Our prototype, Octopus, is an extensible resource manager system for running commercial and scientific parallel applications in the cluster system environment. This system was built to be a platform for experimenting a great variety of resource management issues: resource scheduling algorithms, scheduling structures for distributed system, schemes to support dynamic resource allocation between multiple scheduling paradigms and etc..

It is our intent to port the prototype to the NT environment from the UNIX environment. The porting will provide us a new platform to continue our current experimentation as well as to extend research to new directions. Some of the new directions include: the scalability issues of the scheduling structures across high number of machines of heterogenous capacities/resources, the approaches in integrating application and scheduler fault-tolerance and etc.. The new platform will also be our experimental ground for our new high speed connection technology.

Liana L. Fong
fong@watson.ibm.com (914)784-7269
IBM T. J. Watson Research Center
P.O. Box 704 Yorktown Heights, NY 1059