How to finance hybrid IT

Budget time is rarely fun in the enterprise IT world. Squeezing as much as possible out of a usually constrained budget when there's so much to do is enough to drive the most seasoned professional to despair. Yet, it must be done, and done well, to maintain resources that allow the business to flourish.

In years gone by, the juggling act between capital and operational expenses (CapEx and OpEx) was a constant. With all resources residing in the data center, IT had to invest scarce capital dollars just to keep the ship afloat, while still spending significant amounts on operations.

Then came the public cloud and the opportunity to shift expenses to the OpEx side of the ledger, which many chief financial officers prefer. However, that came with its own challenges, including the potential for skyrocketing operational costs as well as compliance and security requirements that public cloud providers could not always meet. These complications provided the impetus for the hybrid cloud.

In an article in IEEE's Cloud Computing journal, cloud economics guru Joe Weinman notes that hybrid clouds "can offer economic benefits, even when—in fact, particularly when—the unit cost of public cloud services and resources is higher than that of private dedicated resources, a scenario that some reasonably sized, well-run IT shops can and have achieved."

With hybrid cloud, the cost models change. There can still be capital requirements, depending on the state of your in-house infrastructure (it will likely need some software changes, at the very least). Vendors are now being creative in how they price these elements, allowing payments to scale to match business growth. But the public cloud component will be all OpEx.

Capacity on demand

One of the appealing factors of the cloud model is the ability to scale up and down quickly, spinning up and decommissioning resources as required. In theory, that controls costs as well, but as many companies have discovered, it doesn't always work that way. Without strict governance controls, including policies and automation, cloud resources that should have been turned off can remain active and end up adding to costs, to the point that the supposedly cheaper cloud costs significantly more than the in-house resources it replaced.

In a hybrid model, while this is also a risk, the public cloud is often used only for surge capacity to supplement in-house resources, at least initially. As cloud-native applications enter the mix, however, it's even more important that any financial plan involving hybrid cloud include governance controls to determine how public and private clouds can be used in the most cost-effective manner.

Private clouds can also take advantage of vendors' "pay as you grow" plans, in which users receive additional capacity from the outset and pay for it as they grow into it. One of the biggest annoyances for corporate IT is the length of time it takes to purchase and install additional capacity when required. Users want it yesterday, but the vendor may quote weeks or months before delivery. That disconnect in expectations often leads business units to put systems into the public cloud that shouldn't be there, simply because they don't have time to wait for the proper resources.

While missing a business opportunity is expensive, a precipitous leap into the public cloud can be even more costly in a number of ways. If regulatory and compliance requirements for an application aren't met, there could be fines. And if security isn't up to scratch, a data breach could cost millions, and potentially put the company out of business.

However, forecasting capacity requirements can be akin to a black art. IT (or the business) may know that it will need additional private cloud resources later in the buying cycle but can't justify the immediate expenditure when purchasing. That can lead to additional expense later.

Vendors have addressed these issues by introducing hybrid payment plans. The customer commits to a certain capacity, making fixed payments as with any equipment lease. However, the vendor also provides buffer capacity with usage metered and charged for separately. If the customer's usage bursts into the buffer, there's a bill. If not, there is no charge for the buffer that month. As the business grows and begins to consistently devour the existing buffer, additional buffer capacity can be installed as required, and the amount of committed capacity (and its fixed charge) increased. It's a cloud-like way to expand without putting sensitive data in the public cloud.

For less sensitive data, of course, the application can automatically burst to the public cloud.

Not just compute

While many concentrate on the data center aspects of hybrid cloud financing, you also need to factor in telecommunication costs. Data transfer to and from a public cloud isn't free, and while it's nowhere near as expensive as it used to be, public cloud usage still adds to overall corporate bandwidth requirements. In a 2011 paper published in the journal Information Systems Frontiers, researchers Oleksiy Mazhelis and Pasi Tyrväinen argued that "as the volume of data transferred to/from the public cloud increases, a greater portion of the capacity should be allocated to the private cloud."

Don't forget the people

The shift to hybrid cloud doesn't just impact hardware and software. There's an unavoidable people cost as well. Managing a public cloud vendor requires different skills than those for running a data center. IT staff may need training, and in fact, some may need to be replaced. 

Don't expect immediate returns

Moving to a hybrid cloud model doesn't come with instant returns. The upfront costs involved in "cloudifying" existing infrastructure, adapting applications for private or public cloud use, acquiring monitoring and management tools, training, and other expenses have to be considered.

How long will it take? Intel IT built a model of the projected costs for its own hybrid cloud initiatives in 2013, which showed that potential savings would only begin to manifest themselves in year two but would grow steadily after that. At the time, the authors noted, "The hybrid cost savings may vary from those shown, depending on the rate of cost-efficiency improvements in both the public cloud and in Intel’s private cloud. Public and private hosting costs are both decreasing rapidly; however, we believe that due to the operational efficiencies we are putting in place, including new technologies, training efficiencies, and use of open source solutions, hybrid costs will decrease as fast as or faster than public cloud costs. It should be noted that the efficiencies public providers have introduced into large-scale computing are a catalyst to help direct and lead the way for all computing—including private clouds."

While Intel's savings may not be achievable by all—its internal IT operation is extremely efficient, keeping its private cloud costs low—its model still indicates that a hybrid cloud can indeed save money while meeting the operational goals of the company. But to do so, the cost structure of the private cloud must be similar to that of the public cloud provider.

Financing hybrid IT: Lessons for leaders

• Keep a close eye on your operational expenses during cloud deployments.
• Have a plan and goals to determine the effectiveness of cloud and hybrid cloud transitions. If those goals aren't being met, find out why.
• Don't be afraid to de-cloud applications and services if they prove more expensive to operate in the new paradigm.