Cloud storage giant Dropbox has announced that it will be pulling all its data out of Amazon’s cloud. Dropbox stores tens of billions of files on behalf of nearly 500 million users, allowing users of the service to easily store files on the Internet, send them to others, and synchronize access to them across a range of different devices, including laptops, phones, and tablets. Previously, Dropbox relied on the Amazon cloud to get this job done—a cloud computing service powered by online retail-giant Amazon that allows virtually anyone to build and operate their own software without the hassle of having to set up their own hardware. That basically means that Dropbox was storing its users’ billions and billions of files on Amazon’s machines, as opposed to directly on its own machines.
Now, however, that is all about to change. Dropbox will now be using its own in-house storage system, dubbed “Magic Pocket.” Magic Pocket, the brainchild of Dropbox engineers, has been in the works for well over two and a half years now. It makes use of an entirely new type of machine, called Diskotechs, and was built with an entirely new programming language. These Diskotechs are pretty compact, measuring in at 1½ feet by 3½ feet by 6 inches, but they can hold a staggering amount of data—up to a petabyte of data, or a million gigabytes. In fact, just 50 of these ultra-powerful machines have the capacity to store everything mankind has ever written. Pretty impressive, right? Dropbox has already moved about 90 percent of the files it stores onto Magic Pocket, with plans to migrate the rest over to the new system soon.
Why Build Your Own Data Center?
Many are asking why Dropbox would invest all that time and money into building its own datacenter when it could have just kept using the Amazon cloud. The answer to this question is relatively simple: by investing in its own infrastructure, Dropbox stands to save a significant amount of money.
First and foremost, Amazon isn’t letting Dropbox use the Amazon cloud for free. They’re likely paying a pretty penny for the service, which basically means they need to pay for the same infrastructure over and over again. Amazon is making billions off the service, reporting $2.41 billion in revenue for its Amazon Web Services division during the fourth quarter of last year alone. That works out to be approximately 9.6 billion per a year. By investing in its own infrastructure, Dropbox is able to eliminate this cost and, presumably, boost its bottom line. Most experts say the move will generate significant economic value for Dropbox.
“Nobody is running a cloud business as a charity,” explains Aditya Agarwal, the current Dropbox vice president of engineering and a former Facebook engineer. “There is some margin somewhere.”
Secondly, the move is advantageous in terms of disaster recovery. Let’s say Dropbox has to recover hundreds of GBs of TBs of data from a public cloud like Amazon or Carbonite. Not only is this a tall order, requiring quite a bit of time and effort, it also might not even work. However, with its own data center, Dropbox can easily do a local restore inside the data center and recover a customer’s data in a timely manner.
The bottom line is that in the near future we will likely see more major companies abandoning the Amazon cloud in favor of their own in-house infrastructure.