Title
Task distribution with a random overlay network
Abbreviated Journal Title
Futur. Gener. Comp. Syst.
Keywords
grid computing; commodity computers; routing; GRIDS; Computer Science, Theory & Methods
Abstract
We consider a model where commodity service providers are offering commodity computational services to a set of customers. We provide a solution for the efficient distribution of tasks by forwarding the service requests on an overlay network comprised on random cycles. We introduce algorithms for the creation, maintenance and repair of the overlay network. We discuss two algorithms, random wandering and weighted stochastic forwarding, for the allocation of the tasks to providers. Both approaches are highly scalable because the algorithms use only limited local information. As we are designing our approach for use in a commercial setting, there is a requirement that the tasks, being a source of profits, be allocated fairly to the providers. We investigate the fairness of the algorithms and show that adding a random pre-walk can improve the fairness. Through a simulation study we show that the approach provides efficient task allocation on networks loaded up to 95% of their capacity. Published by Elsevier B.V.
Journal Title
Future Generation Computer Systems-the International Journal of Grid Computing Theory Methods and Applications
Volume
22
Issue/Number
6
Publication Date
1-1-2006
Document Type
Article
Language
English
First Page
676
Last Page
687
WOS Identifier
ISSN
0167-739X
Recommended Citation
"Task distribution with a random overlay network" (2006). Faculty Bibliography 2000s. 5969.
https://stars.library.ucf.edu/facultybib2000/5969
Comments
Authors: contact us about adding a copy of your work at STARS@ucf.edu