Project Big Green is IBM's sprawling initiative to increase the energy efficiency of IT. In May 2007, Big Blue announced that it would redirect no less than US$1 billion per year to Big Green, which applies both to solutions IBM offers to customers and to the company's own internal IT operations.
"When we made our commitment, it was top down and bottom up," says Rich Lechner, vice president of IBM's systems and technology group. "From the top, it started with Sam Palmisano. From the bottom up, a lot has come from employees inside IBM using Web collaboration tools. We have an energy-efficient community of 35,000 employees and their families. It all sort of bubbled up. We learned what they cared about."
For Lechner, the reality of Big Green is mainly about coordinating a constellation of IBM technologies to yield greater levels of energy efficiency, from blade servers to virtual machine management. He holds up virtualization and its inherent resource optimization as "the driver," but he also cites technologies associated with IBM's Autonomic Computing initiative, such as IBM's WorkLoad Manager, which helps datacenters self-optimize. As part of the Cool Blue portfolio, the Tivoli team has developed software that continually monitors energy usage across organizations so that energy consumption levels can become a standard component of runtime SLAs.
Greening on the inside
Inside IBM, the biggest green milestone has a "back to the future" gloss. In 2000, Lechner says, IBM began running Linux to the mainframe. In August of last year, the company took that idea to its logical extreme and started moving the workload of 3,900 of its own servers to 30 virtualized System z9 mainframes running Linux. IBM anticipates that it will cut energy consumption by 80 per cent, saving more than $2 million in energy costs.
It's not just about tech choices. As Lechner dryly notes, "IBM has quite a bit of experience in datacenter design." IBM now performs datacenter energy-efficiency assessments for customers, as well as special thermal analyses for high-density computing. Lechner cites experiments with recycling datacenter heat output, with one customer in Switzerland using that thermal surplus to heat a nearby public swimming pool.
Lechner's sharpest observation is that energy efficiency rises to the level of data and applications, which continue to grow exponentially. IT must have a strong commitment to data integration and service-oriented architecture to reduce redundancy -- and to stop increases on the demand side from canceling the benefits of datacenter efficiency.