There's been tons written by pundits about the cloud recently, but I haven't seen any significant in-depth analysis of how implementing compute clouds is integrated with IT Operations. IT OPS is the "guts" of how IT operates day-to-day processes, configurations, changes, additions and problem resolutions. The most popular reference to these processes is ITSM (IT Service Management), and the most popular guide to managing the processes is ITIL (the IT Information Library, v3). Not everyone uses (or even believes in) ITIL, and indeed, it's not required. But it is a convenient way to look at the possible methods/processes IT Ops can bring-to-bear to manage the size and complexity of today's data centers.
Obviously, if you're outsourcing your IT to a Software-as-a-Service provider, you've already obviated most ITSM issues. Somebody else is managing infrastructure for you.
But if you're running your own software in a cloud (say, Amazon EC2) you'd still probably worry about how tomanage & change software configurations; how security is administered; and how new versions are deployed. The only real processes eliminated are those dealing with the hardware -- you still have the software and data management processes to deal with.
Now, if you're operating your own cloud (say you're a hosted services provider, or building an "internal cloud" within your data center) there are still a number of processes to manage -- but also, a number that are conveniently automated or eliminated.
For example, if you look at the 'Service operation" block above, things like Event Management or Problem Management are conveniently automated (if not eliminated) by the "self-healing" aspects of most cloud computing (really utility computing) policy & orchestration engines. Similarly, in the "service design" block, things like capacity management and service level management are similarly automated, and don't require a traditional paper policy.
Consider the types of processes that would be impacted with the use of a truly "elastic" and "self-healing" cloud: "trouble tickets" would be opened and closed automatically and within seconds or minutes. Problem managent would essentially take care of itself. Service levels would be automated. Configurations would be machine-tracked and machine-verified. Indeed, most of the complexity that ITIL was designed to help manage, would be handled by computer, the way complex systems ought to be.
One other quick observation: in a cloud environment, where resources are dynamically and continuously shifted and repurposed, the Configuration Management System (usually a relational database) becomes "real-time", that is, it could change minute-by-minute, instead of daily or weekly, as-is the case with most current CMDB systems.
At any rate, I'd really like to see more in-depth analysis from the IT Ops and/or analyst community to dissect how ITSM is impacted as more IT staffs turn to, or implement, cloud-style automated infrastructures. This way, we can also get out of being "cloud idealists" and become "cloud pragmatists."