“In his piece, “Cloud Computing Meets Energy Management,” William Clifford makes important points about the need to optimize the efficiency of both cloud data centers and on-premise computing. However, a new study released this week challenges his assertion that cloud computing “just transfers the consumption problem to another location.” The findings suggest instead that cloud computing can significantly reduce the overall net energy use of business computing needs.”
When small organizations (100 users) move to the cloud, the effective carbon footprint reduction could be up to a 90% savings by using a shared cloud environment instead of their own local servers
For large corporations, the savings are typically 30% or more. In a case study with a large consumer-goods company, the team calculated that 32% of energy use and resulting carbon emissions could be saved by moving 50,000 e-mail users in North America and Europe to Microsoft’s equivalent cloud offering.
What accounts for these significant energy savings? Think of cloud computing as being like mass transit. The data center is essentially getting computing applications to carpool or take the bus instead of sitting in their own individual servers. However, unlike mass transit, there is no sacrifice in convenience or performance with this move. Consider the disappointing fact that a typical server in a company often runs at about 10% of capacity, meaning there are lots of servers out there drawing power without doing much computing
The economies of scale of cloud data centers allow much higher utilization of servers, dynamic provisioning to better match server capacity to demand, and multi-tenancy to serve thousands of organizations with one set of shared infrastructure.
The efficiency benefits of the cloud won’t be realized unless customers are thoughtful about decommissioning or repurposing unused servers, and cloud providers like Microsoft continue to innovate in
the name of greater and greater efficiency.
For companies with their own large-scale infrastructure, this study identifies the key drivers that will let them optimize for the greatest efficiency as well.