According to IBM's cloud blog (www.ibm.com), cloud computing has been around since the 1950s, when computer service companies purchased mainframe computers and offered computing services to schools and small businesses. Virtual machine capabilites were introduced in the 1970s, allowing mainframe systems to have multiple virtual systems on a single physical node. The introduction of the hypervisor 20 years later allowed multiple operating systems to run alongside each other, sharing the same physical computing resources. With the addition of the internet, computer service companies were able to offer vast arrays of virtual computing capability saving users the costs of home grown data centers.

It seems only logical, then, that owners of home grown data centers would want to take advantage of these web services and abandon their home grown capabilities. But, experience has shown that transitioning older IT systems and software capabilities to these wonderfully efficient web-based virtual environments can be expensive, and may not return benefits sufficient to offset the cost of the transition within a reasonable time frame.

This paper presents a structured paradigm of cost categories and activities to consider as organizations plan for transition of home grown compute and storage services to commercial web-based services. We discuss methods to model the cost drivers and processes to transition compute services and migrate associated data, and then optimze operations costs for cloud Infrastructure, Platform capabilities and shared Software. We also discuss our research and lessons learned as we developed models for estimating cloud costs using real world examples.