14 DAY TRIAL // Amazon can safely say that it has one of the largest and most versatile cloud networks in the world, especially given its ownership of a supercomputer.
The queer thing about Amazon's so-called supercomputer is that it is not really a supercomputer.
In fact, whether or not it exists depends on what Amazon decides is required at any given time.
A HPC (high-performance computing) application is a big set of machinery - building-sized really - composed of many nodes, each running one or more processors and/or GPU compute modules.
Amazon's supercomputer is nothing like that and, one might say, does not exist at all.
Instead, the company has something known as Elastic Compute Cloud, a technology that can access and, in a way, sync some or all of the data centers it has spread around the world.
In other words, Amazon's HPC runs on a virtual layer and can be fielded at the company's whim.
The other advantage is that anyone can access and use the HPC, something very noteworthy, considering that it ranks 42 in terms of performance (worldwide).
Not costing much to start up is another, huge advantage of the virtual machine.
“If you wanted to spin up a ten or twenty thousand [processor] core cluster, you could do it with a single mouse click,” says Jason Stowe, the CEO of Cycle Computing, a company that helps researchers and businesses run supercomputing applications atop EC2.
“Fluid dynamics simulations. Molecular dynamics simulations. Financial analysis. Risk analysis. DNA sequencing. All of those things can run exceptionally well atop the [Amazon EC2 infrastructure].”
Stowe believes that researchers will take to this new solution fast and will soon stop worrying about building their own research station.
“I’ve been doing this kind of stuff for awhile, and I think that five or 10 years from now, researchers won’t be worrying about administering their own clusters,” he says.
“They’ll be spinning up the infrastructure they need [from services like EC2] to answer the question they have. The days of having your own internal cluster are numbered.”