213 0

An Ontology-based Monitoring Approach for Efficient Server Management in Datacenter

Title
An Ontology-based Monitoring Approach for Efficient Server Management in Datacenter
Author
이정민
Advisor(s)
김정선
Issue Date
2014-02
Publisher
한양대학교
Degree
Master
Abstract
With world-wide development of IT service, the size of datacenters are growing rapidly to provide seamless services such as technical support, hosting, managing, and storing, and the number of equipment and facilities which should be managed in datacenters has been exponentially increased. Some issues such as availability, energy efficiency, heat density, datacenter integration, and power consumption have been being magnified by those equipment and facilities mentioned above. Therefore, to manage datacenters efficiently, there are three important elements to be administered; various equipment in datacenters, energy efficiency and power consumption of equipment, and provision of services to clients. Servers are the most important management objects in datacenters to be administered closely. In case of almost datacenters, however, servers are commonly operated inefficiently with low utilization, as a result, they causes additional power consumption unnecessarily from not only themselves but also other equipment to adjust heat generated by them. Therefore, servers should be monitored to manage datacenters efficiently. However, previous server monitoring approaches have usually the managed server state using databases, and also have monitored state of server individually. As a result an administrator should know structure of the datacenter and associated databases that stores observed server state data measured by sensors, and is mandatory to observe data and to analyze relationships among data stored in the databases. To solve these problems, we proposed an ontology-based monitoring approach. We constructed domain ontology for the server state to manage datacenter, and mapped observed data onto domain ontology using semantic annotation that assigns meaning to raw data and defines relationships among them explicitly. As a result, an administrator is not necessary to understand the structure of databases, and data which is not indicated can be monitored by inferring ontology classes. It makes knowledge and process separated, and the monitoring system can become robust to extend and change databases. Moreover, we defined query rules using SWRL (Semantic Web Rule Language) and server state rules using SQWRL (Semantic Query-enhanced Web Rule Language) to provide convenience to monitor servers easily. To describe the proposed approach, we designed the ontology-based monitoring system architecture, and constructed the knowledge system. Then we implemented the pilot system to conduct the verification based-on situation-specific scenarios.
URI
https://repository.hanyang.ac.kr/handle/20.500.11754/130903http://hanyang.dcollection.net/common/orgView/200000423215
Appears in Collections:
GRADUATE SCHOOL[S](대학원) > COMPUTER SCIENCE & ENGINEERING(컴퓨터공학과) > Theses (Master)
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE