Comparative Benefits of Using the Grid
Back to Previous Page
In the Pre-SAMD scenario, if datasets need intensive analysis that is beyond
the capacity of a desktop PC, researchers had to log in to an Athens-enabled
dataset, download the datasets to their PC several times, and then physically
take these files to their nearest High Performance Computer.
The SAMD model provides an integrated environment to manage this workflow
in such a way that time and effort is saved using the single sign-on feature
to access both the data and the compute resources in the one environment rather
than having to negotiate a potentially confusing technical environment.
The following table compares the pre-SAMD and Grid models
Fig. 1 Shows the user having to access resources individually
Fig. 2 Shows the Single Sign On Grid model
- The diagram above shows that social scientists need to access resources
- Without single sign on giving standard authentication and authorisation,
access to the various datasets and High Performance Computers (HPCs)
frequently requires separate login accounts. There may be several
usernames and passwords to remember in the whole operation.
- The distributed resources appear to the user as entirely separate
resources. For example, the researcher may have to download some of
the data in their office and some data from another PC in a separate
location. This data may need to be combined and then sent off to a
large Unix computer to do the high speed analysis. These operations
require separate passwords and a basic knowledge of UNIX commands.
- There is no integrated front-end to all of these operations which
means that different interfaces need to be learned by the social scientist,
including the UNIX command line.
- Along with the data analysis, this all took the best part of a day.
- The diagram above shows that social scientists can access multiple
resources with a single sign-on.
- In an e-science Grid a single sign on gives you access to all the
databanks to which you have an entitlement. For example, this may
be a databank at MIMAS
to which you have an entitlement through your university, or an individual
registration for the French census at the Institut National de la
Statistique et des Etudes.
- Distributed resources appear as a single resource. This means users
can search across all the databanks wherever they are located in a
single search. The network also includes High Performance Computing
facilities for data analysis and the data can be moved directly from
the databank to the HPC.
- The integrated front-end (the GUI) designed by the SAMD technical
team makes the steps of searching for data and computer resources
- Data collection and analysis time was reduced to less than an hour.