CDP released a report last month estimating the energy and carbon emission impacts of a cloud computing architecture (public and private) versus dedicated IT. According to the model, a shift to cloud computing results in an estimated carbon reduction of 50% by 2020.
Where do these carbon (and energy) savings come from? They are a function of two parameters in the model that are assumed to vary across the three architectures: power usage effectiveness (PUE: ratio of total power used by data center to power used by IT) and server utilization rates. Note that for PUE, the lower the number the better, while the opposite is true for server utilization.
While it’s a good first start, it would be helpful to put some confidence bars around the assumptions and check the sensitivity of the final results to various scenarios (best case, worst case, etc.) Also, it’s not clear what the adoption scenarios look like and how, precisely, the PUE’s were forecast. Finally, it’s interesting to contrast these optimistic CDP estimates with what Greenpeace has said about cloud computing.