Science Grid This Week
November 30, 2005 Current Issue | About SGTW | Subscribe | Archive | Contact SGTW  
Ann Chervenak: Connecting Scientists and Data

Ann Chervenak
Ann Chervenak
When data-intensive applications and distributed collaborations meet, the result can be myriad copies of large data sets located around the world. Ann Chervenak helps these collaborations easily find and reliably transfer these data replicas.

Chervenak is a Research Team Leader and Research Assistant Professor at the University of Southern California's Information Sciences Institute, where she works on data management as part of the Globus Alliance core team. The Globus team, which includes researchers from Illinois, California, Sweden and Scotland, develops fundamental technologies behind the Grid.

Two software systems for data management command most of Chervenak's time. The replica location service (RLS), a lower-level service that is a full-fledged part of the Globus Toolkit, registers replicas of data and discovers them later. The data replication service (DRS), still a work in progress, is a higher-level service that integrates the RLS with reliable data movement technology.

"I can ask the DRS to bring 1,000 files to my local site," explains Chervenak. "It will find them on the Grid using the RLS, figure out where all the replicas are, and invoke a reliable file transfer to bring them to me. Once they're successfully transferred, it will register those new files in the RLS so that they can be discovered by someone else."

The design of the RLS and DRS are driven by the needs of applications such as the Laser Interferometer Gravitational-Wave Observatory (LIGO), the Geosciences Network (GEON) and the Earth System Grid (ESG), so Chervenak spends a lot of time working with scientists from those collaborations.

"Collaborations like LIGO have done a lot of work on higher-level services, but their systems are customized," said Chervenak. "We don't want other data-intensive collaborations to have to do these implementations themselves."

With a schedule already full designing data management systems, preparing talks, writing papers and traveling to meetings, Chervenak also finds time to teach graduate courses in grid computing.

"I've taught two graduate seminars since coming to USC, and I enjoy it," said Chervenak. "Teaching these classes is a team effort, with Globus designers and developers giving lectures to the graduate students on their specialties."

—Katie Yurkewicz