How Useful Is Old Information?
IEEE Transactions on Parallel and Distributed Systems
The Power of Two Choices in Randomized Load Balancing
IEEE Transactions on Parallel and Distributed Systems
A taxonomy and survey of grid resource management systems for distributed computing
Software—Practice & Experience
Scheduling Resources in Multi-User, Heterogeneous, Computing Environments with SmartNet
HCW '98 Proceedings of the Seventh Heterogeneous Computing Workshop
Dynamic Matching and Scheduling of a Class of Independent Tasks onto Heterogeneous Computing Systems
HCW '99 Proceedings of the Eighth Heterogeneous Computing Workshop
Heuristics for Scheduling Parameter Sweep Applications in Grid Environments
HCW '00 Proceedings of the 9th Heterogeneous Computing Workshop
Beowulf Cluster Computing with Linux
Beowulf Cluster Computing with Linux
Dynamic Server Allocation for Queueing Networks with Flexible Servers
Operations Research
Optimal Routing In Output-Queued Flexible Server Systems
Probability in the Engineering and Informational Sciences
The Anatomy of the Grid: Enabling Scalable Virtual Organizations
International Journal of High Performance Computing Applications
Dynamically mapping tasks with priorities and multiple deadlines in a heterogeneous environment
Journal of Parallel and Distributed Computing
Fault-aware scheduling for Bag-of-Tasks applications on Desktop Grids
GRID '06 Proceedings of the 7th IEEE/ACM International Conference on Grid Computing
Dynamic scheduling for heterogeneous Desktop Grids
GRID '08 Proceedings of the 2008 9th IEEE/ACM International Conference on Grid Computing
| Hi-index | 0.00 |
This paper deals with designing effective resource management strategies for systems of heterogeneous computers. Each computer is represented as an abstract server, capable of serving different task demands at different rates. We consider a system with I types of independent Poisson task demand arrival streams and J parallel servers with independent non-identical processing time distributions for each arrival type. The decision of routing each type i task immediately upon arrival to a server j is made by comparing the state information of a subset of the J servers. We show that choosing the subset according to a linear programming (LP) problem which maximizes the system capacity can not only significantly reduce the amount of state information required in making the routing decision, but also yield shorter total mean queue length (and hence mean time in system) compared with the policies requiring global state information. In addition, we explore means of limiting flexibility to further reduce the required state information.