Visual Grid Application Development Environments Parametric and Parallel











- Slides: 11

Visual Grid Application Development Environments: Parametric and Parallel Models for Creating Distributed Applications Shoaib Burq, Steve Melnikoff, Rajkumar Buyya Grid computing & Distributed Systems (GRIDS) Lab. The University of Melbourne, Australia www. gridbus. org/ GRIDS

Agenda n n n 2 Motivation Architecture and Java Based Implementation Case Study: Molecular Docking Demo Conclusions and Future Work

Visual Parametric Tool (VPT) n n n 3 Rapid parameterisation for application scientists Creation & maintenance of parameter sweep applications/projects Creation of Nimrod-G run-files Platform independent (Java) Open source

Parametric Sweep Applications n Distribution Model n n n Data and Information Model n n 4 Parameters form basis for parallelism Parameter Studies Declarative language Parameterisation of input files/scripts

Architecture of VPT Data File 2. File independent Parameterization GRIDS Paramete r object GRIDS Paramete GRIDS r object Paramete r object (Generate Plan File) 5 Import Data files Project object 1. File specific Parameterization GRIDS Paramete GRIDS r object Paramete r object Visual Parameterization Tool Plan File Run File parameter X integer default 1 parameter Y integer range 1 to 10 …. variable X constant 1 variable Y index 0 list 1…. Jobs … (Generate Run File) Distributed Data Scheduler

Case Study and Demo n n n 6 Parameterize a molecular docking input file Generate Nimrod-readable runfile > java -cp VPTClasses. jar. Grid. Bus 2. GBProject

Drug Design: Data Intensive Computing on Grid Molecules Protein n 7 Chemical Databases (legacy, in. MOL 2 format) It involves screening millions of chemical compounds (molecules) in the Chemical Data. Base (CDB) to identify those having potential to serve as drug candidates. [Collaboration with WEHI for Medical Science, Melbourne]

Molecular Docking Application 8

Docking Task Definion 9

Conclusion and Future Developments n n 10 It works! To support creation of abstract parameters (e. g. , logical files for HEP community) Support other distributed application construction model Integration with some popular cluster and Grid job management systems

That’s all folks… n 11 Any Question?