On-demand computing promises to dynamically allocate IT resources. But, deciding whose demands take priority will still be your call.
The buzz about "on-demand computing" is getting louder now that vendors such as EMC have joined other storage heavyweights (IBM being foremost among them) in bandying the phrase about. EMC recently announced its OpenScale metered billing system, which charges users based on actual storage usage rather than on capacity on the floor. At CA World in July, Computer Associates hammered home the on-demand theme in nearly every session.
In an on-demand computing scenario, server and storage resources for particular applications are automatically and dynamically provisioned as needed and then charged back to users. It's basically a "pay as you go" model. The charge back part of the process has been around for a while. Storage devices and software management tools have, for some time now, offered reporting and alerting tools that let companies know how much capacity a particular user, user group, department, or business unit is consuming - and then charge for it. But, the automated provisioning of processing power, network bandwidth, and storage capacity is just beginning to be woven into vendors' products. The degree of IT-wide, policy-based automation and reprovisioning described by on-demand visionaries like CA is still more "on the horizon" than "here." But, the horizon is not a distant one, and you'd be wise to begin considering what an on-demand IT model might mean for your organization.
Negotiate Before You Automate
As for the actual IT infrastructure - the hardware, the software, the network - on-demand computing will certainly affect your approach to purchases. CA claims, for instance, that one thing rendered nonessential in an on-demand environment is new hardware. Theoretically, you'll always be optimally using what you have and will only need new hardware if you are approaching 100% utilization across an enterprise that now needs to scale.
But, if systems-wide automation lives up to its promise, you won't be thinking much about the IT resources themselves. You will, however, have to think hard about how to build a charge back structure for the users of those resources. What on-demand computing can't do is automate the process of deciding which users and apps should receive priority access to resources. That requires human negotiation. It will still be up to your organization's various IT stakeholders to settle the territorial disputes that inevitably arise when departmental budgets and resource allocation policies are established. In determining who should be charged, for what IT resources, and how much, your organization will have to openly debate the business value of various users and units. For instance, should some departments be given charge back discounts based on the relative value they bring to the company? Try asking that question in a company-wide forum.