Customized dynamic load balancing for a network of workstations
Journal of Parallel and Distributed Computing
A Case for NOW (Networks of Workstations)
IEEE Micro
Efficient Data Access for Parallel BLAST
IPDPS '05 Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Papers - Volume 01
| Hi-index | 0.01 |
Universities around the world are creating their own computing clusters for student use. This gives each student access to more resources than would be possible in traditional lab settings. However, as more university departments begin to take advantage of these resources, more sophisticated techniques for allocating computer cycles and load balancing are necessary. This paper presents a simple, per-application approach to responding to saturation of a computing node. By accurately determining when a cluster node has become overburdened by multiple users or is unable to respond to new users, the customized program is able to change its behavior and politely hand control over to what would otherwise be inconvenienced applications.