WO2023058782A1 - Hybrid cloud operation method providing workload unit deployment and priority scheduling - Google Patents

Hybrid cloud operation method providing workload unit deployment and priority scheduling Download PDF

Info

Publication number
WO2023058782A1
WO2023058782A1 PCT/KR2021/013562 KR2021013562W WO2023058782A1 WO 2023058782 A1 WO2023058782 A1 WO 2023058782A1 KR 2021013562 W KR2021013562 W KR 2021013562W WO 2023058782 A1 WO2023058782 A1 WO 2023058782A1
Authority
WO
WIPO (PCT)
Prior art keywords
workload
hybrid cloud
workloads
cloud
management system
Prior art date
Application number
PCT/KR2021/013562
Other languages
French (fr)
Korean (ko)
Inventor
김동민
손재기
Original Assignee
한국전자기술연구원
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 한국전자기술연구원 filed Critical 한국전자기술연구원
Publication of WO2023058782A1 publication Critical patent/WO2023058782A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533—Hypervisors; Virtual machine monitors
    • G06F9/45558—Hypervisor-specific management and integration aspects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/30—Monitoring
    • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/30—Monitoring
    • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
    • G06F11/3409—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment
    • G06F11/3433—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment for load management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5061—Partitioning or combining of resources
    • G06F9/5072—Grid computing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/54—Interprogram communication
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/54—Interprogram communication
    • G06F9/546—Message passing systems or structures, e.g. queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06316—Sequencing of tasks or work
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0633—Workflow analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533—Hypervisors; Virtual machine monitors
    • G06F9/45558—Hypervisor-specific management and integration aspects
    • G06F2009/4557—Distribution of virtual machine instances; Migration and load balancing

Definitions

  • the present invention relates to a hybrid cloud management technology, and more particularly, to a method for deploying workloads for machine learning model development in a hybrid cloud operating environment.
  • the present invention has been made to solve the above problems, and an object of the present invention is to appropriately place workloads for machine learning models in a hybrid cloud operating environment composed of a plurality of cloud environments, and to load workloads between clouds. It is to provide a method of rearrangement through the movement of.
  • a hybrid cloud-based workload management system includes a workload-queue in which workloads are stored; and a master for arranging the workload newly stored in the workload-queue to one of the clouds constituting the hybrid cloud.
  • a hybrid cloud-based workload management system includes a workload scheduler that calculates a priority of a workload newly stored in a workload-queue; and a priority-queue in which the calculated priority is stored, and the master may determine a cloud to place the workload based on the priority order.
  • the master may determine a cloud with higher availability as the priority of the workload is higher.
  • the master collects logs of clouds
  • the hybrid cloud-based workload management system may further include a dynamic workload scheduler that rearranges deployed workloads based on the collected logs of clouds.
  • a hybrid cloud-based workload management system includes a collector for collecting logs of deployed workloads; an analyzer that analyzes the collected logs; and a detector for detecting anomaly from the analysis result, and the master may redeploy the workload for which the anomaly is detected by the detector to another cloud.
  • the master may rearrange pre-located workloads based on the priorities of the workloads stored in the priority-queue.
  • Workloads may be workloads for machine learning model development.
  • a hybrid cloud-based workload placement method includes storing a workload in a workload-queue; and deploying the stored workload to one of the clouds constituting the hybrid cloud.
  • a machine learning service developer can be provided with a highly convenient workload placement environment in a hybrid cloud operating environment.
  • a stable workload operating environment can be provided by collecting and analyzing logs of workloads running in a hybrid cloud environment.
  • FIG. 1 is a diagram showing a hybrid cloud-based machine learning service system to which the present invention is applicable;
  • FIG. 2 is a detailed block diagram of a workload management system according to an embodiment of the present invention.
  • FIG. 3 is a flowchart provided to explain a hybrid cloud-based workload placement method according to another embodiment of the present invention.
  • FIG. 4 is a diagram showing the hardware structure of a workload management system according to an embodiment of the present invention.
  • a hybrid cloud-based workload placement/relocation (transition) method is proposed.
  • the priority of the workload is considered, and the workloads are rearranged and scheduled by analyzing logs of future clouds and workloads.
  • a hybrid cloud-based machine learning service system to which the present invention is applicable includes a workload management system 100 and a hybrid cloud 200.
  • the hybrid cloud 200 is a cloud system configured by integrating multiple clouds.
  • the workload management system 100 arranges workloads necessary for machine learning development in the hybrid cloud 200 as containers, and rearranges and manages the workloads deployed in the hybrid cloud 200 .
  • FIG. 2 is a detailed block diagram of a workload management system 100 according to an embodiment of the present invention.
  • the hybrid cloud 200 is further illustrated in FIG. 2 .
  • the hybrid cloud 200 is composed of a plurality of clouds 210 to 240
  • Cloud-1 210 is a cloud provided by Amazon
  • Cloud-2 220 is a cloud provided by Microsoft.
  • Cloud-3 (230) can be configured as a cloud provided by Google
  • Cloud-4 (240) can be configured as a cloud provided by Naver.
  • the workload management system 100 includes a log collector 110, a system monitor 120, a log analyzer 130, a hybrid cloud master 140, and an anomaly detector 150. ), a priority-queue 160, a dynamic workload scheduler 170, a workload scheduler 180, and a workload-queue 190.
  • the log collector 110 collects logs of workloads deployed in the clouds 210 to 240, the system monitor 120 monitors the state of the hybrid cloud 200, and the log analyzer 130 is a log collector ( 110) to analyze the collected logs.
  • the anomaly detector 150 is a component that detects anomalies from the analysis result by the log analyzer 130 and is located in the hybrid cloud master 140 .
  • the hybrid cloud master 140 deploys workloads to the clouds 210 to 240 and redeploys the deployed workloads.
  • the priority order of the workloads stored in the priority-queue 160 is referred to for workload arrangement, and the detection result of the anomaly detector 150 is referred to for workload rearrangement.
  • the hybrid cloud master 140 collects logs of the clouds 210 to 240, and the dynamic workload scheduler 170 determines the clouds 210 to 240 based on the collected logs of the clouds 210 to 240. ) to relocate the deployed workloads.
  • Workloads are stored in the workload-queue 190, and the workload scheduler 180 calculates the priority of a workload newly stored in the workload-queue 190 and stores it in the priority-queue 160.
  • priorities of workloads are stored in the priority-queue 160.
  • the hybrid cloud master 140 arranges workloads in the clouds 210 to 240 based on priority order. do.
  • the hybrid cloud master 140 may rearrange the workloads disposed in the clouds 210 to 240 by comparing priorities of the workloads accumulated and stored in the priority-queue 160 .
  • FIG. 3 is a flowchart provided to explain a hybrid cloud-based workload placement method according to another embodiment of the present invention.
  • the workload scheduler 180 calculates the priority of the newly stored workload in step S310 and prioritizes the priority- It is stored in the queue 160 (S320).
  • the hybrid cloud master 140 determines a cloud to place the stored workload in step S310 based on the priority calculated in step S320 and places it as a container (S330).
  • step S330 the higher the priority of the workload, the higher the availability and the higher the quality / state of the resource. decide
  • the hybrid cloud master 140 collects logs of the clouds 210 to 240 (S340). Then, the dynamic workload scheduler 170 rearranges the workloads deployed in the clouds 210 to 240 based on the logs of the clouds 210 to 240 collected in step S340 (S350).
  • step S350 a workload placed in a cloud with low availability may be relocated to a cloud with high availability.
  • the log collector 110 collects logs of workloads deployed in the clouds 210 to 240, and the log analyzer 130 analyzes the logs collected by the log collector 110 (S360).
  • the anomaly detector 150 detects anomaly from the analysis result in step S360, and the hybrid cloud master 140 redeploys the workload for which the anomaly is detected to another cloud (S370).
  • the workload management system 100 can be implemented as a computing system including a communication unit 101 , a processor 102 and a storage unit 103 .
  • the communication unit 101 is a means for communicating with the clouds 210 to 240 constituting the hybrid cloud 200 and communicating with a terminal of a developer developing a workload.
  • the processor 102 includes a log collector 110, a system monitor 120, a log analyzer 130, a hybrid cloud master 140, an anomaly detector 150, and a dynamic workload scheduler 170 among the components shown in FIG. ), which is a hardware means for performing the function of the workload scheduler 180.
  • the storage unit 103 is a storage medium for implementing the priority-queue 160 and the workload-queue 190 .
  • the workloads assume workloads for machine learning model development, which are merely exemplary.
  • the technical idea of the present invention can be applied even when it is replaced with other types of workloads.
  • machine learning service developers are provided with a highly convenient workload placement environment, and can maximize the use of cloud resources by relocating workloads between cloud environments in a hybrid cloud. , By collecting and analyzing logs of workloads running in a hybrid cloud environment, it is possible to provide a stable workload operation environment.
  • the technical spirit of the present invention can also be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment.
  • technical ideas according to various embodiments of the present invention may be implemented in the form of computer readable codes recorded on a computer readable recording medium.
  • the computer-readable recording medium may be any data storage device that can be read by a computer and store data.
  • the computer-readable recording medium may be ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, and the like.
  • computer readable codes or programs stored on a computer readable recording medium may be transmitted through a network connected between computers.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Quality & Reliability (AREA)
  • Educational Administration (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Computer Hardware Design (AREA)
  • Mathematical Physics (AREA)
  • Debugging And Monitoring (AREA)

Abstract

A hybrid cloud-based workload deployment method is provided. A hybrid cloud-based workload management system, according to an embodiment of the present invention, comprises: a workload queue in which workloads are stored; and a master which deploys workloads newly stored in the workload queue into one of clouds forming a hybrid cloud. Accordingly, in a hybrid cloud operation environment, machine learning service developers can be provided with a highly convenient workload deployment environment, and workload re-placement is possible, and thus, it is possible to maximize the use of cloud resources.

Description

워크로드 단위 배치 및 우선순위 스케줄링을 제공하는 하이브리드 클라우드 운용 방법A hybrid cloud operation method that provides workload unit placement and priority scheduling
본 발명은 하이브리드 클라우드 관리 기술에 관한 것으로, 더욱 상세하게는 하이브리드 클라우드 운영 환경에서 머신러닝 모델 개발을 위한 워크로드들을 배치하는 방법에 관한 것이다.The present invention relates to a hybrid cloud management technology, and more particularly, to a method for deploying workloads for machine learning model development in a hybrid cloud operating environment.
머신 러닝 모델 개발을 위한 워크로드들은 구동되는 서버 및 클라우드 환경, 리소스 정보 등을 고려하여 적정의 클라우드에 배치되어야 높은 서비스 품질을 보장할 수 있다.Workloads for machine learning model development must be placed in an appropriate cloud in consideration of running server and cloud environments, resource information, etc. to ensure high service quality.
이에 따라, 워크로드들을 적정의 클라우드에 배치시키기 위한 방안이 필요하다. 또한, 일단 배치된 워크로드라 할지라도 사후 환경 변경에 따른 지속적인 관리가 요구된다.Accordingly, there is a need for a plan for deploying workloads to an appropriate cloud. In addition, even once deployed workloads, continuous management according to post-environmental changes is required.
이는, 클라우드 환경과 워크로드 상태 모두를 고려하여 복합적인 방식으로 이루어져야 한다.This must be done in a complex way, taking into account both the cloud environment and the state of the workload.
본 발명은 상기와 같은 문제점을 해결하기 위하여 안출된 것으로서, 본 발명의 목적은, 다수의 클라우드 환경으로 구성된 하이브리드 클라우드 운용 환경에서, 머신러닝 모델을 위한 워크로드들을 적정하게 배치하고, 클라우드 간 워크로드의 이동을 통해 재배치하는 방법을 제공함에 있다.The present invention has been made to solve the above problems, and an object of the present invention is to appropriately place workloads for machine learning models in a hybrid cloud operating environment composed of a plurality of cloud environments, and to load workloads between clouds. It is to provide a method of rearrangement through the movement of.
상기 목적을 달성하기 위한 본 발명의 일 실시예에 따른, 하이브리드 클라우드 기반 워크로드 관리 시스템은, 워크로드들이 저장되는 워크로드-큐; 및 워크로드-큐에 새롭게 저장된 워크로드를 하이브리드 클라우드를 구성하는 클라우드들 중 하나에 배치하는 마스터;를 포함한다.According to an embodiment of the present invention for achieving the above object, a hybrid cloud-based workload management system includes a workload-queue in which workloads are stored; and a master for arranging the workload newly stored in the workload-queue to one of the clouds constituting the hybrid cloud.
본 발명의 실시예에 따른 하이브리드 클라우드 기반 워크로드 관리 시스템은, 워크로드-큐에 새롭게 저장된 워크로드의 우선순위를 계산하는 워크로드 스케줄러; 및 계산된 우선순위가 저장되는 우선순위-큐;를 더 포함하고, 마스터는, 운선순위를 기초로, 워크로드를 배치할 클라우드를 결정할 수 있다.A hybrid cloud-based workload management system according to an embodiment of the present invention includes a workload scheduler that calculates a priority of a workload newly stored in a workload-queue; and a priority-queue in which the calculated priority is stored, and the master may determine a cloud to place the workload based on the priority order.
마스터는, 워크로드의 우선순위가 높을수록 가용도가 높은 클라우드를 결정할 수 있다.The master may determine a cloud with higher availability as the priority of the workload is higher.
마스터는, 클라우드들의 로그를 수집하고, 하이브리드 클라우드 기반 워크로드 관리 시스템은, 수집된 클라우드들의 로그를 기초로, 배치된 워크로드들을 재배치하는 동적 워크로드 스케줄러;를 더 포함할 수 있다.The master collects logs of clouds, and the hybrid cloud-based workload management system may further include a dynamic workload scheduler that rearranges deployed workloads based on the collected logs of clouds.
본 발명의 실시예에 따른 하이브리드 클라우드 기반 워크로드 관리 시스템은, 배치된 워크로드들의 로그를 수집하는 수집기; 수집된 로그를 분석하는 분석기; 및 분석 결과로부터 이상을 감지하는 탐지기;를 더 포함하고, 마스터는, 탐지기에 의해 이상이 감지된 워크로드를 다른 클라우드에 재배치할 수 있다.A hybrid cloud-based workload management system according to an embodiment of the present invention includes a collector for collecting logs of deployed workloads; an analyzer that analyzes the collected logs; and a detector for detecting anomaly from the analysis result, and the master may redeploy the workload for which the anomaly is detected by the detector to another cloud.
마스터는, 우선순위-큐에 저장된 워크로드들의 우선순위를 기초로, 기배치된 워크로드들을 재배치할 수 있다.The master may rearrange pre-located workloads based on the priorities of the workloads stored in the priority-queue.
워크로드들은, 머신러닝 모델 개발을 위한 워크로드들일 수 있다.Workloads may be workloads for machine learning model development.
한편, 본 발명의 다른 실시예에 따른, 하이브리드 클라우드 기반 워크로드 배치 방법은, 워크로드-큐에 워크로드를 저장하는 단계; 및 저장된 워크로드를 하이브리드 클라우드를 구성하는 클라우드들 중 하나에 배치하는 단계;를 포함한다.Meanwhile, according to another embodiment of the present invention, a hybrid cloud-based workload placement method includes storing a workload in a workload-queue; and deploying the stored workload to one of the clouds constituting the hybrid cloud.
이상 설명한 바와 같이, 본 발명의 실시예들에 따르면, 하이브리드 클라우드 운용 환경에서 머신러닝 서비스 개발자는 편의성 높은 워크로드 배치 환경을 제공받을 수 있게 된다.As described above, according to the embodiments of the present invention, a machine learning service developer can be provided with a highly convenient workload placement environment in a hybrid cloud operating environment.
또한, 본 발명의 실시예들에 따르면, 하이브리드 클라우드 환경에서 클라우드 간 워크로드 재배치가 가능함에 따라, 클라우드 리소스의 사용을 극대화 할 수 있게 된다.In addition, according to embodiments of the present invention, as workloads can be relocated between clouds in a hybrid cloud environment, the use of cloud resources can be maximized.
뿐만 아니라, 본 발명의 실시예들에 따르면, 하이브리드 클라우드 환경에서 구동중인 워크로드들의 로그를 수집, 분석함에 따라 안정된 워크로드 운용 환경을 제공할 수 있게 된다.In addition, according to embodiments of the present invention, a stable workload operating environment can be provided by collecting and analyzing logs of workloads running in a hybrid cloud environment.
도 1은 본 발명이 적용가능한 하이브리드 클라우드 기반 머신러닝 서비스 시스템을 도시한 도면,1 is a diagram showing a hybrid cloud-based machine learning service system to which the present invention is applicable;
도 2는 본 발명의 일 실시예에 따른 워크로드 관리 시스템의 상세 블럭도,2 is a detailed block diagram of a workload management system according to an embodiment of the present invention;
도 3은 본 발명의 다른 실시예에 따른 하이브리드 클라우드 기반 워크로드 배치 방법의 설명에 제공되는 흐름도, 그리고,3 is a flowchart provided to explain a hybrid cloud-based workload placement method according to another embodiment of the present invention, and
도 4는 본 발명의 일 실시예에 따른 워크로드 관리 시스템의 하드웨어 구조를 도시한 도면이다.4 is a diagram showing the hardware structure of a workload management system according to an embodiment of the present invention.
이하에서는 도면을 참조하여 본 발명을 보다 상세하게 설명한다.Hereinafter, the present invention will be described in more detail with reference to the drawings.
본 발명의 실시예에서는 하이브리드 클라우드 기반 워크로드 배치/재배치(전이) 방법을 제시한다.In an embodiment of the present invention, a hybrid cloud-based workload placement/relocation (transition) method is proposed.
구체적으로, 하이브리드 클라우드 운영 환경에서 머신러닝 모델 개발을 위한 워크로드들을 배치 스케줄링함에 있어 워크로드의 우선순위를 고려하고, 향후 클라우드들과 워크로드의 로그를 분석하여 워크로드들을 재배치 스케줄링한다.Specifically, in the batch scheduling of workloads for machine learning model development in a hybrid cloud operating environment, the priority of the workload is considered, and the workloads are rearranged and scheduled by analyzing logs of future clouds and workloads.
도 1은 본 발명이 적용가능한 하이브리드 클라우드 기반 머신러닝 서비스 시스템을 도시한 도면이다. 본 발명이 적용가능한 하이브리드 클라우드 기반 머신러닝 서비스 시스템은, 워크로드 관리 시스템(100)과 하이브리드 클라우드(200)를 포함하여 구성된다.1 is a diagram illustrating a hybrid cloud-based machine learning service system to which the present invention is applicable. A hybrid cloud-based machine learning service system to which the present invention is applicable includes a workload management system 100 and a hybrid cloud 200.
하이브리드 클라우드(200)는 다수의 클라우드들을 통합하여 구성하는 클라우드 시스템이다.The hybrid cloud 200 is a cloud system configured by integrating multiple clouds.
워크로드 관리 시스템(100)은 머신러닝 개발을 위해 필요한 워크로드들을 하이브리드 클라우드(200)에 컨테니어들로 배치하고, 하이브리드 클라우드(200)에 배치된 워크로드들을 재배치하여 관리한다.The workload management system 100 arranges workloads necessary for machine learning development in the hybrid cloud 200 as containers, and rearranges and manages the workloads deployed in the hybrid cloud 200 .
이와 같은 기능을 수행하는 워크로드 관리 시스템(100)에 대해, 이하에서 도 2를 참조하여 상세히 설명한다. 도 2는 본 발명의 일 실시예에 따른 워크로드 관리 시스템(100)의 상세 블럭도이다.The workload management system 100 performing such a function will be described in detail with reference to FIG. 2 below. 2 is a detailed block diagram of a workload management system 100 according to an embodiment of the present invention.
이해와 설명의 편의를 위해, 도 2에는 하이브리드 클라우드(200)을 더 도시하였다. 도시된 바와 같이, 하이브리드 클라우드(200)는 다수의 클라우드들(210~240)로 구성되는데, 클라우드-1(210)은 아마존에서 제공하는 클라우드, 클라우드-2(220)는 마이크로소프트에서 제공하는 클라우드, 클라우드-3(230)은 구글에서 제공하는 클라우드, 클라우드-4(240)는 네이버에서 제공하는 클라우드로 구성할 수 있다.For convenience of understanding and explanation, the hybrid cloud 200 is further illustrated in FIG. 2 . As shown, the hybrid cloud 200 is composed of a plurality of clouds 210 to 240, Cloud-1 210 is a cloud provided by Amazon, and Cloud-2 220 is a cloud provided by Microsoft. , Cloud-3 (230) can be configured as a cloud provided by Google, and Cloud-4 (240) can be configured as a cloud provided by Naver.
본 발명의 실시예에 따른 워크로드 관리 시스템(100)은, 도시된 바와 같이, 로그 수집기(110), 시스템 모니터(120), 로그 분석기(130), 하이브리드 클라우드 마스터(140), 이상 탐지기(150), 우선순위-큐(160), 동적 워크로드 스케줄러(170), 워크로드 스케줄러(180), 워크로드-큐(190)를 포함하여 구성된다.As shown, the workload management system 100 according to an embodiment of the present invention includes a log collector 110, a system monitor 120, a log analyzer 130, a hybrid cloud master 140, and an anomaly detector 150. ), a priority-queue 160, a dynamic workload scheduler 170, a workload scheduler 180, and a workload-queue 190.
로그 수집기(110)는 클라우드들(210~240)에 배치된 워크로드들의 로그를 수집하고, 시스템 모니터(120)는 하이브리드 클라우드(200)의 상태를 모니터링하며, 로그 분석기(130)는 로그 수집기(110)에 의해 수집된 로그를 분석한다.The log collector 110 collects logs of workloads deployed in the clouds 210 to 240, the system monitor 120 monitors the state of the hybrid cloud 200, and the log analyzer 130 is a log collector ( 110) to analyze the collected logs.
이상 탐지기(150)는 로그 분석기(130)에 의한 분석 결과로부터 이상을 탐지하는 구성으로, 하이브리드 클라우드 마스터(140)에 위치한다.The anomaly detector 150 is a component that detects anomalies from the analysis result by the log analyzer 130 and is located in the hybrid cloud master 140 .
하이브리드 클라우드 마스터(140)는 워크로드들을 클라우드들(210~240)에 배치하고, 배치된 워크로드들을 재배치한다. 워크로드 배치에는 우선순위-큐(160)에 저장된 워크로드의 운선순위가 참조되고, 워크로드 재배치에는 이상 탐지기(150)에 의한 탐지 결과가 참조된다.The hybrid cloud master 140 deploys workloads to the clouds 210 to 240 and redeploys the deployed workloads. The priority order of the workloads stored in the priority-queue 160 is referred to for workload arrangement, and the detection result of the anomaly detector 150 is referred to for workload rearrangement.
또한, 하이브리드 클라우드 마스터(140)는 클라우드들(210~240)의 로그를 수집하는데, 동적 워크로드 스케줄러(170)는 수집된 클라우드들(210~240)의 로그를 기초로 클라우드들(210~240)에 배치된 워크로드들을 재배치한다.In addition, the hybrid cloud master 140 collects logs of the clouds 210 to 240, and the dynamic workload scheduler 170 determines the clouds 210 to 240 based on the collected logs of the clouds 210 to 240. ) to relocate the deployed workloads.
워크로드-큐(190)에는 워크로드가 저장되는데, 워크로드 스케줄러(180)는 워크로드-큐(190)에 새롭게 저장된 워크로드의 우선순위를 계산하여 우선순위-큐(160)에 저장한다.Workloads are stored in the workload-queue 190, and the workload scheduler 180 calculates the priority of a workload newly stored in the workload-queue 190 and stores it in the priority-queue 160.
이에 따라, 우선순위-큐(160)에는 워크로드들의 우선순위들이 저장되게 되는데, 전술한 바와 같이, 하이브리드 클라우드 마스터(140)는 운선순위를 기초로 워크로드들을 클라우드들(210~240)에 배치한다.Accordingly, priorities of workloads are stored in the priority-queue 160. As described above, the hybrid cloud master 140 arranges workloads in the clouds 210 to 240 based on priority order. do.
나아가, 하이브리드 클라우드 마스터(140)는 우선순위-큐(160)에 누적하여 저장된 워크로드들의 운선순위들을 비교하여, 클라우드들(210~240)에 배치된 워크로드들을 재배치할 수도 있다.Furthermore, the hybrid cloud master 140 may rearrange the workloads disposed in the clouds 210 to 240 by comparing priorities of the workloads accumulated and stored in the priority-queue 160 .
도 3은 본 발명의 다른 실시예에 따른 하이브리드 클라우드 기반 워크로드 배치 방법의 설명에 제공되는 흐름도이다.3 is a flowchart provided to explain a hybrid cloud-based workload placement method according to another embodiment of the present invention.
도시된 바와 같이, 새로운 워크로드가 생성되어 워크로드-큐(190)에 저장되면(S310-Y), 워크로드 스케줄러(180)는 S310단계에서 새롭게 저장된 워크로드의 우선순위를 계산하여 우선순위-큐(160)에 저장한다(S320).As shown, when a new workload is created and stored in the workload-queue 190 (S310-Y), the workload scheduler 180 calculates the priority of the newly stored workload in step S310 and prioritizes the priority- It is stored in the queue 160 (S320).
그러면, 하이브리드 클라우드 마스터(140)는 S320단계에서 계산된 우선순위를 기초로, S310단계에서 저장된 워크로드를 배치할 클라우드를 결정하여 컨테이너로 배치한다(S330).Then, the hybrid cloud master 140 determines a cloud to place the stored workload in step S310 based on the priority calculated in step S320 and places it as a container (S330).
구체적으로, S330단계에서는 워크로드의 우선순위가 높을수록 가용도가 높고 리소스의 품질/상태가 좋은 클라우드를 결정하고, 워크로드의 우선순위가 낮을수록 가용도가 낮고 리소스의 품질/상태가 낮은 클라우드를 결정한다.Specifically, in step S330, the higher the priority of the workload, the higher the availability and the higher the quality / state of the resource. decide
이후, 하이브리드 클라우드 마스터(140)는 클라우드들(210~240)의 로그를 수집한다(S340). 그러면, 동적 워크로드 스케줄러(170)는 S340단계에서 수집된 클라우드들(210~240)의 로그를 기초로, 클라우드들(210~240)에 배치된 워크로드들을 재배치한다(S350).Thereafter, the hybrid cloud master 140 collects logs of the clouds 210 to 240 (S340). Then, the dynamic workload scheduler 170 rearranges the workloads deployed in the clouds 210 to 240 based on the logs of the clouds 210 to 240 collected in step S340 (S350).
구체적으로, S350단계에서는 가용도가 낮은 클라우드에 배치된 워크로드를 가용도가 높은 클라우드에 재배치할 수 있다.Specifically, in step S350, a workload placed in a cloud with low availability may be relocated to a cloud with high availability.
한편, 로그 수집기(110)는 클라우드들(210~240)에 배치된 워크로드들의 로그를 수집하고, 로그 분석기(130)는 로그 수집기(110)에 의해 수집된 로그들을 분석한다(S360).Meanwhile, the log collector 110 collects logs of workloads deployed in the clouds 210 to 240, and the log analyzer 130 analyzes the logs collected by the log collector 110 (S360).
이상 탐지기(150)는 S360단계에서의 분석 결과로부터 이상을 감지하는데, 하이브리드 클라우드 마스터(140)는 이상이 감지된 워크로드를 다른 클라우드에 재배치한다(S370).The anomaly detector 150 detects anomaly from the analysis result in step S360, and the hybrid cloud master 140 redeploys the workload for which the anomaly is detected to another cloud (S370).
도 4는 본 발명의 일 실시예에 따른 워크로드 관리 시스템(100)의 하드웨어 구조를 도시한 도면이다. 본 발명의 실시예에 따른 워크로드 관리 시스템(100)은, 도시된 바와 같이, 통신부(101), 프로세서(102) 및 저장부(103)를 포함하여 구성되는 컴퓨팅 시스템으로 구현가능하다.4 is a diagram showing the hardware structure of the workload management system 100 according to an embodiment of the present invention. As shown, the workload management system 100 according to an embodiment of the present invention can be implemented as a computing system including a communication unit 101 , a processor 102 and a storage unit 103 .
통신부(101)는 하이브리드 클라우드(200)를 구성하는 클라우드들(210~240)과 통신하고, 워크로드를 개발하는 개발자의 단말과 통신하기 위한 수단이다.The communication unit 101 is a means for communicating with the clouds 210 to 240 constituting the hybrid cloud 200 and communicating with a terminal of a developer developing a workload.
프로세서(102)는 도 2에 도시된 구성들 중 로그 수집기(110), 시스템 모니터(120), 로그 분석기(130), 하이브리드 클라우드 마스터(140), 이상 탐지기(150), 동적 워크로드 스케줄러(170), 워크로드 스케줄러(180)의 기능을 수행하기 위한 하드웨어 수단이다.The processor 102 includes a log collector 110, a system monitor 120, a log analyzer 130, a hybrid cloud master 140, an anomaly detector 150, and a dynamic workload scheduler 170 among the components shown in FIG. ), which is a hardware means for performing the function of the workload scheduler 180.
저장부(103)는 우선순위-큐(160)와 워크로드-큐(190)를 구현하기 위한 저장매체이다.The storage unit 103 is a storage medium for implementing the priority-queue 160 and the workload-queue 190 .
지금까지, 하이브리드 클라우드 기반 워크로드 배치/재배치 스케줄링 방법에 대해 바람직한 실시예를 들어 상세히 설명하였다.So far, the hybrid cloud-based workload placement/relocation scheduling method has been described in detail with a preferred embodiment.
위 실시예에서, 워크로드들은 머신러닝 모델 개발을 위한 워크로드들을 상정하였는데, 예시적인 것에 불과하다. 다른 종류의 워크로드들로 대체되는 경우에도 본 발명의 기술적 사상이 적용될 수 있다.In the above embodiment, the workloads assume workloads for machine learning model development, which are merely exemplary. The technical idea of the present invention can be applied even when it is replaced with other types of workloads.
본 발명의 실시예에 따르면, 하이브리드 클라우드 운용 환경에서 머신러닝 서비스 개발자는 편의성 높은 워크로드 배치 환경을 제공받고, 하이브리드 클라우드에서 클라우드 환경간 워크로드 재배치가 가능함에 따라 클라우드 리소스의 사용을 극대화 할 수 있으며, 하이브리드 클라우드 환경에서 구동중인 워크로드들의 로그를 수집, 분석함에 따라 안정된 워크로드 운용 환경을 제공할 수 있게 된다.According to an embodiment of the present invention, in a hybrid cloud operating environment, machine learning service developers are provided with a highly convenient workload placement environment, and can maximize the use of cloud resources by relocating workloads between cloud environments in a hybrid cloud. , By collecting and analyzing logs of workloads running in a hybrid cloud environment, it is possible to provide a stable workload operation environment.
한편, 본 실시예에 따른 장치와 방법의 기능을 수행하게 하는 컴퓨터 프로그램을 수록한 컴퓨터로 읽을 수 있는 기록매체에도 본 발명의 기술적 사상이 적용될 수 있음은 물론이다. 또한, 본 발명의 다양한 실시예에 따른 기술적 사상은 컴퓨터로 읽을 수 있는 기록매체에 기록된 컴퓨터로 읽을 수 있는 코드 형태로 구현될 수도 있다. 컴퓨터로 읽을 수 있는 기록매체는 컴퓨터에 의해 읽을 수 있고 데이터를 저장할 수 있는 어떤 데이터 저장 장치이더라도 가능하다. 예를 들어, 컴퓨터로 읽을 수 있는 기록매체는 ROM, RAM, CD-ROM, 자기 테이프, 플로피 디스크, 광디스크, 하드 디스크 드라이브, 등이 될 수 있음은 물론이다. 또한, 컴퓨터로 읽을 수 있는 기록매체에 저장된 컴퓨터로 읽을 수 있는 코드 또는 프로그램은 컴퓨터간에 연결된 네트워크를 통해 전송될 수도 있다.Meanwhile, it goes without saying that the technical spirit of the present invention can also be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment. In addition, technical ideas according to various embodiments of the present invention may be implemented in the form of computer readable codes recorded on a computer readable recording medium. The computer-readable recording medium may be any data storage device that can be read by a computer and store data. For example, the computer-readable recording medium may be ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, and the like. In addition, computer readable codes or programs stored on a computer readable recording medium may be transmitted through a network connected between computers.
또한, 이상에서는 본 발명의 바람직한 실시예에 대하여 도시하고 설명하였지만, 본 발명은 상술한 특정의 실시예에 한정되지 아니하며, 청구범위에서 청구하는 본 발명의 요지를 벗어남이 없이 당해 발명이 속하는 기술분야에서 통상의 지식을 가진자에 의해 다양한 변형실시가 가능한 것은 물론이고, 이러한 변형실시들은 본 발명의 기술적 사상이나 전망으로부터 개별적으로 이해되어져서는 안될 것이다.In addition, although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific embodiments described above, and the technical field to which the present invention belongs without departing from the gist of the present invention claimed in the claims. Of course, various modifications are possible by those skilled in the art, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

Claims (8)

  1. 워크로드들이 저장되는 워크로드-큐; 및a workload-queue in which workloads are stored; and
    워크로드-큐에 새롭게 저장된 워크로드를 하이브리드 클라우드를 구성하는 클라우드들 중 하나에 배치하는 마스터;를 포함하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system comprising a; master for arranging a workload newly stored in the workload-queue to one of the clouds constituting the hybrid cloud.
  2. 청구항 1에 있어서,The method of claim 1,
    워크로드-큐에 새롭게 저장된 워크로드의 우선순위를 계산하는 워크로드 스케줄러; 및a workload scheduler that calculates the priority of a workload newly stored in the workload-queue; and
    계산된 우선순위가 저장되는 우선순위-큐;를 더 포함하고,Further comprising a priority-queue in which the calculated priority is stored,
    마스터는,the master,
    운선순위를 기초로, 워크로드를 배치할 클라우드를 결정하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system, which determines a cloud to place a workload based on priority.
  3. 청구항 2에 있어서,The method of claim 2,
    마스터는,the master,
    워크로드의 우선순위가 높을수록 가용도가 높은 클라우드를 결정하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system that determines a cloud with higher availability as the priority of the workload increases.
  4. 청구항 2에 있어서,The method of claim 2,
    마스터는,the master,
    클라우드들의 로그를 수집하고,collect logs from clouds;
    하이브리드 클라우드 기반 워크로드 관리 시스템은,A hybrid cloud-based workload management system,
    수집된 클라우드들의 로그를 기초로, 배치된 워크로드들을 재배치하는 동적 워크로드 스케줄러;를 더 포함하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system further comprising a dynamic workload scheduler that relocates deployed workloads based on the collected logs of clouds.
  5. 청구항 2에 있어서,The method of claim 2,
    배치된 워크로드들의 로그를 수집하는 수집기;a collector that collects logs of deployed workloads;
    수집된 로그를 분석하는 분석기; 및an analyzer that analyzes the collected logs; and
    분석 결과로부터 이상을 감지하는 탐지기;를 더 포함하고,A detector for detecting abnormalities from the analysis results; further comprising,
    마스터는,the master,
    탐지기에 의해 이상이 감지된 워크로드를 다른 클라우드에 재배치하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system that redeploys workloads whose anomalies are detected by detectors to other clouds.
  6. 청구항 2에 있어서,The method of claim 2,
    마스터는,the master,
    우선순위-큐에 저장된 워크로드들의 우선순위를 기초로, 기배치된 워크로드들을 재배치하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system, characterized in that, based on the priorities of the workloads stored in the priority-queue, the previously assigned workloads are relocated.
  7. 청구항 1에 있어서,The method of claim 1,
    워크로드들은,workloads,
    머신러닝 모델 개발을 위한 워크로드들인 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 관리 시스템.A hybrid cloud-based workload management system, characterized in that the workloads for machine learning model development.
  8. 워크로드-큐에 워크로드를 저장하는 단계; 및storing the workload in a workload-queue; and
    저장된 워크로드를 하이브리드 클라우드를 구성하는 클라우드들 중 하나에 배치하는 단계;를 포함하는 것을 특징으로 하는 하이브리드 클라우드 기반 워크로드 배치 방법.Deploying the stored workload to one of the clouds constituting the hybrid cloud; hybrid cloud-based workload deployment method comprising the.
PCT/KR2021/013562 2021-10-05 2021-10-05 Hybrid cloud operation method providing workload unit deployment and priority scheduling WO2023058782A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2021-0131382 2021-10-05
KR1020210131382A KR20230048667A (en) 2021-10-05 2021-10-05 Workload deploy method in a Hybrid Cloud

Publications (1)

Publication Number Publication Date
WO2023058782A1 true WO2023058782A1 (en) 2023-04-13

Family

ID=85804398

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2021/013562 WO2023058782A1 (en) 2021-10-05 2021-10-05 Hybrid cloud operation method providing workload unit deployment and priority scheduling

Country Status (2)

Country Link
KR (1) KR20230048667A (en)
WO (1) WO2023058782A1 (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150172205A1 (en) * 2013-12-13 2015-06-18 International Business Machines Corporation Dynamically Move Heterogeneous Cloud Resources Based on Workload Analysis
US20160299697A1 (en) * 2015-04-08 2016-10-13 Prophetstor Data Services, Inc. Workload-aware i/o scheduler in software-defined hybrid storage system
KR20170076396A (en) * 2015-12-24 2017-07-04 한국과학기술원 A cost-effective virtual machine offloading scheme for streaming cache server placement in hybrid cloud environment
KR20190076693A (en) * 2017-12-22 2019-07-02 부산대학교 산학협력단 Automatic distributing and predicting usage for workload in heterogeneous cloud environment
KR20210060364A (en) * 2019-11-18 2021-05-26 주식회사 위즈온텍 Edge server system supporting hybrid cloud

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150172205A1 (en) * 2013-12-13 2015-06-18 International Business Machines Corporation Dynamically Move Heterogeneous Cloud Resources Based on Workload Analysis
US20160299697A1 (en) * 2015-04-08 2016-10-13 Prophetstor Data Services, Inc. Workload-aware i/o scheduler in software-defined hybrid storage system
KR20170076396A (en) * 2015-12-24 2017-07-04 한국과학기술원 A cost-effective virtual machine offloading scheme for streaming cache server placement in hybrid cloud environment
KR20190076693A (en) * 2017-12-22 2019-07-02 부산대학교 산학협력단 Automatic distributing and predicting usage for workload in heterogeneous cloud environment
KR20210060364A (en) * 2019-11-18 2021-05-26 주식회사 위즈온텍 Edge server system supporting hybrid cloud

Also Published As

Publication number Publication date
KR20230048667A (en) 2023-04-12

Similar Documents

Publication Publication Date Title
JP6571161B2 (en) Method, apparatus, and system for exploring application topology relationships
WO2019117593A1 (en) Cloud computing device and method for supporting container-based resource allocation
CN102349064B (en) Method and apparatus for scalable monitoring of virtual machine environments
CN103810016B (en) Realize method, device and the group system of virtual machine (vm) migration
WO2014104634A1 (en) System and method for dynamically expanding virtual cluster and recording medium on which program for executing the method is recorded
WO2012002635A1 (en) System test apparatus
US8856585B2 (en) Hardware failure mitigation
WO2012033237A1 (en) System testing method
WO2016047814A1 (en) Resource allocation method using cloud api key, and apparatus therefor
WO2013154365A1 (en) Method and apparatus for performing task scheduling in terminal
CN103365702A (en) System and method for tracking process of lightweight virtual machine under IaaS cloud environment
CN109343931B (en) Load balancing oriented application perception virtual machine scheduling method in IaaS environment
WO2012124841A1 (en) Communication test device and method thereof
WO2013137615A1 (en) Method and apparatus for detecting leak of information resource of device
CN102893261B (en) The idle conversion method of sampling and system thereof
WO2023058782A1 (en) Hybrid cloud operation method providing workload unit deployment and priority scheduling
Koller et al. Unified monitoring and analytics in the cloud
KR20200052798A (en) Log analysis framework device of docker container
EP3561761A1 (en) Image analysis method and device
Wei et al. Using sample-based time series data for automated diagnosis of scalability losses in parallel programs
WO2022131404A1 (en) On-device-based data analysis system and method
CN114185656A (en) Test task processing method, device, equipment and storage medium
CN113031969A (en) Equipment deployment inspection method and device, computer equipment and storage medium
Guo et al. Learning-based characterizing and modeling performance bottlenecks of big data workloads
WO2024117302A1 (en) Method for managing ml model distribution by using aiot platform

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21959994

Country of ref document: EP

Kind code of ref document: A1