Quality of service for wide area clusters
Proceedings of the 8th ACM SIGOPS European workshop on Support for composing distributed applications
EURO-PDP'00 Proceedings of the 8th Euromicro conference on Parallel and distributed processing
Hi-index | 0.00 |
Sharing hardware and software resources among networked workstations has become common practice. Resource management in such systems is complex because of physical distribution of the resources, communication delays and the lack of global state information. This paper describes a scalable load monitoring service and a scalable resource query service developed in the Warp project to provide necessary state information for resource management. Scalability is achieved by reducing network traffic. Techniques include exploiting IP multicast for one-to-many communications, using supply-driven and demand-driven approaches for disseminating different classes of information, compressing information, and tailoring servers only to publish information when certain conditions are met. GUI tools, statistical analysers, and job scheduling are example clients of the services.