DHPC Technical Report DHPC-101
Reconfigurable Scheduling of Long Running Objects
K.A.Hawick and H.A.James
Archived: 22 May 2001
Published in Proc. Int. Conf. on Parallel and Distributed Processing
Techniques and Applications (PDPTA 2001), Las Vegas, USA, June 2001.
Abstract
We employ the new term reconfigurable scheduling to describe the
capability of check-pointing or suspending an already scheduled and
running task and its subsequent restart. It is a particularly useful
feature of a scheduling system in a long running middleware
environment that long running objects can be suspended and later resumed
later without having to be completely restarted with an associated
waste of compute cycles. We have experimented with the serialisation
and multi threading capabilities of the Java runtime environment in
support of reconfigurable scheduling. We describe a model for
computing the savings in efficiency for scheduling long running tasks
under such an environment. We also discuss the potential for real
time scheduling of tasks in a middleware environment when
reconfigurable scheduling is available.
Keywords: reconfiguration; long-running; distributed objects;
parallel computing.
PDF version
Postscript version (gzip compressed)
[
DHPC Adelaide |
DHPC Bangor |
Contacts |
People |
Projects |
Reports
]
webmaster@dhpc.adelaide.edu.au