On choosing a task assignment policy for a distributed server system
Journal of Parallel and Distributed Computing - Special issue on software support for distributed computing
The EASY - LoadLeveler API Project
IPPS '96 Proceedings of the Workshop on Job Scheduling Strategies for Parallel Processing
Using Queue Time Predictions for Processor Allocation
IPPS '97 Proceedings of the Job Scheduling Strategies for Parallel Processing
Practical Heterogeneous Placeholder Scheduling in Overlay Metacomputers: Early Experiences
JSSPP '02 Revised Papers from the 8th International Workshop on Job Scheduling Strategies for Parallel Processing
The Impact of More Accurate Requested Runtimes on Production Job Scheduling Performance
JSSPP '02 Revised Papers from the 8th International Workshop on Job Scheduling Strategies for Parallel Processing
Scheduling From the Perspective of the Application
HPDC '96 Proceedings of the 5th IEEE International Symposium on High Performance Distributed Computing
Benefits of Global Grid Computing for Job Scheduling
GRID '04 Proceedings of the 5th IEEE/ACM International Workshop on Grid Computing
Improving a Local Learning Technique for QueueWait Time Predictions
CCGRID '06 Proceedings of the Sixth IEEE International Symposium on Cluster Computing and the Grid
Backfilling Using System-Generated Predictions Rather than User Runtime Estimates
IEEE Transactions on Parallel and Distributed Systems
Modeling the Impact of Resource Sharing in Backfilling Policies using the Alvio Simulator
MASCOTS '07 Proceedings of the 2007 15th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems
A job self-scheduling policy for HPC infrastructures
JSSPP'07 Proceedings of the 13th international conference on Job scheduling strategies for parallel processing
Hi-index | 0.00 |
Several centralized scheduling solutions have been proposed in the literature for environments composed of several independent computational resources, such as centralized schedulers, centralized queues and global controllers. These approaches use a unique scheduling entity responsible for scheduling all jobs submitted by users. In our previous work we proposed the use of self-scheduling techniques to dispatch jobs which are submitted to a set of distributed computational hosts, which are in turn managed by independent schedulers (such as MOAB or LoadLeveler). In the ISIS-Dispatcher, scheduling decisions are made independently for each job instead of using a global policy where all jobs are considered. In this paper we present how the ISIS-Dispatcher techniques can be used in the XtreemOS architecture for manage the jobs. This system is designed to be deployed in large scenarios that potentially involve thousands of resources. In such systems it is not feasible to make the dispatcher contact to all the systems. It is not realistic to suppose that the dispatcher stores the information about all the resources and where they are located. Obviously, this approach would imply problems of scalability. In this paper we also evaluate the impact about the amount of resource information that the dispatcher can collect during the job scheduling.