정보자원 활용 최적화를 위한 클라우드 적용 HPC 모델 연구
A Study on Cloud-applied HPC Model for Optimizing the Use of iNformation Resources
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Cloud 기술을 활용하여 정보자원을 필요한 만큼 즉시 활용할 수 있으며, 가상화 기술을 통해 다양한 환경의 애플리케이션을 공동활용할 수 있다. HPC의 경우 고성능 컴퓨팅 환경을 지원하면서 정보자원을 효율적으로 활용하기 위한 연구가 진행되고 있다. 본 연구는 Cloud 기술을 적용한 IaaS, HPC container, orchestration 모델의 아키텍처를 분석하고 HPC에서 활용할 수 있는 개선된 모델을 제시한다. 개선된 모델은 고속의 컴퓨팅 및 네트워크 환경을 지원하면서 MPI 기반의 분산병렬 연산을 처리한다. 다양한 사용자의 연산 요청처리를 위해 scheduler를 제공하며 정보자원이 부족한 경우 다른 시스템과 연계하도록 federation을 제공한다. 이를 활용하여 기존 HPC 환경을 제공하면서다른 애플리케이션도 실행가능한 환경을 제공하여 정보자원의 활용을 최적화 한다.
Abstract
Information resources can be utilized immediately as needed by utilizing cloud technology, and applications in various environments can be utilized simultaneously through virtualization technology. In the case of HPC, research is being conducted to utilize information resources efficiently while supporting a high-performance computing environment. This study analyzes the architecture of IaaS, HPC container, and orchestration models to which Cloud technology is applied, and presents an improved model that can be utilized in HPC. The improved model handles MPI-based distributed parallel operation while supporting high-speed computing and network environments. A scheduler is provided to handle operation requests from various users, and federation is provided to link with other systems when information resources are insufficient. This study provides an environment in which other applications can be executed while providing the existing HPC environment to optimize the utilization of information resources.
참고문헌 (24)
- W. Gropp, "Using MPI:Portable parallel programming with the messagepassing interface" (1994), MIT Press
- RL. Graham, "Open MPI : A flexible high performance MPI", Proceedings of the 6th International Conference on Parallel Processing and Applied Mathematics (2005)
- User’s guide for mpich, a portable implementation of MPI, "W. Gropp", Argonne National Laboratory, ~ (1996)
- J. Ma, "SLURM : Simple Linux utility for resource management", Proceedings of Job Scheduling Strategies for Parallel Processing (2003)
- NVIDIA Technical overview, "Introducing the cloudnative supercomputing architecture", ~ (2022)
- J. Poort, "How cloud-based supercomputing is changing R&D", Harvard Business Review (2021)
- D. Kusnetzky, "Supercomputer as a service", Virtualization & Cloud Review (2017)
- ATAM: Method for architecture evaluation, "R. Kazman", ~, pp.CMU, SEI-2000-TR-004
- "InfiniBand Trade Association(IBTA)"[URL]
- A. Romanow, "An overview of RDMA over IP" (2003), PFLDnet
- "Computer-aided engineering"[URL]
- Andrew J, "HPC Container Runtimes", High Performance Container Workshop (2020)
- D. M. Jacobsen, "Shifter : Containers for HPC", Cray Users Group Conference (2016)
- GM. Kurtzer, "Singularity:Scientific containers for mobility of compute" (2017), PLOS
- Priedhorsky R, "Charliecloud:Unprivileged containers for user-defined software stacks in HPC", the International Conference for High Performance Computing, Networking, Storage and Analysis (2017)
- D. M. Jacobsen, "Contain this, unleashing docker for HPC", Proceedings of the Cray User Group (2015)
- "What is cloud native?" (2020)[URL]
- "Deploy to Swarm"[URL]
- Hightower K, "Kubernetes: Up and running dive into the future of infrastructure" (2017), OReilly Media, Inc
- N. Zhou, "Container orchestration on HPC systems through kubernetes", Journal of Cloud Computing (2021)
- K. Sheets, "Stig Telfer, Kubernetes, HPC and MPI", StackHPC, ~ (2018)
- T. Menouer, "KCSS : Kubernetes container scheduling strategy", The Journal of Supercomputing, 77 (2021)
- A. M. Beltre, "Enabling HPC workloads on cloud infrastructure using kubernetes container orchestration mechanisms", IEEE, 2019, ~
- L. Bass, "Software architecture in practice", Addison Wesley