Adaptive Control of Redundant Task Execution for Dependable Volunteer Computing - Cloud, Grid and High Performance Computing: Emerging Applications - page 154

dispatching a task whenever a task enquiry comes,
this dispatch policy waits for several task enquiries
from different workers, and then dispatch tasks
to them at once. It uses a heuristics-based failure
probability estimation method to find an optimized
task-to-worker assignment that minimizes the
overall failure probability of the tasks.
The LFPD policy has been evaluated with real
world trace data sets on a simulator. The evalua-
tion results have been compared with those of two
selected baseline policies. The comparison results
indicate the effectiveness of the LFPD policy.
The results also prove that the LFPD policy can
beat the greedy dispatch policy when the mean
task process time is much smaller than the mean
TTF of the workers. The difference between the
results with two trace data sets is also discussed.
To study how the different type of workers in
the real world may affect the effectiveness of the
LFPD, a trace data set that consists of two types
of workers has been generated from the two real
world trace data sets. The LFPD policy has been
simulated with and without the ability to identify
different type of workers. The results indicate that
worker type identification can provide additional
performance improvement.

Anderson, D. P. (2004). BOINC: A system
for public-resource computing and storage. In
Proceedings of Fifth IEEE/ACM International
Workshop on Grid Computing , (pp. 4-10).

Anderson, D. P., Cobb, J., Korpela, E., Lebof-
sky, M., & Werthimer, D. (2002). Seti@home:
an experiment in public-resource computing.
Communications of the ACM , 45 (11), 56-61.
doi:10.1145/581571.581573

Bolosky, W. J., Douceur, J. R., Ely, D., &
Theimer, M. (2000). Feasibility of a serverless
distributed file system deployed on an exist-
ing set of desktop PCs. ACM SIGMETRICS
Performance Evaluation Review , 28 (1), 34-43.
doi:10.1145/345063.339345

Bolze, R., Cappello, F., Caron, E., Dayd'e, M.,
Desprez, F., & Jeannot, E. (2006). Grid'5000: A
large scale and highly reconfigurable experimental
grid testbed. International Journal of High Perfor-
mance Computing Applications , 20 (4), 481-494.
doi:10.1177/1094342006070078

Brevik, J., Nurmi, D., & Wolski, R. (2004). Au-
tomatic methods for predicting machine avail-
ability in desktop grid and peer-to-peer systems.
In Proceedings of the 2004 IEEE International
Symposium on Cluster Computing and the Grid
(CCGRID04) , (pp. 190-199).

ACKNOWLEDGMENT

Cappello, F., Djilali, S., Fedak, G., Herault, T.,
Magniette, F., & N'eri, V. (2005). Computing
on large-scale distributed systems: Xtrem web
architecture, programming models, security, tests
and convergence with grid. Future Generation
Computer Systems , 21 (3), 417-437. doi:10.1016/j.
future.2004.04.011

This research was partially supported by Grant-
in-Aid for Scientific Research on Priority Areas
#18049003 and Strategic Information and Com-
munications R&D Promotion Program (SCOPE-
S) #061102002.

REFERENCES

Chien, A. A., Calder, B., Elbert, S., & Bhatia, K.
(2003). Entropia: Architecture and performance
of an enterprise desktop grid system. Journal
of Parallel and Distributed Computing , 63 (5),
597-610. doi:10.1016/S0743-7315(03)00006-6

Aldrich, J. (1997). R. A. Fisher and the making of
maximum likelihood 1912-1922. Statistical Sci-
ence , 12 (3), 162-176. doi:10.1214/ss/1030037906