Will Sun/IBM deliver on open cloud computing?
Sun Microsystem's Sun Cloud Compute Service may be a very well-conceived open cloud service, but would an IBM acquisition let it execute?
The clouds seem to be revolving around the Sun today.
As CNET reported earlier, Sun Microsystems has announced the Sun Cloud Compute Service to attendees of their CommunityOne developer event. In short, Sun is following through on its promise of delivering a cloud infrastructure service, initially targeting developers, students, and start-ups.
What I love about this announcement is how well it leverages Sun's extensive cloud-computing DNA. Open source/open APIs, a focus on developers, great use of Sun hardware and software assets--it's all there. In fact, I have to give Lew Tucker, Sun's CTO of cloud computing, and his team credit for being among the first to assemble a cloud offering from scratch that I think will really appeal to developers.
The service revolves around the concept of creating Virtual Data Centers (VDCs), a concept already made popular by 3TERA, VMWare, and others. The idea is that you go to the Sun Cloud, create an account, create one or more VDCs, and use the VDC to allocate servers, storage, IP addresses, network services, etc., as required to build an infrastructure for whatever your purpose may be.
Perhaps the most interesting aspect of the entire offering to me is the API sets offered. Tim Bray offers some detail, but the API itself is available under the Creative Commons "Attribution" license. It's a well-thought-out API set, true to Sun's ethos of "engineering for engineers, by engineers."
For example, storage will run with WebDAV and the Amazon AWS S3 API, both phenomenal standards to choose if ease of integration into existing applications and management tools is your goal.
The compute side will leverage a REST API that returns JSON containing information including links to sub-components of the requested resource. As Tim explains:
To use the API, you need one URI to get started; it identifies the whole cloud. Dereference that, you get some JSON that has URIs for your VDCs (Virtual Data Centers); deference those, and you get more JSON that represents your clusters and servers and networks and so on. This has the nice side-effect that the API doesn't constrain the design of the URI space at all.
What Tim means in that last sentence is that Sun can extend the API easily to include any new resource types or services that Sun may want to add to the offering. Nice.
But can they deliver?
I would rhapsodize about how well Sun hit the sweet spot, if it were not for a couple of things working against them. First, the Amazon Web Services APIs are slowly becoming the de-facto standard for cloud infrastructure APIs. EUCALYPTUS, the academic "build your own cloud" infrastructure uses them. What's more, the Cloud Computing Interoperability Forum is reportedly working on an Amazon compatible API, and even Sun capitulated to the S3 API.
GoGrid is also promoting its API under the Creative Commons Attribution license.
However, Sun's strong engineering and marketing muscle gives it a chance here, and I would remain highly optimistic if not for the other huge piece of news today: IBM is reportedly in talks with Sun to acquire the data center stalwart for $6.5 billion dollars.
As ZDNet's Larry Dignan notes in his coverage, a deal like this one is long overdue:
The companies mesh on the open-source software front, Sun is struggling, and IBM can consolidate some server market share.
Now, IBM loves open source, and loves the developer community just as much as Sun. But Big Blue already has a cloud strategy in place. It is incredibly uncertain what, if anything, will happen with the existing Sun Cloud Compute Service were such an acquisition to happen.
It is conceivable that IBM would keep the service running. On the other hand, IBM engineers have plenty of skin in the cloud game as well, and it is unclear whether a brand new Sun offering could survive a political battle were an acquisition to happen. The good news is that the Sun API is already in the open, so others that choose to implement it would likely be protected from any legal harm. Further development of the API, on the other hand, is far from assured in this scenario.
In the end, the success of Sun's endeavor will likely depend on two things: what any given acquirer decides is right for its cloud business, and the rate of adoption for the nascent Sun Cloud Compute Service offering. Rapid adoption could make it difficult for IBM or anyone else to change the market's mind.
Still, it's a brilliant offering on paper. Let's hope that brilliance makes it to the market before it gets dulled down or snuffed out all together.