download0 view743
twitter facebook

공공누리This item is licensed Korea Open Government License

Title
VM auto-scaling methods for high throughput computing on hybrid infrastructure
Author(s)
최지은김윤희김서영안윤선최재영
Publication Year
2015-06-03
Abstract
Cloud computing provides on-demand resource provisioning and scalable resources dynamically for the efficient use of computing resources. Scientific applications recently need a very large number of loosely coupled tasks to be handled efficiently. In response, current computing environments often consist of heterogeneous resources such as cloud computing. To effectively use cloud resources, auto-scaling methods that consider diverse metrics such as CPU utilization and costs of resource usage have been studied widely. However it still remains a challenge to automatically and timely allocate resources such that deadline violation and application types are considered. In this paper, we propose auto-scaling methods that consider specific conditions such as application types, task dependency, user-defined deadlines and data transfer times within a hybrid computing infrastructure. Our hybrid computing infrastructure consists of local cluster and cloud resources using HTCaaS. We observe noticeable improvements in performance when our auto-scaling methods for bag-of-tasks and workflow applications is applied.
Keyword
Auto-scaling; Hybrid infrastructure; Cloud computing; Bag-of-tasks; Workflows
Journal Title
Cluster Computing
Citation Volume
18
ISSN
1386-7857
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 학술지 발표논문
URI
https://repository.kisti.re.kr/handle/10580/14455
http://www.ndsl.kr/ndsl/search/detail/article/articleSearchResultDetail.do?cn=NART74712637
Export
RIS (EndNote)
XLS (Excel)
XML

Browse