As it turns out, a lot. Recently, a growing number of suppliers have begun talking about hybrid clouds. The idea is that you can get the best of both cloud worlds. That is, the security, reliability and control of an on-premise solution while at the same time tapping the elasticity of the public cloud. EMC CEO Joe Tucci summed it up at EMC World 2011 when he stated that hybrid cloud will become the de facto standard.

Interestingly, as is usually the case, the vendor marketing is ahead of customer adoption. In a recent Wikibon survey of enterprise technology executives, we found that the world of enterprise IT is still very confused about the cloud with 37% of respondents indicating that they have no clear cloud plans or that cloud is hype. Importantly, The most common cloud strategy as seen in the chart below is “we’re headed to the cloud but just getting started” (26%).

Early Days for Enterprise Cloud Adoption (Source: Wikibon 2011)

What’s also relevant is despite vendor rhetoric in this area, less than 10% of respondents believe that they will use a hybrid of public and private clouds. While we fully expect this figure to rise over time, there is little question we are in the early days of hybrid cloud adoption.

Why are so few IT organizations (ITOs) actively pursuing a hybrid cloud approach? I believe the answer boils down to three things: 1) Definition. Most organizations are already doing hybrid clouds – that is a mix of on-premise virtualization and public cloud strategies – but their strategies are not integrated and deliberate at this time; 2) ITOs are still getting their virtualization act together, moving through the application portfolio toward virtualizing mission critical apps. It’s too early for them to aggressively pursue so-called hybrid clouds; 3) The solutions on the market are immature and not fully baked.

Specifically, the hybrid cloud solutions today (e.g. the ones that Tucci refers to) are largely extensions of on-premise virtualization strategies. It’s unclear that is the right approach. VMware for example is pushing the on-premise envelopes by bringing virtualization to the public cloud but it’s unclear that’s what the market wants. The public cloud is dominated by Xen hypervisors and extending enterprise technologies outward may or may not be what customers and service providers desire. Maybe the market wants a different hybrid – e.g. think virtualized data center plus Amazon AWS. But the problem is this approach may be too complicated to manage and not deliver desired results in terms of the ability to move data out to and back from the cloud seamlessly, securely and privately; with enterprise-class SLAs.

One of the more interesting cases of hybrid clouds (‘true’ hybrid clouds) that I’ve seen is Zynga. Dave Cahill recently wrote an excellent summary of Zynga’s approach. In a nutshell, instead of extending an enterprise-class cloud to a consumer cloud, Zynga is building an enterprise cloud on consumer technologies. In other words, it’s building a private cloud based on ‘true’ cloud technologies. As Cahill writes:

Zynga’s private cloud, called zCloud internally, is essentially a commodity cloud based on AWS design principles and using XenServer at the core. Zynga manages its clouds, both AWS and zCloud, through a single pane of glass provided by RightScale. This capability was integral to Zynga’s deployment.

He goes on to point out:

What might get lost in the details is that this is a commodity cloud inside an enterprise, not an “enterprise” cloud inside an enterprise. There is a material difference between the two. Building private clouds using expensive commercial solutions are a submission to the existing, brand conscious, IT strategies. I struggle to understand how companies can extract real cloud economics out of these architectures. Instead, as Cloudscaling’s Randy Bias has argued, it is much easier to start with a commodity cloud and layer services on top of this to achieve the desired level of feature/function you need. Trying to work backwards from a premium priced “enterprise” into commodity cloud economics is an exercise I don’t understand.

The lack of hybrid cloud adoption underscores the problem. That is, to deliver a hybrid cloud and have end-to-end network visibility and management, a company has to actually leverage a public cloud service and be able to deploy a private cloud infrastructure to federate the two together for the formation of a ‘true’ hybrid cloud.

Amazon S3 is public cloud only, with limited web dashboard support. EMC/VMware is generally private cloud only, with no usage-based pricing. So if EMC, for example, tries to help a customer deploy a hybrid cloud infrastructure—in Joe’s words the winning model is going to be both moving the internal infrastructure to private clouds and federating that with a public cloud—well, that might mean EMC hardware in the your private cloud and Amazon managing your public cloud; and in this scenario, who do you hold accountable for any data latency or availability or data loss issues? The EMC/VMware response is to federate VMware clouds working with service providers. But it’s unclear that the economics of this approach will win.

The fact is, it’s not trivial to simply piece together a federated, hybrid cloud storage solution from different vendors. There are far too many moving parts—you can’t see and monitor both ends of the network. It’s difficult to guarantee an end-to-end quality of service level and you have too many cooks in the kitchen—no single throat to choke. In the case of Zynga – it had to architect its own z-Cloud hybrid to succeed. The problem is not every company is Zynga and can afford such an approach or has the need to scale its business like Zynga.

Nirvanix is an example of a company with part of a potential solution here. While it is solely focused on storage – and that’s only one piece of the puzzle, its file system can be used both on-premise and in the public cloud with Nirvanix data centers. I could see this as a potential solution with larger partners that have the go to market juice to compete with Amazon in the public cloud and VMware in the private cloud by doing essentially what Zynga did across a number of customers – that is, architect a commodity cloud solution for the enterprise that could be a true hybrid.

The point is, a ‘true’ hybrid cloud enables you to retain the data you want within the confines of your data center walls and offload less frequently accessed data or archival/backup data to the public cloud (copies/replication/DR); at commodity pricing.

‘True’ hybrid clouds have the ability to obtain storage resources at the click of a button—no POs, no haggling with procurement or annoying sales reps. Cloudbursting means when you need capacity instantaneously to support a spike in workload during a holiday selling season, or during peak trading hours, you can burst out additional capacity from the public cloud and return that capacity when no longer required. And you’re only billed for what you actually use—whether on-premise (as an internal customer) or off.

Managed services providers and corporations looking to partner with a cloud service provider or acquire cloud services for their own company should really take a careful look at the capabilities of all the vendors. Do they own all or just parts of the cloud infrastructure? Can they really deliver on-demand, pay-by-the-drink, elastic compute and storage at commodity pricing? Users should beware of traditional enterprise vendors jumping on the hybrid cloud bandwagon with a business model that is still in its infancy. While there’s an allure to some of the messaging in the market, the Zynga case study underscores that things are changing very quickly and the consumer side, not the enterprise side of the market, is driving the economics of the business.

So buyers beware and do some homework before buying into the hype.

[Cross-posted at Wikibon Blogd]