Sharing Science Gateway Artefacts Through Repositories - Science Gateways for Distributed Computing Infrastructures - page 122

Fig. 9.2 gUSE workflow storage

Usage scenarios : We distinguish between development and execution scenarios.
In the first case, as shown in Fig. 9.2 these storage entities have different roles in the

three phases of the workflow development: graph creation, concrete workflow
con guration, and workflow exportation. Creating the abstract graph of the work-

flow using the graph editor de nes the workflow skeleton by adding its nodes and

the relations among them. This step does not indicate any

file transfer. Therefore,
this step uses the Workflow Storage only, which stores the given information in the
database. In the workflow con guration phase users can add arguments to specify

the resources to be used or the types of workflow nodes and so on. In this phase
input files and executables can be also uploaded. These operations invoke the File

Storage to transfer these

files into the proper folders. Information storage in the

workflow con guration step is supported by both the File Storage and the workflow

storage services. Finally, the export process requires cooperation of all three storage
components. First of all, the workflow configuration and its data must be collected
from the

file storage and the workflow storage. Next this data is compressed and is
exported to the Application Repository as a single zip file.

In the execution scenario the Application Repository supports sharing workflows
among gateway users. They import the workflows from the Application Repository
in order to parameterize and execute them.

9.3.2 SHIWA Work fl ow Repository

The “ Sharing Interoperable Workflows for Large-Scale Scienti c Simulations on

Available DCIs (SHIWA) project (SHIWA 2014) developed and deployed the

SHIWA Workflow Repository. The repository allows users to manage workflows

”