Four Ways to Build a Colocation-Friendly Hybrid Cloud

If you own colocated infrastructure that you want to integrate into a hybrid cloud, there are more solutions available to you today than ever. The intersection between public clouds and colocation has grown so dynamic that simply keeping track of the different approaches can be a challenge.

Let’s walk through the main hybrid cloud strategies available to colocation data center users today, as well as the key vendors associated with each approach.

Building a Colocated Hybrid Cloud with Virtual Machines

Perhaps the most basic approach to integrating colocated infrastructure into a hybrid cloud is to host the workloads on virtual machines, then use a platform like VMware Cloud Foundation to manage them.

This approach may feel the most familiar to many users, because it allows them to keep their workloads inside VMs. It also provides the flexibility of being able to move between different public clouds at will, because the strategy is not dependent on any specific public cloud vendor's tooling.

The strategy does, however, make you dependent on the platform you use to orchestrate your hybrid architecture. It may also restrict you to using VMs to host your workloads and deprive you of the ability to use other types of hosting solutions, like serverless functions, although that is changing in the case of VMware’s hybrid cloud platform, which has now extended beyond VMs thanks to the Project Pacific initiative.

Related:Why AWS’s Hybrid Cloud Play Is Different, and What It Means for Colocation

For a deep dive into the various hybrid cloud strategies available today, be sure to sign up for the upcoming free DCK webinar, titled Hybrid Cloud, the Benefits of NOT Going All-In

Extending Public Cloud Services into Colocation

If you don't mind being dependent on a public cloud vendor, you can use a hybrid cloud framework like AWS Outposts or Azure Stack. These frameworks extend public cloud services directly into colocation facilities or on-premises enterprise data centers, which means you can use a service like AWS EC2 or Azure Functions to host workloads on your own servers.

These services do have their drawbacks. Outposts is particularly restrictive with regard to the types of hardware and cloud services you can use. Azure Stack is somewhat more flexible, but it's still evolving. These solutions may also be less attractive from a cost perspective, because they leave you with a public cloud bill on top of the fees you pay to colocation providers to host your infrastructure, not to mention whatever it costs you to purchase and manage your servers.

Expect this niche to keep growing and getting better, however. Cloud vendors beyond the Big Three are already investing in it (Oracle, for example, has a platform called Cloud@Customer that is similar to Outposts and Azure Stack). Microsoft is gradually rolling out Azure Arc, which is like Azure Stack but without the hardware -- or cloud platform -- restrictions. Google has Anthos.

Related:Why Hybrid Cloud and Colocation Don’t Always Gel

A Kubernetes-Based Hybrid Cloud

If you don't mind running most of your workloads inside containers, you can use Kubernetes to provision and manage a hybrid cloud that spans colocated servers and public cloud infrastructure.

This is the approach driving Anthos, Google's main hybrid cloud solution. It's also now a major part of VMware's cloud strategy. And I'd bet more than a little cash that we'll be seeing IBM making a Kubernetes-centric hybrid cloud play that derives from OpenShift, the Kubernetes platform that Big Blue acquired along with Red Hat. OpenShift is already hybrid friendly; all IBM has to do is figure out how to integrate OpenShift into the broader IBM cloud strategy.

Mimic Public Cloud APIs in Your Colocation Facility

Yet another approach to building a hybrid cloud with colocated infrastructure is to use a platform like Eucalyptus, which lets you use public cloud APIs to run workloads on your own servers.

Unlike frameworks like Outposts and Azure Stack, this type of solution doesn't bring public cloud vendors' own software into your data center. Instead, it essentially mimics their APIs so that you can deploy workloads that run just as they would in the public cloud -- indeed, you can lift-and-shift them into the public cloud virtually instantly, if you want -- but without becoming dependent on the public cloud.

The major limitations here are that only certain public clouds are supported (Eucalyptus works only with AWS, for example), and you're limited to using core IaaS services (mainly compute and storage) to run your workloads. On the other hand, this strategy frees you from the lock-in risks you'd face with a solution like Outposts. It may also be a lot less expensive, because you don't need to pay public cloud bills.

Conclusion

Just a few years ago options for building colo-friendly hybrid clouds were few and far between. You were essentially limited to VM-centric platforms, and even those did not usually integrate private and public data centers very tightly.

Today, however, a new generation of hybrid cloud solutions has changed the game for colocation users. No matter what your priority is -- avoiding public cloud lock-in, containerizing your workloads, minimizing costs, or something else -- there is a hybrid cloud platform available that will work with your colocated infrastructure.

For a deep dive into the various hybrid cloud strategies available today, be sure to sign up for the upcoming free DCK webinar, titled Hybrid Cloud, the Benefits of NOT Going All-In