Self-adjusting resource sharing policies in Federated Grids

The majority of non-coordinated decentralized meta-schedulers in a Federated Grid perform scheduling strategies without taking into account resources' current load or specific resource owners' internal demands, leading to suboptimal schedules. Clearly, these policies increase the number of job migrations, the number of messages generated per re-scheduled job, and also the application makespan. The main purpose of the present study is to analyze the effect of applying self-adjusting resource sharing policies to previously proposed performance based scheduling strategies. For example, when a resource is near saturation or has an internal peak demand, it can decide not to accept new external jobs. On the other hand, when a job owner receives the previous action, it can decide not to submit temporally more jobs to that resource. In this way, the proposed self-adjusting resource sharing policies save time and communication bandwidth by reducing the number of jobs migrations, and thus, avoiding the generation of the corresponding messages per re-scheduled job. At the same time, the new resource sharing strategies improve application makespan and resource performance objective functions while maintaining infrastructure owners complete autonomy. Highlights? Self-adjusting resource sharing policies save time and communication bandwidth. ? Our policies map jobs proportional to the performance of the resources. ? Our algorithms perform a mapping of jobs adjusted to resources real saturation. ? We don't need to negotiate with the rest of the participating infrastructures. ? Our strategies can reach a good makespan within different saturation level scenarios.

[1]  Ian T. Foster,et al.  Agreement-Based Resource Management , 2005, Proceedings of the IEEE.

[2]  Eduardo Huedo,et al.  A decentralized model for scheduling independent tasks in Federated Grids , 2009, Future Gener. Comput. Syst..

[3]  Chris R. Jesshope,et al.  Parallel Computers 2: Architecture, Programming and Algorithms , 1981 .

[4]  Reid G. Smith,et al.  The Contract Net Protocol: High-Level Communication and Control in a Distributed Problem Solver , 1980, IEEE Transactions on Computers.

[5]  Eduardo Huedo,et al.  Benchmarking of high throughput computing applications on Grids , 2006, Parallel Comput..

[6]  Zoltan Juhasz,et al.  Scalability Analysis of the Contract Net Protocol , 2002, 2nd IEEE/ACM International Symposium on Cluster Computing and the Grid (CCGRID'02).

[7]  Li Zhang,et al.  Tycoon: An implementation of a distributed, market-based resource allocation system , 2004, Multiagent Grid Syst..

[8]  Eduardo Huedo,et al.  A framework for adaptive execution in grids , 2004, Softw. Pract. Exp..

[9]  Rajkumar Buyya,et al.  Model-based simulation and performance evaluation of grid scheduling strategies , 2009, Future Gener. Comput. Syst..

[10]  Andrew M. Walker,et al.  Job submission to grid computing environments , 2008 .

[11]  Jonathan M. Garibaldi,et al.  A Multi-agent Infrastructure and a Service Level Agreement Negotiation Protocol for Robust Scheduling in Grid Computing , 2005, EGC.

[12]  Rajkumar Buyya,et al.  Decentralised Resource Discovery Service for Large Scale Federated Grids , 2007, Third IEEE International Conference on e-Science and Grid Computing (e-Science 2007).

[13]  Jennifer M. Schopf,et al.  A performance study of monitoring and information services for distributed systems , 2003, High Performance Distributed Computing, 2003. Proceedings. 12th IEEE International Symposium on.

[14]  Selim G. Akl,et al.  Scheduling Algorithms for Grid Computing: State of the Art and Open Problems , 2006 .

[15]  Eduardo Huedo,et al.  Performance-based scheduling strategies for HTC applications in complex federated grids , 2010, Grid 2010.

[16]  Eduardo Huedo,et al.  A Performance Model for Federated Grid Infrastructures , 2008, 16th Euromicro Conference on Parallel, Distributed and Network-Based Processing (PDP 2008).

[17]  Rajkumar Buyya,et al.  SLA-Based Coordinated Superscheduling Scheme for Computational Grids , 2006, 2006 IEEE International Conference on Cluster Computing.