Cloudonomics: The Business Value of Cloud Computing, + Website

CHAPTER 14

Money Value of Time

If forecasting the future is fallible, what is the alternative? Although it’s often advised that it’s better to be predictive and proactive than reactive, perhaps this wisdom needs to be turned on its head. If we can react instantaneously to any condition with as many—or as few—on-demand resources as we need, forecast error is irrelevant.

To the extent that exactly the right resources are available at exactly the right time, unpredictable variability becomes a nonissue. It is clear that this is the best possible solution and one that, increasingly, the cloud is ready to handle.

However, suppose that instead of instantaneous dynamic response, there is a delay. How much does such a delay cost? The answer is: It depends.1 It depends on a variety of factors, such as the penalty associated with insufficient resources and that associated with excess resources, as well as the degree to which the demand is unpredictably volatile. It also depends on the interaction between customers (or users) and the service process: If there are insufficient resources, is the revenue (or productivity) lost forever, or are transactions queued until resources are available. If they queue up, what is the order in which customers are served? At a high level, if the demand doesn’t vary very much, being slow to respond isn’t that critical: Even a sloth can catch a snail. If application demand is highly volatile, the resource-provisioning function needs to be equally agile: It ...