Once upon a time, the major public cloud providers mostly ignored colocation users. There wasn’t an AWS hybrid cloud solution, or an Azure one. If you wanted to integrate public cloud services with workloads hosted on your own servers inside a colocation data center, you were left to figure out how to do that on your own, using third-party solutions.
That has changed drastically over the past several years thanks to the rollout of new hybrid cloud platforms that make it much easier to connect public cloud services to colocated infrastructure. Exactly how the platforms do that, however, and which features they offer, vary significantly from one public cloud to the next.
Colocation and Public Cloud, a Brief History
Until circa 2018, all of the big-name public cloud providers -- Amazon Web Services, Azure, and Google Cloud Platform -- paid little attention to use cases that required users to consume public cloud services in some way while keeping other parts of their workloads inside a colocation data center.
That was understandable. For most of their history, the public cloud vendors wanted to get as many customers as possible to migrate wholesale to the public cloud. Hybrid cloud solutions that keep some resources on private hardware mean less IaaS revenue for public cloud providers.
That’s not to say that it was totally impossible to incorporate their public cloud services into a hybrid architecture. Customers could use open source platforms like Eucalyptus to achieve this goal. VMware also began offering some hybrid solutions before the public clouds themselves jumped on this bandwagon. And for their part, the cloud providers did at least offer interconnect services that boosted network performance for use cases where customers bridged colocated infrastructure with the public cloud. (Microsoft also began offering a hybrid solution called Azure Pack in the earlier 2010s, but it never really caught on.)
Then, in the span of barely a year, the public cloud providers suddenly revealed a lineup of hybrid cloud solutions that make it much easier for their existing customers to integrate their public cloud environments with colocated hardware -- and for colocation users who are not yet in the public cloud to start consuming some public cloud services.
Those included an AWS hybrid cloud solution, Outposts, announced in 2018; a Google hybrid cloud solution, Anthos, announced in 2019, and two Microsoft Azure hybrid cloud solutions, Azure Stack and Azure Arc, which the company announced in 2016 and 2019, respectively.
Comparing Colocation Offerings from Public Cloud Providers
All of these platforms do the same basic thing: They make it possible to deploy and manage public cloud services on servers running in a colocation facility or on premises. But they have different strengths and cater to different use cases.
AWS Outposts, the AWS hybrid cloud solution, is probably the most restrictive of the hybrid platforms. It requires servers purchased from Amazon itself, and it doesn’t work with third-party clouds, just private hardware.
Amazon’s recent announcements at re:Invent 2020 of EKS Anywhere and ECS Anywhere, which extend the AWS container services to data centers without requiring Outposts, gave colocation users a couple additional ways to run AWS services in colocation facilities. But the options are still quite limited overall.
Given that AWS dominates the public cloud computing market, the limitations of its hybrid solutions make sense. AWS has the least to gain by making it easier for customers to run some workloads outside of AWS data centers.
Azure Stack and Azure Arc
Azure Stack, the Azure hybrid cloud solution, is also a relatively restrictive platform. It requires certified hardware, although Stack-compatible servers are available from a variety of vendors. It also works only with private hardware, not other public clouds.
Azure Arc, on the other hand, is more flexible. It extends the Azure control plane to basically any type of infrastructure, which means colocation customers can manage their data centers using the same tools, and configure them with the same workloads, as they could in the Azure cloud itself.
For now, Azure Arc is still being rolled out. When it’s complete, however, it’s likely to place Microsoft in a dominant position within the hybrid cloud market.
Google’s hybrid cloud solution, Anthos, uses Kubernetes, combined with some other open source tooling, to allow users to deploy applications anywhere and manage them via GCP (specifically, via GKE, GCP’s Kubernetes service).
Like Azure Arc, Anthos supports multiple public clouds and basically any private data center. It also works with any modern hardware: Users can bring the servers they already own, or purchase new ones.
The only major caveat with Anthos is that it’s Kubernetes-oriented. For colocation customers who haven’t yet jumped on the K8s bandwagon, Anthos may be less attractive than a platform like Azure Arc. On the other hand, for those who want to update their legacy workloads and go all-in on cloud-native, Anthos offers an easy path for getting there.
IBM, which is to the Big Three Cloud providers what Charles de Gaulle was to the World War II-era Big Three, has been making a major hybrid cloud play of its own following its acquisition in 2018 of Red Hat.
OpenShift, the Red Hat Kubernetes distribution, is the foundation of IBM’s hybrid cloud solutions. That places IBM and OpenShift in a similar position to Google and Anthos. Perhaps the big difference is that Big Blue markets its hybrid offers as more customizable than those of the bigger-name public cloud providers. That may be a selling point for colocation customers who have large-scale infrastructure and need specialized configurations or services in order to get some of their workloads into a hybrid environment.
In a major change from just a few years ago, there is now a plethora of options for colocation users who want to connect their colocated workloads to public clouds. Which solution is best depends on whether you want multicloud flexibility or just a hybrid cloud, as well as which hardware you want to use and which cloud services you need.