In a blog posting earlier this year, Ian Murdock, Sun Microsystems' vice president of cloud computing strategy and Debian GNU/Linux founder, wondered what might emerge as the cloud equivalent of the Linux distribution.
Murdock pointed out that the cloud computing world today resembles the early days of Linux, during which dabblers with a surplus of time and motivation could assemble and integrate their way to a Linux platform.
However, it wasn't until the Linux distributions emerged and whisked away much of the code-cobbling grunt work that the platform became broadly useful.
Certainly, the assortment of different Web-based services out there could be made more useful through a bit of Linux distributor-like massaging, but a menu of mashed-up Web services won't add up to a platform comparable to Linux.
The problem with that mashed-up picture--and the root, I suspect, of much of the anti-cloud sentiment that emanates from those without anything Web 2.0 to sell--is that it offers too little provision for packing everything up and moving it behind your firewall, or behind your laptop cover, if that's what you'd prefer.
Lately, the term "private cloud" has been turning up quite a bit, and, in many cases, I've seen the phrase met with derision, as though people who wish to organize their workloads a la Amazon or consume software in the style of Salesforce without entering into a blood pact with such an organization don't "get it."
The power of Linux as a platform lies in its capacity not only for taking on almost any task, but also for attacking those tasks almost anywhere. Certainly, you'd be hard-pressed to achieve the level of operational efficiency of an Amazon or a Google, but that's no reason not to seek a more portable future for the cloud.
When I consider what might emerge as the cloud-era equivalent of the Linux distribution, the most likely candidate seems to me to be the Linux distribution itself.
And, as I learned during a talk by Red Hat CTO Brian Stevens at last month's IDC Cloud Computing Forum West, the Linux world's most prominent distributor is working toward bringing just such a reality to pass.
Through a collection of separate projects, clustered around Red Hat's oVirt virtualization management project, the company has been working toward extending Linux's layers to encompass a complete cloud reference implementation.
At the IDC event, Stevens roughly outlined his company's plans to make this reference implementation available in the form of a publicly accessible test cloud running on Red Hat-hosted machines. At the same time--which, according to Stevens, should be around this summer--Red Hat will make the reference implementation available in a downloadable form suitable for installing on a pair of servers on ones' test bench.
As Stevens hastened to point out, Red Hat is not out to become a cloud provider itself; rather, the company views its plans to advance the state of cloud-building as the logical next step in the platform continuum that Red Hat, Ian Murdock and others began in the early '90s.
Since it began life as a bare kernel intended for educational purposes, Linux has steadily accrued higher-level stack layers, which now include the capacity for hosting virtual instances of itself or other operating system environments.
It stands to reason that Linux should continue scaling up, into a building block for any number of private, public or test clouds, each bearing their own set of the slight adaptations through which all technologies evolve.
UPDATE: Just a few hours after I posted this, I caught sight of Mark Shuttleworth's Ubuntu 9.10 name and goals announcement message. One of the goals for that release, and for the 9.04 version that will precede it, are cloud-building capabilities along the lines of what Red Hat has in mind:
"What if you want to build an EC2-style cloud of your own? Of all the trees in the wood, a Koala's favourite leaf is Eucalyptus. The Eucalyptus project, from UCSB, enables you to create an EC2-style cloud in your own data center, on your own hardware."