An open cloud platform is a lot like world peace. Everyone says they want it, but no one is sure how to get it. And in the end, would it really be the ideal state of affairs that we all dream about, or would it just lead to more problems?
To hear people talk, it seems like open cloud platforms and the end of vendor lock-in are just around the corner. VMworld was chock full of announcements from VMware, Citrix, Microsoft and others touting the benefits of the open cloud and the need for fluid environments that can shift data and applications across platforms according to user needs.
Nice talk, but it seems unlikely that the top hypervisor vendors will be able to pull it off in any meaningful way. And maybe that's why it pays to look elsewhere for the means to avoid vendor lock-in.
One possibility comes from Novell, which this week released Cloud Manager, a single-console management stack said to bridge the divide between VMware, Microsoft and Xen virtual servers. The idea is to give enterprises the ability to transcend virtual platforms, which should ultimately allow them to pursue any manner of public/private/hybrid cloud architecture they choose. The system relies on a centralized application server and a series of orchestration servers at each data center. True open cloud service, though, will rely on a set of adapters for the major public cloud platforms like EC2, which are still under development.
Of course, open source almost always leads to Linux, and it's here that you'll probably find the most forward-thinking plans. Red Hat is laying the groundwork for its Cloud Foundations infrastructure, which aims for nothing less than unfettered application and data mobility across platforms. This kind of mobility, the company argues, will require complete independence across hypervisor, OS, middleware and even programming languages, which the company aims to accomplish through its Deltacloud APIs. The set is currently under the purview of the Distributed Management Task Force (DMFT) for consideration as an industry standard, which means that not even Red Hat will have full control of the standard once it's issued.
The problem with open source is that it's only as useful as the industry support it receives, according to tech consultant Krishnan Subramanian. With VMware dominating enterprise virtual environments and Citrix quickly taking control of the public cloud market, nothing significant will happen without support from these two vendors. Perhaps it's better to aim for a universe of federated, interoperable clouds that would serve largely the same purpose. If single-vendor lock-in isn't appealing, would you settle for two or three vendors?
Still, the best way to avoid lock-in is to rely on your own internal infrastructure as much as possible, says processor.com's Chris MacKinnon. Diversity within your own house is attainable through targeted hardware/software combinations, and this, along with hybrid cloud technology, will afford greater flexibility as you venture onto public services. At the very least, if one cloud provider starts jacking up costs or can't meet service agreements, you can more easily switch to another with a mix of in-house technologies.
At this stage of the game, a community of open clouds is certainly feasible, but it likely won't be universal. Traditionally, open technology tended to require a greater deal of integration and compatibility testing than proprietary systems, but whether or not this state of affairs would extend to the cloud is uncertain.
In any event, it seems most enterprises are intent on getting cloud services up and running first, and leaving the fine-tuning for later.