TY - GEN
T1 - UML-based beowulf cluster availability modeling
AU - Song, Hertong
AU - Leangsuksun, Chokchai
AU - Nassar, Raja
AU - Liu, Yudan
AU - Engelmann, Christian
AU - Scott, Stephen L.
PY - 2005
Y1 - 2005
N2 - Cluster Computing is becoming cost effective and popular for its enormous computational power. High availability features need to be included to ensure that cluster computing environments can provide continuous services. In this regard, it is imperative to know the dependability parameters during the conceptual design stages, since these parameters help in facilitating design trade-off and refinement. However, people in system design and development may not be familiar with the analytical modeling techniques. Furthermore, large models are often complicated and out of the logical view of the system. This research strives for an integral analysis and design framework that automatically models system dependability based on Unified Modeling Language (UML) design notations and computes system reliability and availability. A key activity is to investigate UML design semantics and their mapping to availability models such as reliability block diagrams, Markov Chain, Stochastic Petri Net (SPN), etcetera, with minimum end-user requirements to reduce the burden of modeling analysis. The UML-based dependability modeling technique is an enabler to bridge the gap between system architects and developers who are not familiar with dependability modeling formalisms. With our approach, the availability analysis of cluster computing systems can be done with ease.
AB - Cluster Computing is becoming cost effective and popular for its enormous computational power. High availability features need to be included to ensure that cluster computing environments can provide continuous services. In this regard, it is imperative to know the dependability parameters during the conceptual design stages, since these parameters help in facilitating design trade-off and refinement. However, people in system design and development may not be familiar with the analytical modeling techniques. Furthermore, large models are often complicated and out of the logical view of the system. This research strives for an integral analysis and design framework that automatically models system dependability based on Unified Modeling Language (UML) design notations and computes system reliability and availability. A key activity is to investigate UML design semantics and their mapping to availability models such as reliability block diagrams, Markov Chain, Stochastic Petri Net (SPN), etcetera, with minimum end-user requirements to reduce the burden of modeling analysis. The UML-based dependability modeling technique is an enabler to bridge the gap between system architects and developers who are not familiar with dependability modeling formalisms. With our approach, the availability analysis of cluster computing systems can be done with ease.
KW - Availability
KW - Cluster system
KW - Markov Chain Specification
KW - UML (Unified Modeling Language)
UR - http://www.scopus.com/inward/record.url?scp=60749103447&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:60749103447
SN - 9781932415490
T3 - Proceedings of the 2005 International Conference on Software Engineering Research and Practice, SERP'05
SP - 161
EP - 167
BT - Proceedings of the 2005 International Conference on Software Engineering Research and Practice, SERP'05
T2 - 2005 International Conference on Software Engineering Research and Practice, SERP'05
Y2 - 27 June 2005 through 30 June 2005
ER -