WS-PGRADE/gUSE-Based Science Gateways in Teaching - Science Gateways for Distributed Computing Infrastructures - page 214

workflows on various distributed computing infrastructures (currently it only
includes service grid systems; however, the extension of the gateway towards clouds
is in progress).
The University of Westminster operates the SHIWA simulation platform (SSP)
(Terstyanszky 2014) as a generic WS-PGRADE gateway that is utilized in the
module. The SSP connects to various grid systems, including Globus, gLite, and
UNICORE-based grids. One of the connected virtual organizations is the Westfocus
virtual organization (VO), which offers access to the University of Westminster
cluster using on the Globus Toolkit version 4 middleware. Students use this cluster
and VO only for workflow execution.
During the module students go through the whole grid experience. The rst step

is applying for a personal grid certi cate. Westminster operates a registration

authority (RA) within the UK Certi cation Authority, making the process easier to

manage and control. Although the application is done via an external tool called the
Certi cate Wizard, further management of the user certi cate, including uploading

it to the NGS MyProxy Server and downloading a certi cate proxy for a session, is

fully supported by the gateway. Once their user certi cate is ready, students create

and execute sample workflows. As the module aims to introduce distributed
computing at user and business application level only, the tutorials show and
illustrate how these infrastructures can be accessed from user-friendly interfaces
without requiring low-level programming skills. The created workflows include use
cases for traf

c simulation (Delaitre 2005) and molecular modeling (Kiss 2010). In
both scenarios students receive the executables, but create the workflows and
address the parallelization (only by parallel workflow branches or parameter
sweeps) themselves. Figure 16.1 illustrates two typical workflows created by stu-
dents. The left one shows a traf

c
simulators with different input parameters run in different branches of the workflow.
The right graph implements random blind docking in a Monte Carlo simulation
using AutoDock and creates a parameter sweep workflow, with the outgoing port
on the AutoGrid job being a generator port.

c simulation workflow where two identical traf

Fig. 16.1 Workflow graphs for traffic simulation and molecular modeling