I’ve spent 20 years designing, building, supporting, and talking about enterprise data center services. What I’m about to say is controversial to many people who depend on data center services as the engine of their career and livelihood. However, it must be said.
Without a doubt, 2017 is the year that private cloud died. It is not currently dying, it’s already dead.
Defining private cloud
Most of the pushback I’ve seen against the idea that private cloud died falls into two categories. The first is from vendors that provide managed private cloud, and the other is from examples of successful private cloud deployments at enterprises such as Adobe, Walmart, and GE. It’s therefore essential to define private cloud.
SEE: Cloud computing policy (Tech Pro Research)
Private cloud falls into the category of one of those things that’s difficult to pin down. I don’t believe we need to establish some technical definition of cloud–you know it when you see it. I think the simplest way to identify private cloud is to ask the consumers of the service if it meets their expectations of what the cloud should be. For better or worse, public cloud has set the standard for a cloud experience.
Automating provisioning of virtual machines and storage are starting points for modern infrastructure. Scale-out physical infrastructure design and operations also fall into the category of table stakes. A service management platform that enables and supports automation and scale-out operations are again, table stakes not only for private cloud, but all modern IT infrastructure. These attributes alone don’t indicate a cloud delivery model.
I asked Wikibon senior analyst Stu Miniman to define private cloud. Miniman referred to Wikibon’s definition of True Private Cloud (which you can read here) to set the stage. However, Miniman also agreed that public cloud is the measurement stick to which private cloud must compete.
What’s next in 2018
My belief is that most organizations have neither the technical resources nor processes to support managing a private cloud. To start what Wikibon calls a True Private Cloud today is a distraction from what brings real value to business users. There may be exceptions–such as managed private cloud, which eliminates many of my concerns–but it’s true of most private cloud efforts.
I would argue that organizations should focus on reducing the overhead needed to manage legacy IT infrastructure. Automation and DevOps practices help to reduce the resources required to maintain legacy operations. Those financial and personnel resources can then be redeployed to support hybrid cloud and cloud-native initiatives.
The reality remains that legacy data center components will continue to dominate IT. Enterprise IT operations are skilled in managing these resources, and net new resources most likely will be built using cloud-native technologies such as containers and public cloud.
Miniman pointed out that the value point for enterprise IT is in hybrid cloud. His view isn’t too dissimilar from mine. Miniman pointed out solutions like Microsoft AzureStack, VMware on AWS, and Redhat OpenShift as combinations that support hybrid.
My argument is that none of these solutions are actually private cloud platforms. In the case of AzureStack, it’s an on-premises public cloud solution, according to AzureStack chief architect Jeffrey Snover. VMware on AWS is legacy virtualization within the walls of AWS data centers, and Redhat OpenShift is an orchestration platform for containers. None of these solutions offer private cloud experiences.
These solutions do enable integration of public cloud and the private data center. I label this hybrid IT. Hybrid IT is essentially the combination of legacy infrastructure with cloud or multi-cloud solutions. The real challenge and value comes with the integration of legacy IT with cloud-native services, and should be a focus point for the enterprise moving into 2018.