Properly Analyze the Value of the Cloud

Most analyses of the benefits of cloud computing are based on unrealistic total cost of ownership (TCO), return on investment (ROI), and capex/opex calculations -- most of which misunderstand the true value of the cloud to IT service delivery.

To fully understand the potential benefits of cloud computing, a new metric is required.

Both ROI and TCO are problematic in analyzing IT investments. They're inaccurate at best, and were really created to help IT executives justify projects of question value. If you track a single IT investment over time, you may be able to show a short-term positive return, but over three years or more, you're more likely to see either a loss or a break-even scenario.

Once you take into account the additional expenses required to maintain and operate any IT investment, you'll inevitably find costs increasing dramatically. Any attempt to offset these costs would require a costs reduction of 25 percent of more; and so very few CIOs will actually go back and reconcile initial TCO and ROI estimates against the original assertion over time. The result, put simply, is missed expectations.

It's crucial for IT executives to find a metric that allows them to estimate value for investment dollars as accurately as possible, and that metric may well be total service cost (TSC). It works like this:

(Cost of Infrastructure + Cost of Operations + Cost of Software + Cost of Risk) - Billed Usage = TSC

Cost of Infrastructure is always a computable amount. It may be more challenging for private cloud configurations, but it can be done by allocating and tracking percentages of the overall costs for the infrastructure used, including utilities, network, storage, and more.

Cost of Operations, similarly, must include all costs involved with service delivery, including information management, quality and assurance, and traditional network, systems and data center operations. In looking at Cost of Operations, service oriented architecture (SOA) serves as a key design paradigm; services are made of other services. Packaging each service simplifies the costing model for the service as a whole, allowing you to amortize the cost of the operational service across all services that are dependent on them.

Cost of Software is affected by the method you use to calculate Cost of Operations. If you include help desk software and network monitoring software as part of your Cost of Operations, you won't include them here. If, on the other hand, you put all software costs in this bucket, you'll need to track a percentage of all software used to deliver a particular service. The former approach is probably easier to calculate than the latter, as it only requires that you track new licenses, renewed licenses, training, etc., for a particular service. Service boundary modeling can be of value here. If two services rely on the same software, you can build and package a service around that software and centralize the costs for it there.

Cost of Risk assumes that there's a measurable cost associated with your service being down or breached. If that occurs, your overall costs can skyrocket, wiping out any other benefits you may have realized. While including Cost of Risk can vastly increase your TSC, it's more than worthwhile as it will greatly improve its accuracy over time.

Billed Usage focuses on the expected value of the service to the business itself. You'll have to work with marketing, sales and accounting in addition to IT, which can also help others better understand what IT does for the business, to determine an appropriate unit of billing. And while billing may sometimes make use of a chargeback mechanism -- a.k.a. "funny money" -- this metric is crucial. If your company were to consider outsourcing particular services, this metric allows executives to compare the costs of providing the service internally to the pricing of the outsourced service. The metric will also help you understand the degree to which the service is valued and used by the business.

If your infrastructure comes from a public cloud provider, you can count on fixed costs per unit of CPU, storage and memory, but you'll still have to get an accurate estimate of your usage over a given time period to calculate your TSC. And since the cloud can be hit with DDoS and other attacks, that usage (and, therefore, those costs) can increase unexpectedly.

Be sure to include the salaries of all personnel monitoring and running the service in your Cost of Operations, noting of course that those services can also be outsourced. If you do choose to outsource operational services, though, note that your low startup costs, low development costs and low management costs may be offset by a much higher Cost of Risk than you would face if you were to host the service in a co-located facility and use your own employees to run it.

Ultimately, keep in mind that calculations like capex/opex, TCO and ROI are unreliable ways to understand the costs associated with cloud computing.

TSC, while rudimentary, is a far better way to cost and analyze your cloud computing options.

This article originally appeared on Internet.com's CIO Update Web site.