Most enterprise executives expect to gain a number of key capabilities out of the cloud. Scalability ranks high on the list, as I mentioned in my previous post. But there is also broad interoperability, lower overall IT costs and the ability to gain steady access to the latest hardware and software platforms on the market.
Naturally, this has led to heightened interest in open cloud technologies. After all, what good is placing data and applications on the cloud if it can access only a limited portion of the overall infrastructure, or worse, winds up on distributed versions of the same silo-based architectures that plague legacy data center environments?
To many, then, an open cloud is the only way to go. As Ubuntu’s Mark Baker put it recently:
An open cloud world embraces hardware and software interoperability, is relatively free from binding contracts and allows for the free movement of data, applications and workloads from one cloud platform to another. For organizations that can’t afford downtime or the loss of their critical workloads — which is nearly everyone — there is no alternative to this flexible approach to cloud. In short, many principles of an open cloud mimic those of an ideal free market economy, creating something of a ‘free cloud economy.’ Embracing these fundamental principles, open cloud providers will compete to deliver better services because customers can easily take their workloads elsewhere.
His take is that vendors of all stripes will have no choice but to embrace open cloud technologies, namely OpenStack, because to do otherwise will limit their products’ ability to interact with the broader IT ecosystem, greatly diminishing its value in an increasingly cloud-centric enterprise.
Indeed, the recent release of OpenStack Havana has sent the open source community into a frenzy, because it is said to contain a number of critical features for the enterprise. Among them are new metering and orchestration tools that allow organizations to extend control of their data environments to third-party infrastructure, as well as Secure Socket Layer (SSL) support across all service APIs and firewall as a service capabilities that are intended to allay many of the security concerns that IT executives continue to voice over the cloud.
However, not everyone is convinced that OpenStack is the cloud nirvana that some backers claim. According to CloudPro’s Adrian Bridgwater, a close look under the hood reveals that it is more geared toward enabling providers to offer open clouds, rather than enterprises to deploy them within private settings. For one thing, OpenStack still lacks the kind of broad scalability that the enterprise is looking for. When you are talking about ramping up computer, storage, networking and related management/middleware stacks in a coordinated fashion, don’t expect OpenStack to provide an out-of-the-box solution. Like open source products of the past, there is still a fair amount of integration to take care of. And if you already have a substantial presence on AWS, you still have a fairly substantial migration challenge to port it over to an open platform.
The main issue for OpenStack, then, is confronting the high level of expectation that organizations have for the system, says Mirantis’ Anne Friend. She has listed a number of key mistakes that enterprise executives have made with open cloud technology, such as the idea that the absence of licensing fees equates to no real costs at all, and that open platforms can be easily integrated using in-house staff. As well, open system providers often have their own lexicon, making it difficult to mix and match capabilities that go by different terms. And more often than not, IT executives who do get in over their heads with open platforms are loathe to admit their mistake and simply compound their problems by deploying ever more complicated solutions.
Of course, it’s not fair to lay all of these problems at the feet of OpenStack or the open cloud community. True, some of the more strident boosters present open technology as the ultimate solution for everything that ails the enterprise, but in most cases there is a broad understanding that an open cloud, while not perfect, is a more desirable environment than a single-vendor proprietary one.
But the takeaway here shouldn’t be that open is better than closed, or vice versa, or that the enterprise needs to pursue interoperability and extensibility in the cloud to the exclusion of all else. Every solution must be judged on its merits – that is, does it solve a current problem or open up a new opportunity?
In that light, it would seem that an open cloud would suit many enterprise needs, but not all.