Last week in the New York Times, Steve Lohr wrote an article, In Tough Times, Build Fewer Data Centers. He makes a compelling case that "The smartest, greenest, most efficient data center is the one that a company decides it doesn’t need to build." Building a new data center is the natural reaction to a need for increased processing power. It all starts when a new application is sold into an organization or a new division is spun up due to an acquisition, for example. The users demand more processing power and the business demands more out of its IT infrastructure. Thus begins the vicious cycle of server provisioning, data center construction, new hardware costs, new management costs, and increased power and cooling costs. What starts as a need for new features and functions can quickly turn into a $10M-100M+ data center project.
During times of IT budget reductions, this bolus of immense cost is hard to swallow. Fortunately there is solution that can create Data Center Cost Avoidance. Using virtualization, processing power can be harvested from existing systems to avoid building new data centers. Traditional servers are provisioned to run a single operating system and a small number of applications. Typically they use only a small fraction of their processing power. Under sever virtualization, multiple operating systems (maybe a dozen or more) are placed on a single server and collectively use more of the system’s capabilities. Reducing the number of servers by a ratio of 12:1 or greater has a tremendous impact on all the costs associated with IT infrastructure and avoids that new center.
However to be successful, the old standard response to new demands placed on the infrastructure need to change. Instead of the standard purchase order being generated for a new server, a new system is spun up on an existing server all at reduced costs and increased business agility.
Toby J. Velte, Ph.D.
Avanade Inc | Minneapolis
My Fast Company Profile
Co-author of Green IT: Reduce Your Information System's Environmental Impact While Adding to the Bottom Line.