When an On-Premises, Public Cloud Might be Right for You

You may think that the terms on-premises and public cloud are contradictory, but you'd be wrong. A fairly new alternative to both public clouds and private data centers is to incorporate the two.

Companies like HyperGrid will happily deploy their hyper-converged infrastructure hardware into your private data center, yet allow you to use the equipment with a "pay-as-you-go" pricing model. Think of it like using any of the major public cloud service providers such as AWS, Azure or Google Cloud -- but located in your private server room. This on-premises, public cloud approach makes for an interesting choice with its own unique set of benefits and drawbacks.

The protection and control of sensitive data is one popular reason why companies are unwilling to migrate to the cloud. Many IT security administrators simply don't trust their data leaving the protection of a privately-operated infrastructure. However, those concerns are largely eliminated if the cloud provider were to deploy a hyper converged solution inside a data center -- that is managed and secured by the customer. That way, the business can benefit from a cloud consumption model, while operating in a friendly and trusted environment. For those that want the utmost in security and governance, this might be the way to go.

Bandwidth and latency constraints to public cloud resources is another issue that is solved using an on-premises, public cloud deployment model. Public cloud deployments of services that are latency sensitive -- or consume large amounts of bandwidth can be a constrain on an enterprise's Internet and WAN links. A commonly used example would be virtual desktop infrastructure (VDI) deployments. To reduce latency and eliminate potential bottlenecks caused by hundreds or thousands of VDI sessions across links to public clouds, you now have the option of deploying them in an on-premises solution. All while gaining the benefit of not having to manage the underlying compute, storage and network infrastructure components.

Clearly, there are cases where an on-premises, public cloud solution makes sense. Yet, this architecture is not without its own set of pitfalls. For example, in order for any service provider to actually deploy equipment inside your data center, you’d better have a fairly large project ready to go on day one. The claim of on-premises public cloud providers, such as HyperGrid, is that there is no up-front cost and you use the same pay-as-you-go model that all other public cloud providers use. And while this might be true, they’re going to want you to start paying for services as soon as it’s deployed.

If you do have an immediate need for a hyper-converged system on-premises, yet you don't want to manage it, you do have options other than an on-premises, public cloud provider. One method would be to simply buy your own hyper-converged system, then have a third party managed service provider (MSP) deploy and administrate it for you. While you would incur upfront CAPEX costs with the purchase of hardware/software and licensing, there are plenty of multi-year financing options available from resellers. And ultimately, this may end up being the lower-cost option.

Another thing to consider when weighing the pros and cons of an on-premises solution is the fact that you’re still going to have to manage and pay the costs of running it in your data center – or pay to rent colocation space. Everything from electricity to cooling and the data center’s underlying infrastructure must be maintained to keep the service providers gear operating properly. These costs must be factored into the overall pay-as-you-go payments to the service provider. Once you factor both expenses, you can determine the true cost of ownership for the solution.

Ultimately, an on-premises, public cloud isn’t going to be right for everyone. But there will be a niche group of enterprise companies that have the need to maintain applications and data in-house, yet don’t want to bother with the management of the underlying server, storage and networking infrastructure components. And if you believe you have a project that fits those characteristics, then it’s nice to know the option is now available.