Cloud Computing: Is the Cloud Green?

There’s no question that a single mammoth cloud data center is more energy efficient than the sum of the many in-house data centers that it can replace.  The cloud is more efficient because it can enable better utilization of resources and because of its size is able to focus on best practices.

Data centers today account from about 2 percent of all energy consumption, and that percentage has been growing.  Because of the migration to large aggregated cloud data centers, Pike Research estimates that by 2020 data center energy consumption will drop by 31 percent from the energy consumption levels of 2010, resulting in a savings of $16 billion in energy costs.  The report also forecasts that greenhouse gas (GHG) emissions from data centers should drop by 28 percent over the next 10 years.

Eric Woods, analyst at Pike Research, said “The growth of cloud computing will have a very significant positive effect on data center energy consumption. Software as a service, infrastructure as a service, and platform as a service are all inherently more efficient models than conventional alternatives, and their adoption will be one of the largest contributing factors to the greening of enterprise IT.”

The report is generally good news, but it opened up an opportunity for an overall criticism of the industry by Greenpeace analyst Gary Cook who said that “the sector is happy to talk about its potential to lower carbon emissions by 15 percent by 2020, but thus far, IT companies are still taking an incremental approach instead of providing transformative solutions at the scale and speed for which they are known. The corporate sector, and particularly IT, must work today, not tomorrow, to change the status quo and intervene at important junctures to speak up for strong climate and energy policy.”

Others, like Rod Tucker, a researcher at the University of Melbourne in Australia, found the research flawed.  The Pike report overlooks the energy involved in transferring the data back and forth through the cloud and the many routers involved in that data transfer.  Tucker found that when very large amounts of data are moved on a frequent basis to and from the cloud that the equation changes and in-house computing actually consumes less energy.

Factoring in data transfer to the calculation of energy consumption makes the problem of comparing efficiencies much more difficult because it requires being able to understand and measure various usage scenarios.  So, it seems like the cloud can be green, but no always.