Cloud computing services have become integral to the operations of a growing number of enterprises that are under pressure to streamline tasks and cut costs. They've tried to do so through better utilization of scalable storage, automation and programmable interfaces. Along the way, many of the stakeholders – service providers, businesses teams, IT managers – have become caught up in debates about private, public and hybrid cloud terminology. The prominence of these terms makes sense because they're often associated with different levels of security and control – but how different are they, really?
The evolving perception of what constitutes a private cloud
In the past, major public cloud providers such as Amazon Web Services have positioned the private cloud as something that's out of step with modern IT needs. That is, they argue that running most services on-premises or in a data center, behind the corporate firewall, can't achieve the level of elasticity and price variability associated with infrastructure-as-a-service.
Going further than that, many observers have disputed the fundamental notion of the private cloud. Five years ago, InformationWeek's Andrew Murray drew the line between storage virtualization and real cloud storage, concluding that there was no such thing as a private cloud since many solutions advertised as such didn't connect to any third-party infrastructure.
In more recent times, field experts have similarly warned organizations about "cloud washing," or the phenomenon of rebranding older services – often related to virtualization – as true cloud computing solutions. They've also taken the opportunity to point out the flaws – complexity and suboptimal resource utilization – that beset many private cloud implementations.
Writing for TechTarget, Mark Eisenberg outlined the issue as one of squaring the technological features and implementation of private cloud with the implicit goal of any cloud project: Namely, to solve problems at scale.
"Private clouds can meet all of the requirements and provide the value of cloud computing; however, additional effort is required to implement and apply them," Eisenberg wrote. "Remember, clouds solve Web-scale problems, and those methods can help fix some enterprise-scale problems."
Eisenberg focused on resource pooling, broad network access for storage, on-demand self service and rapid elasticity as some of the essential features that make a private cloud work. Fortunately, many vendors have been working on products that address these requirements, such as Seagate's Kinetic Open Storage that uses Ethernet-connected drives to eliminate the storage server tier.
These solutions point the way toward more capable data centers and more robust private and hybrid clouds. Still, there's some uncertainty how private environments fit into the changing cloud picture – or if they count as "clouds" at all.
Private and hybrid setups redefine what "cloud" means
Amazon has softened its stance toward the private cloud, realizing that companies have some workloads that they may never be willing to migrate from data centers to public infrastructure. Security and control are some of the most commonly cited reasons for keeping applications and data close to the vest in this way. Organizations may choose to keep substantial portions of their setups on-site or in a data center and use the public cloud only as needed for bursting.
The importance of on-premises IT has produced increasingly elaborate infrastructure that mimics the core characteristics of the public cloud. While Murray once argued that heavily virtualized environments weren't clouds because they didn't rely on someone else's instances, this formulation may now be outdated. Writing in an Intel forum, HP's Christian Verstraete took issue with the classic arguments against private cloud, i.e., that it isn't on-demand, scalable, self-provisioned and measured (OSSM).
Instead, Verstraete pointed out that some private solutions meet most if not all of these requirements and play an important role in converged infrastructure strategies. For example, with a growing number of enterprises running services from different locations, it now makes sense to measure usage regardless of origin – and measurement is arguably the hardest of the OSSM requirements for private cloud to meet.
Moreover, observers may be throwing the baby out with the bath water, seeing a few cloud-washed implementations and thinking that the private cloud just doesn't exist. This attitude overlooks the private clouds that actually meet most OSSM criteria, plus it fixates on the public/private/hybrid labels when the distinction may be becoming increasingly secondary for companies that simply want to set up environments that work well.
"If through self-provisioning I can initiate a service, well, it seems I am in an 'on-demand' environment," Verstraete explained. "The combination of virtualization and automation allow me to provision that service transparently. Again, within the concept of converged cloud, whether the actual resource is provided within the enterprise (through a private cloud) or made available by a service provider (public cloud) does not make a difference."
With the rise of the hybrid cloud, this stance makes a lot of sense. Some companies aren't just drawing a line between private and public but finding ways to use both, in cases such as cloud bursting that require extra capacity for supporting peak workloads. Piston Cloud Computing CEO Joshua McKenty colorfully noted that enterprises are taking an approach akin to buying underwear (private cloud) but renting a tuxedo (public cloud) – it's not a case of either/or, but instead two different solutions addressing separate requirements.
Gartner famously noted that enterprises have high aspirations for these types of hybrid cloud deployments but aren't quite there yet with implementation. In this sense, the path toward hybrid cloud mirrors that of private cloud a few years ago, when companies often expected to set up a perfectly functioning environment in no time but ran into roadblocks en route, Eisenberg noted.
Ultimately, many realized that the "private cloud" wasn't a monolithic entity but rather a term that could apply to a lot of different use cases. Examples include development and testing, enhanced virtualization and even something that Forrester Research characterized as "public cloud lite" – the ability to make resources available quickly but with a wide range of customization options. Going forward, it makes sense to look more at how each type of cloud can address a company's particular requirements and less at differences in terminology.