As I've noted a number of times, cloud computing promises both greater IT agility and reduced costs. No one disputes the agility issue. Compared to traditional resource deployment that can drag on for weeks or months, cloud computing offers the enormous improvement of a timetable measured in minutes.
On the subject of cost, however, there is no consensus. I discussed the subject a couple of weeks ago here. The other perspective made itself known last week with this opinion piece in InformationWeek. The author evaluated the historical price changes in AWS S3 storage vs. on-premises, disk-based storage and concluded that AWS isn't cost competitive. Moreover, he asserted that AWS storage isn't realizing (or at least, isn't passing on) the benefits of Moore's Law, since its prices aren't dropping as rapidly as on-premises storage pricing.
I don't want to get into a debate about the article, as I believe there's ample evidence that, for a broad range of uses, cloud storage is more cost-effective than on-premises alternatives. One has only to look at the enormous growth of Amazon S3 to confirm that a very, very large number of organizations have come to the same conclusion.
The CFO View of Clouds
However, the question of cloud pricing is an interesting one. The reason for this is clear: While agility is primarily an IT-focused topic, pricing brings in financial analysis, the province of the CFO. How CFOs evaluate cloud pricing will have a large impact on adoption. If CFOs are convinced that cloud computing offers financial benefits, one can expect that there will be significant impetus toward adoption, particularly since, in many companies, the CIO reports to the CFO.
Today, CFOs have a love/hate relationship with cloud computing. With time, their stance will inevitably shift toward a preference for cloud computing and away from on-premises installations such as those described in the InformationWeek column.
Why CFOs Hate the Cloud
Let's start with the hate part of the equation, as it's relatively easy to understand. Cloud computing services that are priced according to resource consumption carry costs that vary over time. If you use a lot of resources one month, the bill will be higher than the bill for another month when resource use was light. Variable, unpredictable pricing messes up financial forecasting, including the all-important cash flow projections. CFOs are instinctively uncomfortable with financial commitments that cannot be estimated ahead of time.
One way to solve this, of course, is to purchase what are often referred to as "external private cloud" services. In this approach, the customer commits to a certain level of resource availability from a cloud provider. This offers the benefit of a fixed cost. On the other hand, this model also poses the problem of requiring a forecast of resource use, with the associated risk of under- or over-provisioning. Amazon offers a variant of this approach with its reserved instances -- by pre-paying a fee upfront, the buyer obtains a lower cost per resource unit used. As an example, by pre-paying $69 per year, one can obtain a small virtual machine price of 3.9 cents per hour, a 51 percent savings compared to the standard eights cents per hour for an on-demand machine.
Sign up for MIS Asia eNewsletters.