Re: よくわかんないけど
投稿者： melancholy_night 投稿日時： 2006/03/01 20:28 投稿番号： [28448 / 38959]
論文見つけました。ネットワーク技術分野のまともな研究者らしいです。
Low-Cost Fault-Tolerance Protocol for Large-Scale Network Monitoring
JinHo Ahn A1, SungGi Min , YoungIl Choi A3, ByungSun Lee A3
A1 Dept. of Computer Science College of Information Science Kyonggi University
San 94-6 Yiuidong Paldalgu, Suwonsi Kyonggido 442-760, Republic of Korea
jhahn91@hanmail.net
A2 Dept. of Computer Science & Engineering Korea University 5-1 Anamdong
Sungbukgu, Seoul 136-701, Republic of Korea sgmin@korea.ac.kr
A3 Network Technology Lab. Electronics and Telecommunications Research
Institute Yusong P.O. Box 106 Taejon 305-600, Republic of Korea
{yichoi,bslee}@etri.re.kr
Abstract:
Distributed hierarchical network monitoring model has been proposed to solve
scalability problem of centralized model. In this distributed model, a top-level
monitoring manager, called main manager, obtains aggregate management
information from mid-level managers, named domain managers, forming a
hierarchical structure. However, if some of monitoring managers crash, network
elements cannot be continuously and correctly monitored until the managers are
repaired. To address this important, but previously unresolved issue, this paper
presents a new fault-tolerance protocol for domain managers, named $DMFTP$,
allowing the managers to efficiently utilize their organization structure. Therefore,
this protocol can minimize failure detection overhead and the number of live
managers affected by each manager node crash. Also, it tolerates concurrent
manager failures and, after the failed managers have been repaired, ensures
their immediate and consistent recovery.
http://springerlink.metapress.com/(tojolt45lvszkh45jjbzxv55)/app/home/contribution.asp?referrer=parent&backto=issue,50,113;journal,1213,331 3;linkingpublicationresults,1:105633,1
Low-Cost Fault-Tolerance Protocol for Large-Scale Network Monitoring
JinHo Ahn A1, SungGi Min , YoungIl Choi A3, ByungSun Lee A3
A1 Dept. of Computer Science College of Information Science Kyonggi University
San 94-6 Yiuidong Paldalgu, Suwonsi Kyonggido 442-760, Republic of Korea
jhahn91@hanmail.net
A2 Dept. of Computer Science & Engineering Korea University 5-1 Anamdong
Sungbukgu, Seoul 136-701, Republic of Korea sgmin@korea.ac.kr
A3 Network Technology Lab. Electronics and Telecommunications Research
Institute Yusong P.O. Box 106 Taejon 305-600, Republic of Korea
{yichoi,bslee}@etri.re.kr
Abstract:
Distributed hierarchical network monitoring model has been proposed to solve
scalability problem of centralized model. In this distributed model, a top-level
monitoring manager, called main manager, obtains aggregate management
information from mid-level managers, named domain managers, forming a
hierarchical structure. However, if some of monitoring managers crash, network
elements cannot be continuously and correctly monitored until the managers are
repaired. To address this important, but previously unresolved issue, this paper
presents a new fault-tolerance protocol for domain managers, named $DMFTP$,
allowing the managers to efficiently utilize their organization structure. Therefore,
this protocol can minimize failure detection overhead and the number of live
managers affected by each manager node crash. Also, it tolerates concurrent
manager failures and, after the failed managers have been repaired, ensures
their immediate and consistent recovery.
http://springerlink.metapress.com/(tojolt45lvszkh45jjbzxv55)/app/home/contribution.asp?referrer=parent&backto=issue,50,113;journal,1213,331 3;linkingpublicationresults,1:105633,1
これは メッセージ 28444 (melancholy_night さん)への返信です．