TY - GEN
T1 - Performance evaluation of scheduling mechanism with checkpoint sharing and task duplication in P2P-based PC grid computing
AU - Gil, Joon Min
AU - Song, Ui Sung
AU - Yu, Heon Chang
PY - 2009
Y1 - 2009
N2 - An important issue in the PC grid computing environment that is characterized by volatility and heterogeneity is the minimization of execution time for all tasks. This paper proposes a scheduling mechanism to reduce such execution time by means of both checkpoint sharing and task duplication under a peer-to-peer (P2P) architecture. In the mechanism, the checkpoint executed by an individual peer (i.e., a desktop PC) is used as an intermediate result and executed in other peers via its duplication and transmission. As a result, as the checkpoint is close to a final result, the reduction of execution time for each task becomes higher. Ultimately, turnaround time can be reduced. Moreover, an analytical model with an embedded Markov chain is presented to evaluate the transmission cost and execution time of our scheduling mechanism. The performance of our scheduling mechanism is also compared with that of the existing mechanism operating on client-server architecture. The analytical results show that our scheduling mechanism is superior to the existing mechanism with respect to the reduction of execution time.
AB - An important issue in the PC grid computing environment that is characterized by volatility and heterogeneity is the minimization of execution time for all tasks. This paper proposes a scheduling mechanism to reduce such execution time by means of both checkpoint sharing and task duplication under a peer-to-peer (P2P) architecture. In the mechanism, the checkpoint executed by an individual peer (i.e., a desktop PC) is used as an intermediate result and executed in other peers via its duplication and transmission. As a result, as the checkpoint is close to a final result, the reduction of execution time for each task becomes higher. Ultimately, turnaround time can be reduced. Moreover, an analytical model with an embedded Markov chain is presented to evaluate the transmission cost and execution time of our scheduling mechanism. The performance of our scheduling mechanism is also compared with that of the existing mechanism operating on client-server architecture. The analytical results show that our scheduling mechanism is superior to the existing mechanism with respect to the reduction of execution time.
KW - Checkpoint sharing
KW - Embedded markov chain
KW - P2P-based PC grid computing
KW - Task duplication
UR - http://www.scopus.com/inward/record.url?scp=67650134557&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=67650134557&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-01671-4_41
DO - 10.1007/978-3-642-01671-4_41
M3 - Conference contribution
AN - SCOPUS:67650134557
SN - 9783642016707
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 459
EP - 470
BT - Advances in Grid and Pervasive Computing - 4th International Conference, GPC 2009, Proceedings
T2 - 4th International Conference on Grid and Pervasive Computing, GPC 2009
Y2 - 4 May 2009 through 8 May 2009
ER -