Building a Consolidated Data Center
A research institution with 30,000 students, 10,000 employees, a medical center
and 10 schools, Vanderbilt University generates and manages massive amounts
of data. When our data storage needs expanded in the late 1990s at a rate of
1,000 percent annually, the university’s IT team quickly realized it must
take definitive action.
The university was beginning to face the cost containment and technology management
issues associated with maintaining several decentralized databases spanning
university and medical center operations. While most universities that run teaching
hospitals maintain separate IT systems, Vanderbilt found economy and efficiency
in sharing systems while functioning as one organization across academic and
clinical programs. To maintain our ability to meet the university’s IT
needs, improve access, and lower costs, the IT team determined it must consolidate
Vanderbilt’s vital data and storage infrastructure.
Integration was also a priority. Our legacy systems were unable to transform
and organize data into the business intelligence the university needed. For
example, Vanderbilt’s management could not effectively analyze and manage
the costs of services provided by the university’s hospital, which accounted
for 70 percent of the medical center’s expenses. Without the means to
make data available through Web browsers, Vanderbilt's users also lacked easy
access to vital information.
Deployment in the Next Five Years
Vanderbilt University addressed its data storage, business intelligence, and
budget needs with Linux and Oracle technology. The implementation, launched
in 2003, consists of a three-node data warehouse built on Oracle Real Application
Clusters on Red Hat Linux teamed with Oracle Application Server. We use one
three-server cluster for the production data warehouse and a two-node cluster
for testing. If any server in a cluster should fail, the remaining servers continue
to operate seamlessly, ensuring high availability.
Vanderbilt projects that the implementation will yield a savings of 185 percent
during the next five years, taking into account the performance and availability
of the new systems, and the efficiency gained in managing a single vendor’s
system.
The new data warehouse system will eventually serve nearly all decision support
needs for the university and its medical center. Currently, an elaborate labor-tracking
application provides labor decision support management. System users can now
easily extract and format vital data into reports for informed decision-making.
Additional warehouse data includes alumni and fund-raising records, financial
information, and academic and student records.
Modeled for Success
In 2002, the IT team created a five-year cost model and demonstrated, based
on a five-year technology history, that trying to limit costs by buying incrementally
would cause huge spikes in capital investments every few years. We would also
lose productivity in trying to catch up. This reasoning resonated with senior
administrators and won support for the project.
We chose the Linux operating system to reap the benefits of its low costs
and scalability. The team determined that we could utilize three Linux servers
for the price of one UNIX server. We also realized that achieving the required
performance and availability from the university’s existing UNIX platform
would be significantly more expensive than it would be in the Linux environment.
Vanderbilt also chose Oracle because the university’s studies proved
the technology could provide savings up front in hardware costs and deliver
additional economies over time.
New Application Server Improves Scalability, Simplifies Management
The need for higher availability, clustering functionality, and scalability
drove Vanderbilt to select Oracle Application Server to replace our existing
BEA WebLogic software. We found that Oracle offered greater scalability and
simpler management capabilities. Capacity challenges, for example, are now addressed
by simply adding a server to the environment and making minor configuration
changes. Now, we can bring a server online in just a few hours with minimal
system management burden.
The school’s new application infrastructure will have approximately
30,000 users, including students, faculty, and staff. We expect users to experience
performance increases of up to 50 percent from using features such as Web caching.
The One-Enterprise Approach
While we considered many vendors for our data warehouse initiative, few shared
Vanderbilt’s vision of a consolidated data center for university and medical
center operations. The majority attempted to sell separately to the university
and medical center. This represented a fundamental IT growth strategy stumbling
block until working with Oracle, which took the time to recognize that Vanderbilt
operates as one enterprise.
The university anticipates building on the benefits of a Linux-based environment
by extending its power across all remaining databases. With the university’s
data storage and processing rapidly expanding, we expect to add 20 processors
a year—now a simple and iterative exercise.
Vanderbilt launched its Linux and consolidated database initiative to contain
IT costs as systems scale and consolidate data for improved and secure access.
Careful planning and analysis has built a solid foundation for future growth
and allowed the university to maintain a single IT infrastructure across the
academic and healthcare organizations.