is moved and allocated to the active edge node for readjustment.

That is, the processing unit 220 determines an edge node that is not currently involved in an actual operation due to a small number of created endpoint objects as an inactive edge node, and sets some application copies allocated to the determined inactive edge node to an active edge. You can move it to a node to allocate it.

In the above example in which the number of application replicas allocated to all edge nodes is counted as '3', the processing unit 220 determines that the endpoint object created in Location B is '1', which is significantly less than the threshold value '5'. By determining an edge node as an inactive edge node, the number of allocated application copies is readjusted from three to one, and the remaining two application copies are moved to Location A, which is an active edge node, to be allocated.

In another embodiment, the processing unit 220 determines the preference rule ratio for each of the edge nodes according to the connection frequency with the IoT device. That is, the processing unit 220 may give priority to individual edge nodes according to the number of times the IoT device is contacted.

For example, for Location A, where the number of times of contact with the IoT device is large and the connection frequency is the highest, the processing unit 220 may determine a preference rule ratio of '80', which is the highest level, and then to Location C with the highest connection frequency. A preference rule ratio of '15' may be determined for each other, and a preference rule ratio of '5' may be determined for Location B having the lowest connection frequency.

Thereafter, the processing unit 220 may perform a readjustment in which the number of application copies is relatively increased and allocated to the edge node having the determined preference rule ratio high. That is, the processing unit 220 may optimally readjust the number of application copies to be allocated to each edge node according to the determined size of the preference rule ratio.

For example, the processing unit 220 readjusts the allocation of 6 application copies to Location A with a preference rule ratio of '66.66', readjusts the allocation of 2 copies of the application to Location C with a preference rule ratio of '22.22', and the preference rule ratio It is possible to rebalance the application copy to Location B of '11.11' to 1 allocation.

In another embodiment, the dynamic resource distribution apparatus 200 may readjust the number of application copies allocated to individual edge nodes in direct proportion to the number of request traffic.

To this end, the counting unit 210 may count the number of request traffic generated from the IoT device for each edge node during a selected period. That is, the counting unit 210 may count the number of request traffic that occurs simultaneously in a predetermined period in the connected IoT device.

Thereafter, the processing unit 220 may readjust the number of application copies for each edge node to match the counted number of request traffic during the period. That is, the processing unit 220 may determine the number of application copies allocated to each edge node during the period to be equal to the counted number of request traffic.

Through this, the dynamic resource distribution apparatus 200 can appropriately adjust the number of application copies allocated for each edge node so as to be optimal for the currently occurring request traffic, so that insufficient or unnecessary application copies do not exist.

In another embodiment, the dynamic resource distribution apparatus 200 may periodically monitor the real-time state of the client, and dynamically adjust the number of service copies according to the requested resource for each edge node.

To this end, the dynamic resource distribution apparatus 200 may additionally configure the monitoring unit 230 . That is, the monitoring unit 230 may serve to monitor the status of the IoT device, which is a client.

Here, the status of the IoT device may be a communication frequency band used by the IoT device, a status of ambient noise, whether there is an error in a connection protocol, a loss rate during transmission of the request traffic, and the like.

Thereafter, the processing unit 220 may further consider the monitoring result to determine the number of the application replicas to be readjusted. That is, the processing unit 220 may increase or decrease the number of application copies allocated to the edge node connected to the IoT device according to the status of the IoT device checked through monitoring.

For example, as a result of monitoring by the monitoring unit 230, when the ambient noise status of the IoT device is so bad that there are many problems in normally delivering the requested traffic, the processing unit 220 allocates to the edge node connected to the IoT device A rebalance can be made to reduce the number of application replicas being made.

According to an embodiment of the present invention, in consideration of the Kubernetes resource allocation problem, it is possible to provide a container-based dynamic resource distribution method and apparatus for an IoT edge computing environment, which provides a resource provisioning function.

In addition, according to the present invention, it is possible to distribute copies for each application service in proportion to the amount of requests to each edge node in consideration of the client's request traffic status.

In addition, according to the present invention, by periodically monitoring the client's real-time status and dynamically adjusting the number of service copies according to the requested resource for each edge node, the overall system throughput can be improved.

The present invention proposes a method for providing a resource provisioning function in consideration of the conventional Kubernetes resource allocation problem.

The dynamic resource distribution apparatus 200 distributes copies for each application service in proportion to the request amount to each edge node in consideration of the client's request traffic state. For example, in the condition of FIG. 1, the dynamic resource distribution apparatus 200 of the present invention provisions six application service replicas to the edge node Location A, and provisions one application service replica to the edge node Location B, and the edge node You can provision two replicas of the application service at Location C.

In addition, the present invention can improve the overall throughput of the system by periodically monitoring the real-time status of the client and dynamically adjusting the number of application service replicas according to the requested resource for each edge node.

The present invention proposes a technology for dynamically distributing container-based resources in an IoT edge computing environment.

That is, in order to provide high scalability and high availability of IoT application services, the present invention can be applied to an edge computing platform that operates multiple container-based replicas that provide the same service.

In addition, it is an object of the present invention to distribute an appropriate number of replicas to each edge node in consideration of the amount of traffic requested to each edge node in the cluster environment of the edge node.

The dynamic resource distribution device 200 creates an Endpoint object for storing the information in the Kubernetes platform environment in order to collect information on the number of reception requests for each edge node.

The dynamic resource distribution device 200 creates an Endpoint object by periodically counting the number of requests to access the application of each edge node using the user-space proxy mode among the proxy modes of Kubernetes regarding the distribution module of user request traffic. .

In other words, the master node in charge of Kubernetes can read the endpoint object on each edge node to know the network traffic status.

In addition, the scheduler of the master node can use the priority rule called nodeAffinity to specify the edge node to which the replica for each application will be distributed. There are two types of nodeAffinity: Required and Preferred.

The mandatory rule means that the selected edge node must satisfy the requirements of replicas for each application. In the preferred rule, the weight field for each edge node is specified from 1 to 100, and each edge node with a high weight value can be selected. Applications can be given priority.

In a Kubernetes cluster, each node has a key-value label.

Therefore, the configuration of the nodeAffinity rule that should place a replica on a particular node is based on the node's label.

The dynamic resource distribution apparatus 200 may utilize a preference rule to allocate a copy of each application.

First, the label of each edge node is determined based on the node's location.

Edge node A has a value determined by using the key as the location and the value as the location name (eg, the label of edge node A: location - location A).

In addition, the value of the weight field of each edge node is determined by the rate of requesting access to the application from the location.

3 is a diagram illustrating setting of nodeAffinity for scheduling according to an embodiment of the present invention.

As shown in FIG. 3, the rate of requesting access to the “IoT-app-1” application from edge node A is 80%, and the rate of requesting access to edge node B is distributed at 20%.

Accordingly, in FIG. 3 , the weight is set to 80 ( 310 in FIG. 3 ) for the edge node A and 20 ( 320 in FIG. 3 ) for the edge node B.

4 is a view for explaining the components and operation of the technology proposed in the present invention.

The present invention can operate on the Kubernetes platform, which is an open source project for container orchestration.

As shown in FIG. 4 , the present invention can be implemented in the form of a control module for application services on a master node of Kubernetes.

That is, a worker node that actually operates in the IoT edge infrastructure environment operates as an edge node, and the kube-proxy of each worker node can count the number of incoming requests for each application deployed to each edge node.

In the control module on the master node named ElssticFog, it is possible to periodically collect the latest information on the application, edge node and network status.

In addition, the master node calculates the ratio of the request amount received from each edge node based on the collected information and updates the nodeAffinity rule.

Finally, the Kubernetes scheduler can redistribute the number of replicas per application according to the nodeAffinity rule.

5 is a diagram for explaining an operation algorithm in ElasticFog according to an embodiment of the present invention.

The operation algorithm of FIG. 5 describes an algorithm operating in the ElasticFog control module on the Kubernetes master node among the components of FIG. 4 .

The applicationList of FIG. 5 may be a list of applications running in the cluster, and the nodeList may be a list of edge nodes in the cluster.

The LocationInfo of FIG. 5 represents a node list (eg, location A:[node1]. location B: [node2]) of each location, and the applicationRequestInfo may store the number of requests received for each application in each edge node (eg, node1). : 800, node2: 200).

The weightInfo of FIG. 5 may indicate the ratio of requests received at each location of the application (eg, location A: 80, location B: 20).

With the Kubernetes API, nodeList and applicationList can be searched through the search function.

In addition, the locationInfo of FIG. 5 is updated based on the label of the edge node, and the applicationReuestInfo may be collected by reading the Endpoint object corresponding to the application.

The operation algorithm may be repeated for each application in the cluster.

That is, in the present invention, the number of replicas can be dynamically adjusted in proportion to the request amount for each application.

The present invention calculates the number of requests received by an application in a cluster and each location through nodeList, locationInfo, and applicationRequestInfo.

Then, the present invention calculates the proportion of requests received at each location and updates the weightInfo.

In addition, the present invention updates the weight value of each location according to the nodeAffinity rule based on the value of weightInfo.

Finally, according to the present invention, when the nodeAffinity rule is updated for each application, the Kubernetes scheduler redistributes the copies on each node.

According to the present invention, it can be utilized as an edge computing infrastructure for the Internet of Things, and in particular, by dynamically distributing operator resources according to the user's request rate ratio, the system processing performance of the edge computing infrastructure can be improved.

Hereinafter, a work flow of the dynamic resource distribution apparatus 200 according to embodiments of the present invention will be described in detail with reference to FIG. 6 .

6 is a flowchart illustrating a container-based dynamic resource distribution method for an IoT edge computing environment according to an embodiment of the present invention.

The container-based dynamic resource distribution method for the IoT edge computing environment according to the present embodiment may be performed by the dynamic resource distribution apparatus 200 .

First, the dynamic resource distribution apparatus 200 counts the number of allocated application replicas for the edge nodes included in the cluster in association with the cluster constituting the middle layer of the Kubernetes-based edge computing ( 610 ). Step 610 may be a process of counting the number of application service replicas currently allocated to the edge node.

1, an initially set number (eg, three) of application copies may be allocated to each edge node, and in this case, the dynamic resource distribution apparatus 200 allocates for each edge node, You can count the number of application replicas to be '3'.

In addition, the dynamic resource distribution apparatus 200 readjusts the number of application copies required by each of the edge nodes in consideration of the request traffic state in the IoT device ( 620 ). Step 620 may be a process of dynamically determining the number of application copies to be allocated to individual edge nodes in consideration of the degree of request input from the IoT device connected to the edge node.

In the above example in which the number of application copies allocated to all edge nodes is counted as '3', the dynamic resource distribution device 200 connects the four IoT devices according to the amount of request traffic generated by the four IoT devices. It is possible to determine the number of required application replicas in the edge node to be '6'. Thereafter, the dynamic resource distribution apparatus 200 may readjust the number of application copies allocated to the corresponding edge node from three to six.

In determining the number of application replicas required by the edge node, the dynamic resource distribution device 200 may create an endpoint object corresponding to the number of request traffic from the IoT device connected to each edge node. . The endpoint entity may be a means for temporarily storing request traffic input from the IoT device when processed by the edge node.

In Kubernetes, which is the basis of the present invention, the dynamic resource distribution device 200 can create the endpoint object in real time by interworking with the amount of request traffic inputted in the same period from the IoT device. For example, as each of the four IoT devices connected to a specific edge node generates request traffic during the same period, the dynamic resource distribution device 200 may create an endpoint object for each of these four IoT devices. .

Thereafter, the dynamic resource distribution device 200 determines that, for an active edge node in which the number of created endpoint entities exceeds a set limit, the request traffic state of the IoT device is determined to be busy, and the number of application copies is increased. to rebalance the allocation.

That is, the dynamic resource distribution device 200 receives a large amount of request traffic from the connected IoT device, and thus determines an edge node in which a plurality of endpoint objects are created beyond a threshold as an activity edge node, and the determined activity You can make adjustments to the edge node to increase the number of application replicas than the current one.

Here, the threshold value may be a reference value that allows it to be determined whether an edge node is a node currently processing an actual job, based on the endpoint entity, and an arbitrary value ( For example, it can be set to '5').

In the above example in which the number of application replicas allocated to all edge nodes is counted as '3', the dynamic resource distribution device 200 determines that the endpoint object created in Location A is '6', which is more than the limit value '5'. , by determining the corresponding edge node as an active edge node, the number of allocated application replicas can be readjusted from 3 to 6.

In another embodiment, the dynamic resource distribution device 200 determines that the requested traffic status in the IoT device is free for an inactive edge node in which the number of created endpoint entities is less than or equal to the threshold or '0'. , at least a portion of the application copies allocated to the inactive edge node are moved to the active edge node to be allocated and readjusted.

That is, the dynamic resource distribution device 200 determines an edge node that is not currently involved in actual work as an inactive edge node because the number of created endpoint objects is small, and copies of some applications that have been allocated to the determined inactive edge node can be allocated by moving it to the active edge node.

In the above example in which the number of application replicas allocated to all edge nodes is counted as '3', the dynamic resource distribution device 200 indicates that the endpoint object created in Location B is significantly less than the threshold '5'. In this case, the corresponding edge node is determined as an inactive edge node, the number of allocated application copies is readjusted from three to one, and the remaining two application copies are moved to Location A, the active edge node, to be allocated. can

In another embodiment, the dynamic resource distribution device 200 determines the preference rule ratio for each of the edge nodes according to the connection frequency with the IoT device. That is, the dynamic resource distribution device 200 may give priority to individual edge nodes according to the number of times of contact with the IoT device.

For example, for Location A, where the number of times of contact with the IoT device is large, and the connection frequency is the highest, the dynamic resource distribution device 200 may determine a preference rule ratio of '80', which is the highest level, and has the next highest connection frequency. A preference rule ratio of '15' may be determined for Location C, and a preference rule ratio of '5' may be determined for Location B having the smallest connection frequency.

Thereafter, the dynamic resource distribution apparatus 200 may perform readjustment in which the number of application copies is relatively increased and allocated to the edge node having the determined preference rule ratio high. That is, the dynamic resource distribution apparatus 200 may optimally readjust the number of application copies to be allocated to each edge node according to the determined size of the preference rule ratio.

For example, the dynamic resource distribution device 200 readjusts the allocation of 6 application copies to Location A of the preference rule ratio '66.66', and readjusts the application copies to Location C of the preference rule ratio '22.22' to 2 allocations, It is possible to readjust one application copy to Location B with a preferred rule ratio of '11.11'.

In another embodiment, the dynamic resource distribution apparatus 200 may readjust the number of application copies allocated to individual edge nodes in direct proportion to the number of request traffic.

To this end, the dynamic resource distribution apparatus 200 may count the number of request traffic from the IoT device generated for each edge node during a selected period. That is, the dynamic resource distribution device 200 may count the number of request traffic that occurs simultaneously in a predetermined period in the connected IoT device.

Thereafter, the dynamic resource distribution apparatus 200 may readjust the number of application copies for each edge node to match the counted number of request traffic during the period. That is, the dynamic resource distribution apparatus 200 may determine the number of application copies allocated to each edge node during the period to be equal to the counted number of request traffic.

Through this, the dynamic resource distribution apparatus 200 can appropriately adjust the number of application copies allocated for each edge node so as to be optimal for the currently occurring request traffic, so that insufficient or unnecessary application copies do not exist.

In another embodiment, the dynamic resource distribution apparatus 200 may periodically monitor the real-time state of the client, and dynamically adjust the number of service copies according to the requested resource for each edge node.

That is, the dynamic resource distribution apparatus 200 may monitor the state of the IoT device, which is a client.

Here, the status of the IoT device may be a communication frequency band used by the IoT device, a status of ambient noise, whether there is an error in a connection protocol, a loss rate during transmission of the request traffic, and the like.

Thereafter, the dynamic resource distribution apparatus 200 may determine the number of the application replicas to be readjusted by further considering the monitoring result. That is, the dynamic resource distribution device 200 may increase or decrease the number of application copies allocated to the edge node connected to the IoT device according to the status of the IoT device checked through monitoring.

For example, as a result of monitoring the dynamic resource distribution device 200 , when the ambient noise status of the IoT device is so bad that there are many problems in normally delivering the request traffic, the dynamic resource distribution device 200 is connected to the IoT device Rebalancing can be done to reduce the number of application replicas allocated to the edge nodes that become

According to an embodiment of the present invention, in consideration of the Kubernetes resource allocation problem, it is possible to provide a container-based dynamic resource distribution method and apparatus for an IoT edge computing environment, which provides a resource provisioning function.

In addition, according to the present invention, it is possible to distribute copies for each application service in proportion to the amount of requests to each edge node in consideration of the client's request traffic status.

In addition, according to the present invention, by periodically monitoring the client's real-time status and dynamically adjusting the number of service copies according to the requested resource for each edge node, the overall system throughput can be improved.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or apparatus, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

As described above, although the embodiments have been described with reference to the limited drawings, those skilled in the art may apply various technical modifications and variations based on the above. For example, the described techniques are performed in an order different from the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

200: dynamic resource distribution device
210: count unit 220: processing unit
230: monitoring unit

Claims (13)

With respect to the cluster that constitutes the middle tier of Kubernetes-based edge computing,
counting the number of allocated application replicas for the edge nodes included in the cluster; and
Re-adjusting the number of application copies required by each edge node in consideration of the request traffic status from the IoT device
A container-based dynamic resource distribution method for an IoT edge computing environment comprising a. According to claim 1,
The step of re-adjusting the number of application replicas,
creating an endpoint entity, corresponding to the number of request traffic from the connected IoT device, to each edge node; and
Re-adjusting to increase the number of application replicas by judging that the request traffic status from the IoT device is busy for an active edge node in which the number of created endpoint entities exceeds a set threshold
A container-based dynamic resource distribution method for an IoT edge computing environment comprising a. 3. The method of claim 2,
The step of re-adjusting the number of application replicas,
For an inactive edge node in which the number of created endpoint entities is less than or equal to the threshold value or '0', the status of request traffic from the IoT device is determined as free, and at least a portion of an application copy allocated to the inactive edge node moving to the active edge node and assigning it to readjust
Container-based dynamic resource distribution method for IoT edge computing environment further comprising. According to claim 1,
The step of re-adjusting the number of application replicas,
determining, with each of the edge nodes, a preference rule ratio according to a connection frequency with the IoT device; and
Re-arranging to allocate a relatively increasing number of application replicas for the determined edge node having a high preference rule ratio
A container-based dynamic resource distribution method for an IoT edge computing environment comprising a. According to claim 1,
Counting the number of request traffic from the IoT device generated during a selected period for each edge node
further comprising,
The step of re-adjusting the number of application replicas,
Rebalancing the number of application replicas for each edge node to match the counted number of request traffic during the period
A container-based dynamic resource distribution method for an IoT edge computing environment comprising a. According to claim 1,
monitoring the status of the IoT device; and
Determining the number of the application replicas to be readjusted, further considering the results of the monitoring
Container-based dynamic resource distribution method for IoT edge computing environment further comprising. With respect to the cluster that constitutes the middle tier of Kubernetes-based edge computing,
a counting unit for counting the number of allocated application copies for the edge nodes included in the cluster; and
A processing unit that readjusts the number of application copies required by each of the edge nodes in consideration of the request traffic status from the IoT device
A container-based dynamic resource distribution device for an IoT edge computing environment comprising a. 8. The method of claim 7,
The processing unit,
To each edge node, create an endpoint object corresponding to the number of request traffic from the connected IoT device,
For an active edge node in which the number of created endpoint entities exceeds a set threshold, rebalancing to increase and allocate the number of application copies by determining that the request traffic status in the IoT device is busy
Container-based dynamic resource distribution device for IoT edge computing environment. 9. The method of claim 8,
The processing unit,
For an inactive edge node in which the number of created endpoint entities is less than or equal to the threshold value or '0', the status of request traffic from the IoT device is determined as free, and at least a portion of an application copy allocated to the inactive edge node , which moves to the active edge node and allocates
Container-based dynamic resource distribution device for IoT edge computing environment. 8. The method of claim 7,
The processing unit,
According to the connection frequency with the IoT device, each edge node determines a preference rule ratio,
For the edge node having the determined preference rule ratio high, readjustment is performed by relatively increasing the number of application replicas and allocating
Container-based dynamic resource distribution device for IoT edge computing environment. 8. The method of claim 7,
The count unit,
Count the number of request traffic from the IoT device generated during the selected period for each edge node,
The processing unit,
For each edge node, readjusting the number of application replicas to match the counted number of request traffic during the period
Container-based dynamic resource distribution device for IoT edge computing environment. 8. The method of claim 7,
Monitoring unit for monitoring the status of the IoT device
further comprising,
The processing unit,
Further considering the results of the monitoring, determining the number of application replicas to rebalance
Container-based dynamic resource distribution device for IoT edge computing environment. A computer-readable recording medium recording a program for executing the method of any one of claims 1 to 6.

Images