When private clouds first emerged more than a decade ago, there was essentially just one deployment model: a DIY deployment approach where companies set up private cloud services themselves on their own infrastructure.
Today, the evolution of the private cloud ecosystem offers a much richer set of deployment options. They range from the DIY approach to a fully-managed, turnkey deployment strategy, with some other deployment strategies falling in between those two options.
This article examines the deployment strategies that are possible for modern private clouds, and evaluates the pros and cons of each.
1Self-Service Private Cloud Deployment (i.e., the DIY approach)
It’s still certainly possible to use a DIY private cloud deployment strategy. (The more formal term for this approach would be to call it a self-service option.) Under this approach, you take private cloud software and deploy and manage it on your own infrastructure.
This deployment strategy gives users the highest degree of control, and it could potentially save money – if (and this is a big ‘if’) a team truly has the in-house expertise it needs to deploy and manage a private cloud on its own.
Generally speaking, however, a self-service deployment model comes with several drawbacks:
- Because the private cloud ecosystem now relies heavily on open source platforms like OpenStack (which is different from a decade ago, when proprietary private cloud frameworks were more common), there is no official support channel that self-service deployment teams can turn to when they run into issues.
- The scale of today’s clouds is larger than ever, making a DIY deployment harder to manage at the scale that companies typically need.
- In almost all cases, self-service deployments take a long time to complete: At least six months, as a rule of thumb. That delay comes with a considerable productivity loss, which may not be enough to offset the cost savings of a self-service deployment model.
- Even if a team has the expertise required to deploy a private cloud on its own today, the ever-changing nature of the cloud ecosystem means that it may not have the skills it needs to integrate whichever platforms or tools appear tomorrow.
- Why and How to Run Machine Learning Workloads on Kubernetes - July 23, 2021
- Learning About Kubernetes Admission Controllers and OPA Gatekeeper - July 20, 2021
- Fargate vs. Kubernetes: Which Container Orchestration Solution Do You Need? - July 19, 2021
Overall, these drawbacks make self-service deployment a poor approach for most private clouds today.
2Hybrid Self-Service Deployment Using Public Cloud Services
The appearance of new hybrid and multi-cloud frameworks – like Azure Stack, Azure Arc and Google Anthos – has enabled a new take on the self-service private cloud deployment model. Today, you could take one of these frameworks and use it to extend the control plane of a public cloud into your own data center in order to build your private cloud.
This is very different from deploying an open source platform like OpenStack to build a private cloud; because in this approach, you depend centrally on a public cloud to make your private cloud tick. There is thus a degree of lock-in at play (which may be lesser or greater, depending on which specific framework you use: Azure Stack will lock you in more than Anthos, for example, because Azure Stack works only with Azure, whereas Anthos works with any public cloud provider).
The upside of this approach, however, is that it requires somewhat less effort and expertise than a traditional DIY deployment. The vendors behind modern hybrid and multi-cloud cloud frameworks provide support services. They don’t offer fully-managed clouds – that is a different deployment model, as we’ll discuss below – but they don’t leave you on your own as much as you would be if you deployed a platform like OpenStack, without any outside assistance.
3Fully-Managed Private Cloud Deployments
A third deployment option is to take advantage of fully-managed services to deploy private clouds.
There are actually different subcategories within this overarching approach to deployment.
Fully-managed hybrid cloud
One subcategory involves using a fully-managed hybrid cloud service, like AWS Outposts. This is similar to the hybrid self-service deployment model described above, with the exception that Outposts provides a turnkey solution for deploying a private cloud using AWS services.
Managed Kubernetes services in public clouds
Another subcategory is to deploy a private cloud using a managed public cloud service like Amazon EKS or Azure Kubernetes Service. With this approach, you can build a private cloud using a relatively painless and almost turnkey deployment process. In most cases, you’ll be limited to hosting your private cloud on public cloud infrastructure, although in some cases you can use on-premises infrastructure as well.
The major drawback in all cases, however, is that currently, the managed services that offer this type of deployment option are based on Kubernetes. That means they’re useful only if you are able to deploy your applications using containers and work within the Kubernetes ecosystem. The big-name public cloud providers don’t offer managed services for deploying a private cloud framework like OpenStack.
Third-party management services
The third subcategory of fully-managed private cloud deployment options is to adopt a service like Platform9’s, which lets you deploy and manage Kubernetes or OpenStack on virtually any private infrastructure – your own data center, a public cloud or both at the same time – without the hassle of having to set up and manage the deployment yourself.
This option gives you greater flexibility and protection from lock-in, because you are not tied to one public cloud platform or one private cloud platform. It also saves you from having to master the private cloud ecosystem in order to keep abreast of the latest developments and integrations. With a managed service, you can deploy the newest private cloud technologies without having to know much about how they work at a low level.
There are a number of different ways to deploy private clouds today. The major differences between them lie in how much effort and expertise they require on the part of users, as well as how much flexibility they offer (in terms of the infrastructure, private cloud platforms and cloud services you can use).
Each option has its own advantages and disadvantages. But generally speaking, if flexibility and freedom from lock-in are priorities, a managed private cloud service offers the greatest value.