What Containers and Kubernetes Mean for Colocation Data Centers

At first glance, containerization of applications -- one of the biggest IT trends of the past decade -- may appear to mean little for the colocation data center industry. Containers became popular because they simplify application deployment, while the hardware and data center infrastructure underneath stays the same.

Yet in several ways that may be less than obvious, containers have opened important opportunities for the colocation industry. Those opportunities could drive more organizations to migrate from the public cloud to colocation data centers while also increasing competition between colocation providers.

Containers in a Nutshell

Software platforms like Docker make it possible to deploy applications inside environments that are isolated from each other but that don’t require traditional, VMware-style virtualization. (Although there’s a strong case for running containers inside virtual machines.)

Over the past decade, containers (and complementary tools like Kubernetes, which helps manage containerized applications) have become massively popular for a variety of reasons. They make it possible to move applications from one server to another without having to reconfigure or rebuild the application. They consume fewer hardware resources than virtualization hypervisors. They start in just a few seconds, whereas conventional bare-metal or virtual machine environments could take a few minutes to boot. They are built mostly with open source technology, reducing acquisition costs and lock-in concerns.

How Containers Change Colocation

Again, it’s perhaps not immediately obvious what any of this has to do with colocation in particular. Sure, you could deploy a containerized application in a colocation data center, just as you could deploy it anywhere else. But why should the colocation industry care?

Well, for several reasons.

From Public Cloud to Colocation

Perhaps most important, containers promise to make it easier for companies to move workloads between public clouds and other data centers.

That’s because, as noted above, a containerized application can typically move seamlessly from one server to another, with little reconfiguration required. This makes containerized workloads different in yet another big way from those that run in virtual machines or on bare metal.

If you are running VMs or bare-metal servers in a public cloud and want to migrate to a colocation center, you’ll need to rebuild a lot of your environment from scratch. You can’t just drag-and-drop an AWS EC2 environment onto a colocated server. This requirement creates tremendous inertia for companies that already have a VM-based infrastructure up and running in the cloud.

But if your applications that run in the public cloud (or, for that matter, in a private or a hybrid cloud) are containerized, you can move them to a colocation data center in a matter of weeks, if that. You can also move them back to the cloud, if you want, just as easily.

For this reason, containers are likely to accelerate the trend of companies moving workloads between clouds and colocation centers, while lessening the level of commitment that comes with deploying applications on one type of platform or the other.

Private and Hybrid Clouds on Colocated Infrastructure

Along similar lines, containers and management tools like Kubernetes simplify the work of building scalable, service-based private or hybrid clouds that are hosted on colocated infrastructure.

In the past, if you wanted to turn your colocated infrastructure into a private or hybrid cloud, you needed a platform like OpenStack or vCloud, which are not exactly easy to scale or to integrate with public cloud services. With containers and Kubernetes, however, you enjoy enormous scalability by default.

If you want to build a hybrid architecture, containers enable you to run the same technologies on your private, colocated infrastructure as those running in the public cloud, given that all of the public cloud providers offer full-fledged support for Docker and Kubernetes. (You certainly can’t say the same about OpenStack or vCloud.) This makes it easier to interface between public and private services in order to build a hybrid cloud that relies in part on colocated infrastructure.

Getting More Out of Colocated Hardware

A third benefit of containers for colocation customers is that containers promise to deliver more bang for your buck, in the sense that they offer more efficient utilization of hardware resources. In other words, because containers don’t impose the overhead of virtualization, they allow you to run more applications on the same server, at the same cost.

To be sure, the performance efficiency of containers relative to VMs is limited. It’s not as if containers will cut colocation costs in half; at best, they might reduce it by ten or fifteen percent in extreme cases, and less than five percent in typical cases. Still, that’s not nothing, and it makes colocation that much more attractive.

Moving Between Colocation Centers

I mentioned above that containers make it easier to move between the public cloud and colocation centers. They do the same thing for migrating from one colocation provider to another.

In this respect, containers are poised to encourage more competition between colocation companies. When customers can pack up and move to another provider with little effort, colocation vendors will need to offer lower pricing, broader geographic reach and higher reliability in order to stay at the top of the market.

Conclusion

It would be an overstatement to say that containers are revolutionizing the colocation industry. Overall, containers and the “cloud-native” turn that they have helped to inaugurate won’t disrupt colocation as fundamentally as they have other sectors of the IT industry. Nonetheless, containers will chip away at certain facets of colocation by bringing greater portability, cost-efficiency and competition to colocation centers.