The recent OpenStack conference (you can check out the OpenStack conference session and keynote videos here) has gotten me thinking, not just about where private cloud is going now, but also where, in the beginning of the era of broken trust in centralized computing, it might go in the future. Here’s what I mean.
There is no doubt that, as my colleague Charlie Babcock wrote, “it just got a lot easier to build OpenStack clouds in a place where they were once deemed unlikely: atop the VMware-virtualized environment of enterprise data centers.”
But I’ve also been doing some thinking about how OpenStack — or any other widely available, open source solution for compute and storage — could set the stage for an era of open source private cloud and parallel the era of the personal computer. Are we about to see the rise of the personal cloud? Maybe. Here’s what I’m thinking.
Amazon introduced the notion of “the mainframe of cloud,” if you will. A gigantic provider that allows you to take advantage of low cost, high-scale timeshare computing, and use as much as you want at any given time. But the downside? A loss of control.
I’ve been first in line among enterprise computing thinkers to pooh-pooh the Neanderthal notion that “if it’s not under IT’s direct span of control, then it’s inherently ungood.” So I don’t believe in private cloud because of unjustified control issues. There are, of course, use cases for private cloud, as Randy Bias’s “Cloud Repatriation” session eloquently discussed at the conference.
But let’s skip the whole enterprise discussion for a moment. As the consumer goes, so does the enterprise. I think we all understand that by now. Or at least, I hope we do. 🙂
We’ve just begun an era of broken trust. Welcome to the era of the NSA breaking the Internet, collusion by private companies in arbitrary spying, and a resulting distrust of proprietary/closed source products, at least by anyone who actually knows what they’re doing.
It is clear to me that in the same way that the rugged individualist was a huge adopter of the PC, that the rugged individualist will start to use cloud computing. I’m not sure that it will be native OpenStack, but OpenStack or something like it will be a basic platform upon which more complex services get built. Think of use cases like a personal file sync tool that doesn’t rely on AWS and Dropbox, instead built on OpenStack and other open source tools, something like OwnCloud, SparkleShare, and CodeLathe’s Tonido, among others.
Ecosystem will be a huge challenge for the personal cloud. Commercial providers like Dropbox, Apple, and Google are following the time-honored script of subsidizing their developer programs in order to drive end-user adoption. And if Apple’s “walled garden” model makes its way to other device manufacturers, where the device maker is the final arbiter of which software is allowed in its app store, that could be another challenge to a truly personally-owned cloud.
Yet, I think there are counter-currents to consider.
Here’s a thought experiment: ask an enterprise security geek about which consumer cloud backup solution you should consider. That person will first tell you that you’re better off using a local backup. If you press the issue, that person will then tell you to use the one that allows you to set your own private encryption key, set on the client, not the server, where the backup provider tells you that if you lose it, sorry, out of luck. (Such is the case with CrashPlan.)
Another counter-current: there’s the Maker movement, whose members believe that life is hackable, and whose members love to create their own products (even mobile phones) from widely-available parts. To be sure, this is related to the love of fixing, and a loathing of closed systems expressed best by the folks at iFixit, who, to be fair, are in business to provide tools and materials to fixers, but who also publish free fixing manuals and express indignation when manufacturers create products that are sealed and difficult to repair.
I stopped hosting my own servers back in 2004 (and removed the 19″ rack from the living room – that’s a tale in and of itself) because the economics and ease of using shared, external hosting made sense. Well, OK, my spouse hated having the rack in the living room, too. But the economics and availability also played a role. Do you really want all of your stuff to have a single point of failure?
But self-hosting may be coming back in a different form: pay a provider a low rate to host your own box in their data center. These don’t need to be huge honkin’ servers, even Raspberry Pi hosting can work for some personal use cases.
That’s where I’m thinking that open source cloud stacks come in: distributed object stores like Swift don’t necessarily have to live at at hosting provider like Rackspace. They could live on just about any hosting site, and if you’re paranoid, maybe they live at your trusted friend or family member’s house. Open source protocols — especially those that are developed with an international cast of characters — have the biggest chance of being secure. A quick and easy way to implement the stack on your equipment could spawn a personal cloud revolution.
There’s a lot of thinking still do be done before I’m convinced that personal cloud could work. But as we think, let’s remember the use case. In an era of broken trust, secure business is not the only use case. Bigger things could be at stake: life, liberty, and the pursuit of happiness.