SAS Grid Computing at
TTU
Through partnerships with SAS Institute, the Center for Advanced Analytics and Business
Intelligence at TTU, and the High
Performance Computing Center at TTU,
What is SAS?
SAS is a large-scale,
integrated system for data management, scientific computation, optimization,
and information delivery. Its
functionality overlaps and largely subsumes functionalities of most
commonly-used mathematical, statistical, applications development, database,
and graphical software packages.
What can the SAS grid be used for?
Any scientific computing
project that involves repetitive, essentially independent time-consuming tasks
is an ideal candidate for grid computing.
Examples include
- Simulation of stochastic systems
- Evaluation deterministic systems for many
different starting conditions
- Crunching data that can be partitioned (eg, the Search for Extraterrestrial Intelligence, or
SETI)
How does the SAS grid work?
See the following diagram:
What projects has the SAS grid been
used for at TTU?
Clinical trials
simulation
TTU
researchers Westfall and Tomoiaga have a grant with a Pharmaceutical company to
develop a SAS grid for clinical trials simulation. Clinical studies can cost $10,000,000+, yet
they may fail, at huge cost to the company, or they may be successful but
wasteful. To reduce failures and costs,
clinical studies are done “in silico,”
and optimized for study design and analysis.
This requires massive simulations.
SAS is used for data generation, complex statistical analysis, graphical
report generation, all with a graphical user interface at the client side, with
an underlying grid to speed computations from days to minutes.
Evaluating significance
of DNA pattern searching
TTU
researchers
Research support for
evaluating operating characteristics of financial event study tests:
- TTU researchers Hein
and Westfall (2004) developed a bootstrap method for event study
testing. The SAS grid used to
evaluate their methodology reduced compute time from a week to an
hour.
- TTU researchers Bremer and English evaluated the
performance of many event study procedures by sampling from CRISP
tapes. The SAS grid reduced compute
time from over a year to two weeks.
Research on the SAS grid
itself
By
a consortium of TTU researchers: Bremer,
R., Perez, J., Smith, P. and Westfall, P., (2004). Grid
Computing at Texas Tech University using SAS. Proceedings
of the14th Annual South- Central SAS
User’s Group Regional Conference, 64-72.
Details on the SAS grid, from SAS
Institute documentation
Automation of
enterprise grid computing capabilities
The platform
suite for grid management
White papers:
Grid
computing using SAS
Contact for further details
Dr. Peter Westfall,
Director, Center for Advanced Analytics and
Business Intelligence, peter.westfall@ttu.edu