![](bg8.png)
enable neuroscientists to exploit the benefits of
cloud computing. Users and organizations no longer
need to be concerned about hardware requirements
or storage capacities as the cloud provides scalability
on demand. Also, our implementation shows that it
is feasible to share data securely between users
belonging to different organizations.
However, while implementing our solution, we
faced compatibility issues with the GW SDK and the
MATLAB environment, and we had to adapt the
.NET API according to the SPM settings. We were
also interested in integrating the Cloud Data
Management Interface (CDMI) implementation
(Livenson and Laure, 2011) for transferring our data
to the cloud, but it was not feasible with our solution
since the CDMI implementation on Azure storage
services missed some features and the .NET CDMI
client library was not as functional as the Java
version. Therefore we had to opt for a solution that
transferred our granular files as a single unit to the
cloud storage.
As we mentioned earlier in section 2.1, each
SPM job is composed of several steps and in real life
it would be interesting to measure the overheads
associated with using Microsoft Azure for deploying
the SPM. For instance, a thorough performance
analysis could be carried out to compare the stage-
in/stage-out time for data, the length of time that
jobs stay in the queue before being run, and other
metrics such as the execution or completion time for
jobs. Also, it could be useful to run real SPM jobs in
different VMs with different memory capacities and
CPU cores to discover the limitations that our
prototype might pose in real life. The possibility of
designing parallel jobs through breaking the SPM
jobs into smaller tasks that can be run in parallel also
can be interesting to achieve better performance.
Finally, to empower users with browsing and data
management capabilities of the storage,
implementing a MATLAB GUI would be useful.
This feature will enable users to see contents of the
containers that belong to different research groups
for data sharing purposes.
ACKNOWLEDGEMENTS
We gratefully acknowledge the help of those who
organized, reviewed and provided motivation for
this paper, especially: Fredrik Ullén, Lars Forsberg,
Rita Almedia from Karolinska Institute and Åke
Edlund, Ilja Livenson from the Royal Institute of
Technology. Special thanks also to Genet
Edmondson for language comments. This research
was sponsored by the EU within the 7FP project
“VENUS-C” under grant agreement number
261565.
REFERENCES
Beno. Retrieved April 7, 2011, from http://phiwave.source
forge.net/howto_parallel/#Parallel_SPM_batch_script
ing_on.
Dean, J. and Ghemawat, S. (2004). MapReduce:
Simplified Data Processing on Large Clusters, Sixth
Symposium on Operating System Design and
Implementation, San Francisco, CA.
Djordjevic, I. and Dimitrakos, T. (2005). A Note On the
Anatomy of Federation, BT Technology Journal,
Volume 23, Issue 4.
Generic Worker Complete Documentation. Retrieved
June 4, 2012, from http://resources.venus-c.eu.
Foster, I. et al., (2007). OGSA Basic Execution Service,
Version 1.0, GFD-RP-R-P.108.
Hwang, K., Fox, G., and Dongarra, J. (2011). Distributed
and Cloud Computing: From Parallel Processing to
the Internet of Things, Morgan Kaufmann Publishers.
Livenson, I. and Laure, E. (2011). Towards Transparent
Integration of Heterogeneous Cloud Storage
Platforms, Proceedings of the Fourth International
Workshop on Data-Intensive Distributed Computing.
MCR (MATLAB Runtime Compiler). Retrieved October
24, 2011, from http://www.mathworks.com/products/
compiler.
Microsoft Windows Azure. Retrieved July 24, 2012, from
http://www.microsoft.com/windowsazure.
MPI (Message Passing Interface). Retrieved April 7, 2011,
from http://www.mcs.anl.gov/research/projects/mpi/.
PSPM (Parallelized SPM). Retrieved April 7, 2011, from
http://prdownloads.sourceforge.net/parallelspm/.
Savva, A. (Editor), (2005). Job Submission Description
Language (JSDL) Specification. Version 1.0.
SPM (Statistical Parametric Mapping). Retrieved April 7,
2011, from http://www.fil.ion.ucl.ac.uk/spm/.
VENUS-C Deliverable D6.1, (2011). Report on
Architecture, http://www.venus-c.eu.
VENUS-C FP7 Project, (2010). Grant Agreement No.
261565, http://www.venus-c.eu.
CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience
336