Grid computing facilitates archive backup and recovery

Moving to a data grid rather than relying on traditional long-term archival backup can provide cost savings of up to 75% for hospitals, according to researchers at the University of Southern California Keck School of Medicine.

Community-sized hospitals could save 76% of their archive backup costs using the data grid, said Brent J. Liu, Ph.D. an assistant professor of research of radiology and biomedical engineering at Keck.

The current approximate cost to replicate a 5 to 7-TB PACS archive for one year is $467,000, including server and storage. By using the grid for clinical data recovery, a federated facility would have to provide only additional storage space costs to support the storage. This would amount to about $114,000, or 24% of the current cost, a savings of $353,000, he said.

One of the first uses of grid computing by medical informatics was the concept of a federation of PACS-exploiting technology as a cooperative backup archive. The application could help save precious data during disaster recovery.

Essentially, a clinical site willing to join a data grid federation would have to provide a set amount of storage space. In return, the site would have the same amount of storage within the federated data grid continuously available for access when the onsite PACS archive encounters a problem, Liu said.

"A federation of PACS can be created, allowing a failed PACS archive to recover its image data from others in the federation in a seamless manner," he said.

Liu described the idea in a recent paper (Comput Med Imaging Graph 2005;29:95-102).

Clinical data stored within the grid are separate from any of the federated clinical sites.

PACS have matured significantly during the past 10 years, yet they remain weak in the area of clinical data backup, he said.

"Current backup solutions are expensive or time-consuming, and the technology is far from foolproof," Liu said. "Many large-scale PACS still encounter downtimes of hours or days, crippling daily clinical operations."

Facilities with storage area network technology already implemented for long-term storage can benefit from a data grid since they will only need to partition a section of the SAN for data grid purposes without having to purchase new devices.

"Not only do you get fully redundant backup, you get virtually continuous availability, with 99.999% uptime," Liu said.

A prototype testbed data grid is in operation composed of one research laboratory and two clinical sites. The Globus 3.0 Toolkit, codeveloped by the Information Sciences Institute at USC and Argonne National, is used to achieve grid connectivity.