Hybrid cloud computing seems to be a less trendy topic these days than it was seven or eight years ago, when hybrid architectures were exploding in popularity. Yet the hybrid cloud ecosystem has changed tremendously in just the past few years--so much so, in fact, that today’s hybrid clouds are fundamentally different from their predecessors. A hybrid cloud created today looks almost nothing like what passed as hybrid cloud for most of the 2010s.
Here’s a look at the new hybrid cloud definition, and why organizations should rethink their hybrid cloud strategies.
The First Generation of Hybrid Cloud
Hybrid clouds have been around since the dawn of modern cloud computing in the mid-2000s. Theoretically, they were defined as architectures that used a central management interface to control workloads that ran partly on on-premises infrastructure and partly in a public cloud.
From the start, however, many organizations ignored the “central management interface” part of the hybrid cloud definition. They instead assumed that if you used on-premises resources and a public cloud at the same time in some way, you had a hybrid cloud environment--even if you lacked a centralized way to manage both parts of your hybrid cloud architecture.
For this reason, as a 451 Research analyst explained to one journalist in 2014, “A true hybrid cloud is rare, and only about one in 10 enterprises have it.”
Changing the Hybrid Cloud Definition
Today, however, true hybrid clouds are becoming much less rare. This is thanks to several new trends and technologies that make it much easier to manage workloads seamlessly across on-premises and cloud-based infrastructure:
One important trend is the explosive popularity of multicloud architectures.
Arguably, the hybrid cloud model is a form of multicloud. In that sense, the multicloud trend hasn’t encouraged more hybrid cloud adoption; if anything, it has pushed more organizations to use multiple public clouds at the same time, rather than building more hybrid architectures that combine on-premises resources with a single public cloud.
Yet what the multicloud shift has done is encourage cloud architects and developers to think in more cloud-agnostic terms. To deploy multicloud workloads efficiently, it’s important to avoid being locked in to a specific cloud vendor’s services or tools, so that you can port them from one public cloud environment to another with ease.
If you design your workloads with this goal in mind, it typically becomes easier to move workloads between on-premises infrastructures and the public cloud, too. Multicloud has made hybrid strategies more practical from this viewpoint.
Another change that has radically changed the hybrid cloud landscape is the soaring popularity of Kubernetes.
You can deploy Kubernetes anywhere--on-premise, in the public cloud or even on a Raspberry Pi, if you really want. It’s by no means a technology specific to hybrid clouds.
However, Kubernetes has made it easier for organizations to implement hybrid cloud architectures by providing a central platform that can span on-premises and cloud-based infrastructures.
This isn’t a primary use case for most Kubernetes distributions. But it is for some, including Rancher and Anthos, both of which use Kubernetes (along with some other tools) to make it easy to unify hybrid cloud workloads. (As a side note, I’m not sure Anthos counts as a Kubernetes distribution, exactly, but it’s close to it.)
3. Azure Stack and AWS Outposts
Anthos is Google Cloud’s answer to similar hybrid cloud frameworks that were released by the other two Big Three clouds: Azure Stack and AWS Outposts.
The latter two frameworks are different from Anthos in many key respects. They are tied to specific cloud platforms (specifically, Microsoft Azure and Amazon Web Services), whereas Anthos can work with any public cloud service. Azure Stack and Outposts also require certain types of hardware for building the on-premises components of hybrid clouds.
Still, like Anthos, Azure Stack and Outposts--all three of which became available just in the past few years--have greatly simplified the effort required to build a hybrid cloud in which on-premises and cloud resources are seamlessly integrated, using a consistent set of APIs and management tools.
The Modern Hybrid Cloud Definition
All of these changes have redefined the meaning of hybrid cloud.
Today’s hybrid clouds are characterized by much tighter integration between on-premises and cloud-based services. From the perspective of applications, there is often no difference at all, because the applications use the same APIs no matter where they are running within a hybrid cloud. At the same time, developers and IT teams can use the same deployment and management interfaces for all parts of their hybrid clouds.
In short, the bar has been set much higher for hybrid cloud architectures over the past few years. Gone are the days when you could plausibly claim to have a hybrid cloud simply by running some VMs in a public cloud and storing some data for them on-premises.
Modern hybrid clouds are built around infrastructure-agnostic development and deployment techniques, and they often rely on sophisticated architectural platforms, like Anthos or Azure Stack, that simply did not exist until the recent past.