Supercomputer As a Service
gubm writes "Nearly one and a half years after making a stunning entry into the global supercomputer list with Eka, ranked as the fourth-fastest supercomputer in the world, Computational Research Laboratories (CRL), a Tata Sons' subsidiary, has succeeded in creating a new market for supercomputers — that of offering supercomputing power on rent to enterprises in India. For now, for want of a better word, let us call it 'Supercomputer as a Service.'"
Or, we could call it what everyone else is calling it. Grid computing or sometimes cloud computing.
Dual Opteron < $600
Isn't this how it was done back in the day, with supercomputer time "leased" to companies who needed it?
My uncle used to work for Minnesota Supercomputer Center and that's how he explained it to me; seemed pretty simple to my 12-year-old mind back then.
Rent-A-Hal. "I'm sorry Dave, I've been repo'd"
The idea is very old, and contrary to the article there are plenty of people offering similar services: http://news.softpedia.com/news/Rent-Your-Own-Supercomputer-for-2-77-per-Hour-82166.shtml, http://www.hoise.com/primeur/00/articles/weekly/AE-PR-04-00-20.html, http://www.linuxplanet.com/linuxplanet/reports/4590/2/, etc.
Is their offering cheaper? Unfortunately the article didn't tell us.
Isn't this what the Storm botnet and the conflicker botnet are doing already? Say instead of storm doing something useful like fold proteins, find ET, or pull us out of this financial mess couldn't a person with a lot of money use the botnets for a useful purpose instead of spam or a denial of service attack?
Its not only nothing new, we never stopped renting high-performance computing time. In some cases, it's ancient supercomputers that aren't all that super any more, but that the applications are so large and difficult to port to other machines, we just kept using them.
Brett
Supercomputing as a service is nearly as old as computers are. Granted they were called mainframes.
Frankly I'm amused at how we seem to be regressing 30 years. I expect any day to see dumb terminals and a prognostication that soon the world will need only a few [cloud] computers.
we forgot to call it a cloud, and put it in a grid configuration after we service oriented it
to leverage our core concepts. obviously all the packets will fall out of it.
Good people go to bed earlier.
For want of a better word? Um, guys, we have a better word for it: timesharing service bureau. We came up with it back in the 60s to describe a business that bought these hugely powerful, hugely expensive things called "mainframes" and sold access to them to customers. Customers could load their software and data onto the TSB's mainframes and run their programs there, paying for only the compute time they needed as they needed it. The TSB would also charge per kilobyte per month for disk storage (data and programs) and per minute for terminal connect time. Replace "mainframe" with "supercomputer" and you've got this new service (minus the connect-time charges since we're no longer using dial-up modems).