William Clifford, 10.07.10, 01:00 PM EDT (courtesy, www.forbes.com http://www.forbes.com/2010/10/07/energy-management-electricity-technology-cio-network-cloud-computing.html?partner=technology_newsletter)
If data centers aren't optimized for efficiency, cloud computing can amplify an energy problem.
Over the past year cloud computing has boomed into a $16.5 billion market. Revenue is projected to grow at an annual rate of 27% over the next four years. Two of the many advantages businesses see in cloud computing are the ability to cut costs and to reduce energy consumption by limiting in-house data center operations.
That said, The Wall Street Journal recently noted that the number of servers deployed in cloud applications is expected to triple to 1.35 million within the same time period. While cloud computing benefits the way businesses share information, the data centers that power the cloud could become a major cause of climate change.
Below are ways that both cloud services companies and businesses looking to moving data and applications to the cloud can assure they are pursuing this mission in the most cost and energy efficient way.Why? Because if they are not optimized for efficiency, then move the data and computing needs of numerous businesses into the data center of one cloud provider compounds the budget, resource and energy management issues that lead those businesses to adopt a cloud solution in the first place. Additionally, businesses that don't examine their own data center performance after migrating to the cloud (few businesses do a complete migration; most run a hybrid solution with some cloud-based and some on-premise data and applications) run the risk of not seeing the promised cost and productivity gains with cloud computing.
Right-Sizing the Data CenterAs mentioned above, few companies move to the cloud 100%. They move some data and applications but keep other things on-premise, especially sensitive information. However, as items are transitioned from on premise to the cloud, companies end up with idle servers which are no longer in use (or used to capacity) but still consume significant power and cooling. These "ghost servers" are used as development machines, storage servers for low access data, or servers dedicated to one high availability application. This happens at cloud data centers as well, unless they are effectively managed.
Ghost servers often run at 15% to 20% utilization or less, but consume power and emit heat at levels approaching a fully utilized server. By finding the optimum capacity, effectively balancing headroom with risk tolerance (service level requirements), companies can recover the 8% to 10% of resources (read: money) that are typically wasted.