Session
Computer Science and Information Systems
Description
ITU-PRP provides a Parallel Programming Framework for Java Developers on which they can adapt their sequential application code to operate on a distributed multi-host parallel environment. Developers would implement parallel models, such as Loop Parallelism, Divide and Conquer, Master-Slave and Fork-Join by the help of an API Library provided under framework. Produced parallel applications would be submitted to a middleware called Parallel Running Platform (PRP), on which parallel resources for parallel processing are being organized and performed. The middleware creates Task Plans (TP) according to application’s parallel model, assigns best available resource Hosts, in order to perform fast parallel processing. Task Plans will be created dynamically in real time according to resources actual utilization status or availability, instead of predefined/preconfigured task plans. ITU-PRP achieves better efficiency on parallel processing over big data sets and distributes divided base data to multiple hosts to be operated by Coarse-Grained parallelism. According to this model distributed parallel tasks would operate independently with minimal interaction until processing ends.
Keywords:
Parallel computing, distributed computing, java, ITU-PRP
Session Chair
Kozeta Sevrani
Session Co-Chair
Galia Marinova
Proceedings Editor
Edmond Hajrizi
ISBN
978-9951-437-31-8
First Page
42
Last Page
54
Location
Durres, Albania
Start Date
8-11-2014 9:45 AM
End Date
8-11-2014 10:00 AM
DOI
10.33107/ubt-ic.2014.57
Recommended Citation
Spahi, Enis and Altilar, D. Turgay, "ITU-PRP: Parallel and Distributed Computing Middleware for Java Developers" (2014). UBT International Conference. 57.
https://knowledgecenter.ubt-uni.net/conference/2014/all-events/57
Included in
ITU-PRP: Parallel and Distributed Computing Middleware for Java Developers
Durres, Albania
ITU-PRP provides a Parallel Programming Framework for Java Developers on which they can adapt their sequential application code to operate on a distributed multi-host parallel environment. Developers would implement parallel models, such as Loop Parallelism, Divide and Conquer, Master-Slave and Fork-Join by the help of an API Library provided under framework. Produced parallel applications would be submitted to a middleware called Parallel Running Platform (PRP), on which parallel resources for parallel processing are being organized and performed. The middleware creates Task Plans (TP) according to application’s parallel model, assigns best available resource Hosts, in order to perform fast parallel processing. Task Plans will be created dynamically in real time according to resources actual utilization status or availability, instead of predefined/preconfigured task plans. ITU-PRP achieves better efficiency on parallel processing over big data sets and distributes divided base data to multiple hosts to be operated by Coarse-Grained parallelism. According to this model distributed parallel tasks would operate independently with minimal interaction until processing ends.