The company increased its computer server performance per megawatt-hour of power used by a factor of 2.1 over the period from June 2007 through June 2010.
22 Mar 2011 -- In May 2007, IBM Global Technology Services stated that it "expects to double the computing capacity of its data centers within the next three years without increasing power consumption and associated carbon dioxide emissions". The company's analysis of this metric has been completed and IBM met its goal, increasing computer server performance per megawatt-hour (MWh) of power used by a factor of 2.1 over the period from June 2007 through June 2010.
In determining its results against this goal*, IBM evaluated the server inventory it manages for its own operations and for clients, which consisted of over 125,000 machines by 2010. These machines are housed in the extensive raised floor data center space owned or leased and operated by IBM Global Technology Services.
IBM's estimated average server performance/power metric in June 2007 was 200 RPE2 per MWh, where RPE2 stands for relative performance estimate, a measure of server computing capacity determined by IDEAS International. In June 2010, the average server performance/power metric was 426 RPE2 per MWh -- slightly exceeding the goal of a two-fold increase in energy efficiency.
IBM realized this increase in computing capacity per unit of energy consumed through the following:
- Replacing older server equipment with new equipment with higher performance per unit of power consumed.
- Virtualizing applications and consolidating them onto fewer servers with higher power performance capabilities. This resulted in the reduction of the number of servers required to perform a given workload, and the removal of older, less efficient servers.
- Improving the cooling efficiency in data centers, resulting in a reduction in the energy required to cool and operate the data centers.
IBM's achievement of its goal also provides significant business benefit to the company. One example is the IBM Lexington Data Center1, where all three of the above methods delivered significant savings:
In addition, the Lexington project provides an illustrative example of the benefits our clients may achieve through similar actions.
*Note: Not included in this goal was equipment and space used to provide back-up (as opposed to routine) operational support known as business continuity and resiliency services. This computing capacity goal also did not cover storage and networking equipment.