Tumult nas nuvens

por Brendon Nafziger, DOTmed News Associate Editor | April 29, 2010




For a health care center, the most annoying part is that these massive image archives are largely fallow. "Ninety-five percent of the access [to archived data] will take place in the first three years, after that access patterns drop," Kermani tells DOTmed News. According to other estimates, cessation of use is even faster: after 90 days, most radiological images aren't pulled again.

Still, just because they're ignored doesn't mean they can be destroyed. Because of HIPAA regulations, the images have to be shelved for a specified amount of time - typically, around seven years. But this medium-to-long-term storage brings with it demands of maintenance, security and back-ups, and the associated expense and manpower needed to ensure the data is safe.

This is where cloud computing could truly shine. In effect, it's about outsourcing those responsibilities to outside professionals in the same way you outsource natural gas delivery to National Grid.

The utility model

"The premise of cloud computing is gaining access to computational resources...in a manner that is analogous to how people consume utilities like electricity, water and telephone systems," Kermani says.

"It all comes down to economics. If I could use storage as a utility, I don't have to own it. I don't have to back it up," Kermani says. "I don't have my own generator in my backyard making power, why do I need my own data center?"

Or as Ryan Howard, CEO of Practice Fusion, a free cloud-based electronic medical record service out of San Francisco, Calif., puts it, "When you need water, you don't build your own water pipe to the lake. When you need electricity you don't build your own power grid. Cloud computing is the same concept."

According to Kermani, while the term - and the buzz - is new, the idea isn't. It goes back to at least the 1960s, with the creation of Multics, an early operating system-cum-computer developed by professors at MIT, who wanted to build a system where individual users would share the computing resources of a common infrastructure. They called it timesharing. In the 1990s, it emerged again as application service providers, and by the beginning of this millennium, it was going by the moniker grid computing.