Granted Patent B2

US 12,693,906 · App. 17/745,069 · Granted Jul 28, 2026

Inventors: Chunghan Lee (Chiyoda-ku, JP); Byoungkwon Choi (Daejeon, KR); Jinwoo Park (Daejeon, KR); Dongsu Han (Daejeon, KR)

Assignees: TOYOTA JIDOSHA KABUSHIKI KAISHA; KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

G06F9/505G06F9/5016

Abstract

Claims (28)

1 . A container management device including a processor, wherein the processor is configured to:

acquire, for a service including interconnected respective microservices installed with containers for executing processing, a historical workload relating to the service, connection information that is information relating to how the microservices are interconnected, and a service chain for propagation of respective processing related to a workload from a microservice at a front end to a microservice at a back end;

acquire resource information regarding the historical workload;

use the historical workload to generate a prediction model expressing a relationship between the workload of each of the microservices and a resource usage to find a resource usage of each of the microservices from the acquired historical workload, the acquired connection information, and the acquired service chain, the prediction model is a linear regression model when an amount of data in the resource information is greater than a predetermined threshold, and the prediction model is a Gaussian process when the amount of data in the resource information is less than or equal to the predetermined threshold;

determine whether the workload has been generated in the service;

upon determination that the workload has been generated in the service, acquire the workload currently generated in the service, the service chain of the workload, and the connection information;

estimate a current workload of each of the microservices using the acquired workload currently generated in the service, the service chain, and the connection information;

use the prediction model for each of the microservices to find the resource usage, and predict a number of containers required; and

control container installation at a same moment for the respective microservices by installation with the predicted number of containers for each of the microservices.

2 . The container management device of claim 1 , wherein the processor is configured to:

further acquire resource information that is information regarding the historical workload and a resource usage for each of the microservices; and

employ the resource information to generate the prediction model.

3 . The container management device of claim 2 , wherein the prediction model is a predetermined regression model, or is a regression model determined by a Gaussian process.

4 . The container management device of claim 3 , wherein the processor is configured to select the predetermined regression model or the regression model determined by a Gaussian process according to an amount of data in the resource information.

5 . The container management device of claim 1 , wherein the processor is configured to:

identify from the service chain which microservices processing will propagate through;

estimate the workload for each of the microservices; and

employ the prediction model to find the resource usage from the workload estimated for each of the microservices.

6 . The container management device of claim 1 , wherein the connection information is a tree topology, a bus topology, a star topology, a ring topology, or a fully connected topology.

7 . A non-transitory storage medium storing a container management program executable by a computer to perform processing comprising:

acquiring, for a service including interconnected respective microservices installed with containers for executing processing, a historical workload relating to the service, connection information that is information relating to how the microservices are interconnected, and a service chain for propagation of respective processing related to a workload from a microservice at a front end to a microservice at a back end;

acquiring resource information regarding the historical workload;

using the historical workload to generate a prediction model expressing a relationship between the workload of each of the microservices and a resource usage to find a resource usage of each of the microservices from the acquired historical workload, the acquired connection information, and the acquired service chain, the prediction model is a linear regression model when an amount of data in the resource information is greater than a predetermined threshold, and the prediction model is a Gaussian process when the amount of data in the resource information is less than or equal to the predetermined threshold;

determining whether the workload has been generated in the service;

upon determination that the workload has been generated in the service, acquiring the workload currently generated in the service, the service chain of the workload, and the connection information;

estimating a current workload of each of the microservices using the acquired workload currently generated in the service, the service chain, and the connection information;

using the prediction model for each of the microservices to find the resource usage, and predicting a number of containers required; and

controlling container installation at a same moment for the respective microservices by installation with the predicted number of containers for each of the microservices.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (30)