There are varying definitions of "hybrid cloud" depending upon whom you ask and what they're trying to sell you. But they all have this in common with one another: They refer to a single software hosting platform whose resources come from multiple sources, some from the public cloud (e.g., Amazon AWS, Microsoft Azure, Google Cloud Platform) and some -- perhaps many -- from your organization's own data center.
What the phrase really means
Here is a definition you can apply to today's technology and to modern data centers: A hybrid cloud is the sum total of an enterprise's digital assets that share a common cloud infrastructure platform. This means, they're sharing the same network, and are managed, or at least overseen, using the same tools. Many enterprises have their own data center assets, and they lease applications or storage from a software-as-a-service (SaaS) provider. That's not the same thing as a hybrid cloud. All of the organization's software, including its applications, data, and database services, are hosted by a combination of private assets and public service providers.
The "cloud" part of hybrid
You might think that, after well over a decade, it would seem somewhat redundant for us to define "the cloud" here and now. Commonly, it refers to pay-as-you-go services from a commercial service provider. But there's a technical meaning that has evolved somewhat from the common meaning, and that applies specifically to what's being "hybridized" on a hybrid cloud platform:
A cloud, no matter where its servers are physically located and whom they serve, enables all of its resources to be implemented, deployed, and managed as though they were part of a collective unit. A cloud platform produces a cloud, whether it's AWS or your own virtual data center. If your enterprise has its own servers, in owned facilities or colocation centers anywhere in the world, and they are all managed on one platform to provide services to users from anywhere on that platform, that's a cloud.
Software is at the center of the cloud
Now, you may be thinking, wait just a doggone moment here, that sounds an awful lot like the definition of hyperconverged infrastructure (HCI) published here not very long ago. And you're right, doggone it, it certainly does.
The difference has to do with the services that the software performs on a cloud platform. When we refer to software, a discrete service is a task that the programs perform; an application enables such services.
OpenStack is one example of a platform that produces private clouds, or hybrid clouds once services from outside suppliers come into the mix. Though it has many uses today, OpenStack's original purpose was to enable its users to cluster together as many servers as they can, to host virtual machines (VM) privately the same way Amazon AWS and others were starting to do publicly. Enablement in this context meant pooling resources together, not unlike what HCI does. Some years back, I helped coin the phrase cloud dynamics to mean thinking about computing resources, such as processing power, like pooled commodities rather than rows and blocks of servers.
A private cloud uses servers and facilities ("infrastructure") that an enterprise owns in their entirety, to host services the way Amazon does. But now that the services belonging to any enterprise's data center arsenal are comprised, if only in some small part, of services and applications from a public cloud provider, it's difficult to identify any cloud in production anywhere today that is entirely private.
The "hybrid" part of cloud
The phrase "hybrid cloud" came to prominence a long, long time ago (around 2012) when it was a very new thing for the parts of an organization's data center to reside partly on-premises and partly on third-party, commercial services. For instance, to have your customer data residing someplace in Salesforce's cloud platform, your business operations data residing on-premises, and your analytics supplied by Salesforce and Tableau and Microsoft, suggested that a variety of sources were contributing to this bigger picture of a single data center. In the minds of many IT leaders in business today, this is what it means to have a hybrid cloud: To operate a data center whose components are partly owned and partly leased.
But as you've just seen, "the cloud" is rapidly evolving. It used to be a blanket phrase that meant, whatever IT resources are "out there," owned by someone else. Then it came to refer to those IT resources that are provided to organizations as services -- like commodities that are paid for as they are used, perhaps on a transactional basis, rather than leasing the server or the facilities that host those services.
Now that phrase is changing again. Today's applications have components that are hosted in multiple locations. Connectivity has become so rapid and so ubiquitous that these components no longer have to be centralized in one data center facility to be useful or practical. Colocation makes it feasible to distribute computing, data storage, and networking resources wherever it's most beneficial. A new and emerging concept called the compute edge -- a spot where it's faster to collect the data from multiple users or remote sensors, process it immediately, and report the results -- is drawing processing power away from massive, hyperscale facilities, and in some cases towards processing centers so small they fit in the back of a pickup truck.
From many, one
If your organization manages a distributed application -- perhaps one that employs microservices that wink into existence, get their instructions from the orchestrator, do their jobs, contribute to the collective, and wink out -- then pieces of your business' functionality may be scattered all over the planet right now. Most likely, your enterprise already operates a hybrid cloud.
In this new frame of reality, a hybrid cloud is a virtual data center gathered together from multiple locations and service hosts, but operated and managed as though they were one place. Quite possibly, all data centers already operate in this manner. (We could say "virtually" all of them, but that term's been taken.) It's containerization that not only makes this methodology more feasible but more efficient, analogous to how the interstate highway system with lots of little vehicles ended up being more efficient than the Transatlantic Railroad with fewer, bigger ones.
Where containerization and orchestration enter the picture
Just as there are many definitions of "cloud" in the context of computing, thanks to the success of marketing and, for some reason, the proliferation of tech blogs, there are a number of ways one can define a "container" in computing. This is not a safety net to allow me to throw yet another definition out there and defend it as just as accurate as all the others.
A container (of the style popularized by Docker) includes just enough of a program to be functional on a server, and hosted by its operating system. Its isolated, so it communicates with the rest of the world only through an Internet Protocol network. An orchestrator, such as Kubernetes, coordinates several containers together into an application.
Now, doesn't that mean Kubernetes manages a cloud? If that platform enables a user to provision fundamental services and applications automatically, like a public cloud provider would, then the answer is yes. You don't have to be hosted by VMs to be a cloud.
Containerized applications are designed to be highly distributed, as well as to incorporate services from the outside world -- from the public cloud. So if a distributed application being orchestrated by Kubernetes, hosted on-premises, includes services such as Amazon Relational Database Service from the public cloud, then isn't the combined product of these services a hybrid cloud? Yes, it is.
Why do we need a business revolution to hybridize a cloud?
So if essentially any cloud platform that combines owned services with outsourced ones is a hybrid cloud, then why is the concept still being "productized," as something of exclusive and unique value into which enterprises should invest? Hybrid cloud remains a thing because it still appears to offer something organizations still desperately need in their data centers: Flexibility.
"Today, we see that fundamentally, there's two paths: Hybrid cloud and public cloud," pronounced VMware CEO Pat Gelsinger, during an August keynote address to the VMworld 2018 conference in Las Vegas. "And these paths are complementary and co-existing. But today, each is being driven by unique requirements and unique teams.
"Largely, hybrid cloud is being driven by IT and operations," Gelsinger continued, "the public cloud being driven more by developers and line-of-business requirements. And it's a multi-cloud environment."
Later in that same address, the CEO admitted to his audience that his company is well aware of how the hybrid cloud concept is evolving. "We're growing in our definition of what the hybrid cloud means," he said, "Through VMware Cloud Foundation, we've been able to unify the private and the public clouds together as never before." He's referring to his company's product for organizations intent on deploying hybrid clouds -- data center platforms for their business users that incorporate services supplied by public cloud providers that also happen to be VMware partners.
I asked Gelsinger later, if VMware succeeds in this effort to "unify" (his word) these two classes of platform, then will there be a time when the phrase "hybrid cloud" has become archaic?
"I think in many regards, the task of technological evolution is essentially to sediment," he responded.
"When I began my career, I was strapping flip-flops together to build registers on a single-board computer," the CEO continued. "So I think we're sort of constantly sedimenting. We're barely getting an enterprise-grade container offering, and now, oh! It's being left over and left behind by function-as-a-service. Hey, and then we're going to have the next thing pass function-as-a-service. We're constantly sedimenting these technologies so that we can build and innovate those higher levels of abstraction, that allow us to go faster and do more things. So will it become an archaic term at some point in time? I hope so. To some degree, that is the exact strategy of VMware."
Agility as the new driver
Larry Socher is the global lead for the Infrastructure Services Growth and Strategy practice at consultancy firm Accenture. During a session at VMworld 2018, Socher told attendees that, although his firm's clients tend to still be interested in hybrid cloud, their reasons are changing. In the beginning, they were sold on the promise of moving data center costs from capital to operating expenditures. Since then, the architecture of the underlying infrastructure has metamorphosed into something beyond the wildest dreams anyone may have had in 2012.
Also: How to buy storage CNET
As a result, the interval between Accenture clients focusing on cost savings and then focusing on speed of innovation, Socher said, became no longer than 18 months. "Agility became a lot bigger driver," he told the audience.
Cloud adjacency as a backseat driver
In 2015, Accenture acquired the unfortunately named advisory firm Cloud Sherpas, incorporating its consultants into its "Cloud First" applications team initiative. At first, remarked Socher, they set about to advise clients on how to build a business model around the development of cloud-based infrastructure in its data centers, how to adopt more viable and cost-saving services from public cloud providers, and how to draw up and follow extensive roadmaps for realigning businesses around those services. Applications were still considered principal drivers -- for example, Salesforce's game-changing SaaS -- but only as incentives towards the broader goal of infrastructure improvement.
Over time, the modernization of applications moved from being the incentive to being the goal. Realizing this goal, as Accenture advised clients, would often mean recomposing software development and IT support teams around smaller teams with more granular, more attainable objectives. "I think what's interesting about the operating model, it really is a multi-dimensional function," he said. "You've got to deal with so many different elements as you look at it.
"The client's definition of 'private cloud' -- the whole world of what we call 'private,' 'public,' 'hybrid' -- is really morphing right now," Socher continued. "We see a lot of our clients who still feel that they want to maintain a huge private cloud data center. At the same time, they don't want to be in the data center business. So we see a lot more interest in, 'Can I move into colocation facilities for cloud adjacency?' Obviously, Equinix is going through the roof."
Also: Report: Hybrid cloud strategies growing fast TechRepublic
By "cloud adjacency," Socher is referring to one of the star features offered by colocation providers such as Equinix, Digital Realty, and RagingWire: Private, often fiber optic, connectivity directly to public cloud providers. Indeed, enterprise colo providers' connections to Azure, AWS, and Google are quite likely visibly, appreciably faster than direct Internet connections between the enterprise themselves and these same providers. Colos sell this premium connectivity as a service, often borrowing a cloud economics model -- for example, Equinix Cloud Exchange.
Although they may not say it up front, colo providers are now in the business of selling hybrid clouds -- in some cases, reselling the very public cloud services that drove customers to consider a hybrid model in the first place.
Why can't a hybrid cloud be built just like any other cloud?
For the past four decades, businesses have been told they need a mindset shift or a culture change or a digital transformation in order to convert themselves into the best possible customers for some piece of technology, like being told you can't really appreciate a certain food unless and until you've had tongue surgery. Accenture has constructed a list of dozens of individual business alterations, subdivided into three "waves," all of which its advisors believe are necessary before an organization can appreciate the complete value of a hybrid cloud.
I asked Accenture's Larry Socher, isn't it possible for someone to just make the call and adopt a hybrid cloud without having to metamorphose into a wholly new form?
"You will not get the value out of it," he responded. Referring to an Accenture client in the life sciences business, he said it was absolutely necessary for that client to change the way it operates.
"The real value is using analytics, automation, and eventually AI to transform how you operate," Socher continued. "Whether it's public or private, I don't care. That's what the real promise is: to make me more agile, more efficient, one guy managing more things. And the only way to do that is to transform the people and the culture."
Also: Why 2017 was the year private cloud died TechRepublic
From a purely technical perspective, a hybrid cloud is not dependent upon an organizational remodeling. True, there are new and better ways to manage and execute a business strategy, especially if that business has a sound plan for investing in on-premises resources and public cloud resources, where they make the most sense.
But it's time for honesty: If the components of a cloud platform are, as the word "cloud" suggests, independent of their location or any physical binding to particular servers, then all clouds are either hybrid clouds now or can become hybrid clouds with one change to a config file. From an administrator's perspective, that's the only digital transformation that may be required.
Yes, the phrase is meaning less and less, but only because the technology to which it refers is making more and more sense.
Learn more -- From the CBS Interactive Network
- VMware touts multi-cloud strategy with expanded hybrid cloud portfolio by Asha Barbaschow, ZDNet
- For Microsoft, the hybrid cloud is the destination, not a staging post by Jack Schofeld, Jack's Blog
- Hybrid cloud: A cheat sheet by James Sanders, TechRepublic