jump to navigation

CLOUDS Lab to Release GridSim Toolkit 5.2 November 27, 2010

Posted by Rajkumar Buyya in CLOUDS Lab News, Grid Computing.
add a comment

CLOUDS Lab to Release GridSim Toolkit 5.2
November 2010

The Cloud Computing and Distributed Systems (CLOUDS) Laboratory at
the University of Melbourne, Australia has released the next-version of
Grid simulation software, the GridSim Toolkit 5.2.

This new version of GridSim brings a new package (i.e. gridsim.fta) that
provides:

* A failure generator based on the Failure Trace Archive (FTA) format.
The provided packages read the tabbed version of the failure traces and generate
a list of events that show the availability/unavailability pattern for each machine
in a grid resource. For more details, see the FTA web site (http://fta.inria.fr) and
the following publication:

Derrick Kondo, Bahman Javadi, Alex Iosup, Dick Epema, The Failure Trace Archive:
Enabling Comparative Analysis of Failures in Diverse Distributed Systems,
In the 10th IEEE/ACM CCGrid, Melbourne, Australia, May, 2010.

* one allocation policy for scheduling of bag of tasks on cluster systems in the
presence of resource failures.

* An implementation of the parallel jobs workload model of DAS-2 for modelling
the workload of clusters. Details about the workload model are available in the
following publication:

Hui Li, David Groep, and Lex Wolters, “Workload Characteristics of a
Multi-cluster Supercomputer”, LNCS 2005, Volume 3277/2005, 176-193.

The version 5.0 of GridSim brings a new package (i.e. gridsim.parallel) that
provides:

* A graphical user interface for debugging resource allocation policies.
* Several allocation policies for scheduling parallel jobs on clusters and
and supercomputers. The policies include aggressive backfilling, conservative
backfilling, advance reservation with conservative backfilling, selective
backfilling and agressive backfilling with multiple resource partitions. The
allocation policies implemented are based on the following research papers:

Ahuva W. Mu’alem and Dror G. Feitelson, Utilization, Predictability,
Workloads, and User Runtime Estimates in Scheduling the IBM SP2
with Backfilling. IEEE Transactions on Parallel and Distributed
Systems, 12:(6), pp. 529-543, 2001.

Barry G. Lawson and Evgenia Smirni, Multiple-Queue Backfilling
Scheduling with Priorities and Reservations for Parallel Systems,
2002 Workshop on Job Scheduling Strategies for Parallel
Processing (JSSPP), pp. 72-87, 2002.

Srividya Srinivasan, Rajkumar Kettimuthu, Vijay Subramani, and
Ponnuswamy Sadayappan. Selective Reservation Strategies for Backfill
Job Scheduling. In JSSPP 2002, LNCS 2537, Springer-Verlag Berlin Heidelberg,
pp. 55-71, 2002.

* An implementation of the workload model by Lublin and Feitelson for modelling
the workload of clusters and supercomputers. Details about the workload model
are available in the following publication:

Uri Lublin and Dror G. Feitelson, The Workload on Parallel Supercomputers:
Modeling the Characteristics of Rigid Jobs. J. Parallel & Distributed
Comput. 63(11), pp. 1105-1122, Nov 2003.

* A data structure to facilitate the scheduling of jobs and advance reservations.
This data structure is described as an appendix on Marcos Assuncao’s PhD thesis.

* Logging features using Java logging API.

In addition, other improvements have made GridSim more compatible with
Java 5 and 6. Some bugs have been fixed and some changes in SimJava were
required to enable a simulation to be paused and resumed.

All components developed as part of the GridSim Toolkit are released as
“open source” under the GPL license to encourage innovation and pass full
freedom to our users.

The early version of our GridSim toolkit has been used/downloaded by several
academic and commercial organizations around the world including:
University of Southern California (USA), California Institute of Technology (USA),
Argonne National Labs (USA), University of Manchester (UK), CERN,
Universidad de Santiago de Compostela (Spain), Indian Institute of Technology,
Tsinghua University (China), Sun Microsystems, IBM Research, Unisys, HP,
Northrop Grumman Information Technology, British Telecom and EMC Corp.

The contributors to the GridSim software (from early to new versions) are:
* Rajkumar Buyya, CLOUDS Lab @ The University of Melbourne.
* Manzur Murshed, GSCIT @ Monash University, Australia.
* Anthony Sulistio, CLOUDS Lab @ The University of Melbourne.
* Gokul Poduval and Chen-Khong Tham,
Dept. of Electrical & Computer Engineering @ National University of Singapore.
* Marcos Dias de Assuncao, INRIA Lyon @ Ecole Normale Superieure de Lyon
* Uros Cibej and Borut Robic, Faculty of Computer and Information Service,
The University of Ljubljana, Slovenia.
* Agustin Caminero, Department of Computing Systems,
Universidad de Castilla La Mancha (UCLM), Spain.
* James Broberg, CLOUDS Lab @ The University of Melbourne.
* Saurabh Garg, CLOUDS Lab @ The University of Melbourne.
* Bahman Javadi, CLOUDS Lab @ The University of Melbourne.

To download the GridSim software, please visit the CLOUDS Lab web site at
http://www.cloudbus.org/gridsim/

Join the GridSim mailing lists at
http://sourceforge.net/projects/gridsim