Database Trends and Applications Vol. 16 No.7
DB2 on Linux Powers High Performance Database Research Center
By Walt Jordan
Now that you have the data, what can you do with it? That
has been a question that has confronted researchers since
commercial imaging satellites were launched several years.
The satellites have the capacity to stream terabytes and
terabytes of data to earth-capturing images of
neighborhoods, farm land, erosion patterns and a lot of
other useful information. Making that data easily
accessible to a broad cross-section of users has long been
a challenge.
That is the challenge being addressed by Naphtali Rishi,
and his team of researchers at the High Performance
Database Research Center (HPDRC) at Florida International
University. Rishe has led the development of Terrafly, a
Web-enabled system designed to aid in the visualization of
remotely sensed and spatial data. Funded in part by
multimillion grants from NASA and the National Science
Foundation, TerraFly, the research center's highest profile
project (www.terrafly.com) is based on a 20 terabyte
database, which includes textual, remotely sensed and
graphical data such as maps. The data can be viewed and
manipulated using standard Web browsers. As users "fly
over" remotely sensed data, other digital data associated
with each specific location can also be viewed. "TerraFly
is intended to be the world's largest database accessible
through the Web," Rishe told DBTA in an exclusive
interview.
TerraFly is being developed at NASA' Regional Applications
Center, which is a part of the HPDRC. According to Rishe,
the project could have a wide range of applications. For
example, potential homeowners could survey different
possible neighborhoods and demographic statistics, the
educational facilities, the location of shopping areas and
other pertinent information could be displayed. Or people
planning a vacation could explore different getaway spots
and then automatically access the online reservation
system. Indeed, the opportunities for TerraFly seem so
enormous that officials at Florida International University
estimate that it may generate up to $1 billion year in
revenues marketing the technology.
TerraFly is powered by a DB2 database running on Linux on
an SP Supercomputer. "I like the efficiency of DB2," said
Rishe. "I look to cost effectiveness of DB2 on Linux. And
I like the good adherence to standards." IBM has been a
generous supporter of the HPDBRC. Florida International is
a participant in the IBM DB2 Scholars Program, which
supports organizations dedicated to setting new standards
for database application excellence, FIU receives DB2 site
licenses for development purposes at no cost. This helped
FIU get its project off the ground cost efficiently.
Moreover IBM has provided a $6.7 million grant to fully
outfit the High Performance Database Research Center with
state-of-the-art hardware and software.
"The SP is the biggest horsepower unit I have in my
possession," said Rishe.
The TerraFly project is part of an extensive research
program into database technologies conducted by the 90
professionals associated with the HPDBRC including 10
principal investigators. The specific research questions
involved with TerraFly include developing algorithms
designed to manage large amounts of imagery and to link
imagery with vector and point data. Finally, Rishe and his
team are looking at ways to more efficiently deliver data
over the Internet. So far, the project has been awarded
three patents and has several other patents pending.
Rishe's research into semantic/object-oriented databases
has led to the development of two applications. The first
component is called a Semantic Wrapper and is an open
middleware system that provides a semantic view of
relational databases and enables queries using Semantic
SQL. "It is a non-invasive enhancement of relational
databases allowing empowerment of the willing users to
easily pose complex queries to existing databases via
familiar tools," Rishe said.
The Semantic Wrapper reverse engineers a relational
database to create semantic schema. The process also helps
DBAs take inventory, document and analyze legacy databases.
"It lets you take control of your legacy databases," Rishe
told DBTA. "You get a map of the database logic and can
understand what kind of queries can deliver."
Rishe has also developed a complete semantic DBMS. The
semantic database stores the meanings of information as
facts about objects. With a semantic database, no keys are
needed and any query can be run ad hoc. Rishe hopes to
commercialize both the Semantic Wrapper and the Semantic
DBMS. More information about he products can be found at
http://hpdrc.cs.fiu.edu/
Rishe's work has won strong endorsements from the research
community. After a demonstration, William Campbell, head of
NASA Goddard Applied Information Sciences Branch, wrote,
"Your semantic database research is specifically germane to
many of the challenges NASA and other data generating
institutions currently face I the management and retrieval
of information relevant to our user community."
Copyright Notice