One of the key benefits cloud computing promises to deliver is that of flexible, usage-based pricing for IT resources: use more, pay more; use less, pay less. This model gives businesses transparency into actual consumption and eliminates – at least in theory – the problem of managing insufficient or idle capacity. The result is optimized efficiency and a more agile enterprise.
The problem is, cloud pricing models aren’t yet capable of measuring IT usage in a way that’s granular enough to bill for it. Consider, for example, CPU minutes, a basic pricing unit for cloud services. Unless they’re running a mainframe, most internal and external IT groups today don’t have the monitoring tools to measure and log CPU minutes. Moreover, a CPU minute on one machine can represent more or less processing than a CPU minute on another machine, so normalization is a challenge as well.
One potential solution is to charge for access time to a server “instance,” regardless of how many CPU minutes are required or used. While this model provides some of the benefits of increased transparency and improved demand management, it still isn’t really a usage-based model, because an “instance” can be defined in the agreement to mean 750 hours of server time, and that turns out to be a full month. The “size” of the instance is just linked to a specific hardware configuration, so although you seem to be paying for hours, you’re really buying access to a specific server configuration for a month, which isn’t all that different from what we had before.
Another issue is marketing hype. Since there are no real rules about how some of the terminology is used, one cloud vendor may use the same words to describe interesting new ways to deliver services that another uses to simply slap a cloud label on a traditional delivery mechanism.
Insight into the details of pricing and service agreements is essential, as is an apples-to-apples comparison of cloud options. CIOs need to understand the true costs that will be associated with different pricing models, the different situations in which one or another scheme is more appropriate, and how their choices will support or undermine their overall IT and business strategy.
Ultimately, if you don’t account for the full range of business and technical requirements posed by the enterprise, the cost of a cloud solution risks becoming “in addition to” rather than “instead of.”
For another perspective on cloud pricing, read this recent CST blog entry by Stanton Jones, CIO of TPI.