As part of IBM’s cloud team, I spend a lot of time talking and writing about cloud infrastructure technologies and how organizations can benefit from a private cloud–whether it’s leveraging virtualization as an onramp to cloud or advancing the cloud with orchestration.
One topic I’ve been tracking that’s central to cloud management is how to control the costs of running the cloud, especially with respect to all the energy waste in data centers. IBM’s research estimates that up to 50 percent of an average air-cooled data center’s energy consumption and carbon footprint today is not caused by computing, but by powering the necessary cooling systems.
Recent studies by the New York Times provide additional data–that data centers expend as much as 90 percent of their energy consumption simply running at full capacity to ensure availability, even if they’re only using a fraction of that capacity at any given time.
And so, the energy waste in running data centers is still very much the dirty secret of cloud computing. IT demand has outpaced technology improvements. For companies like Amazon or Google, or any other cloud provider, ensuring the availability of their services currently outweighs all other considerations. The same is true for most large enterprises building intricate cloud systems, basing their business strategy on cloud as an enabler of speed and agility.
This makes initiatives such as SuperMUC all the more compelling. SuperMUC, the world’s fastest water-cooled supercomputer, is an IBM Research project that hopes to make a serious dent in the goal toward zero-emission data centers. In a nutshell, IBM researchers are using water cooling as an alternative to traditional data center cooling, a method that could reduce energy consumption by 40 percent while at the same time improving peak performance.
I talked with Dr. Bruno Michel, who manages Advanced Thermal Packaging at IBM’s Zurich Research Lab, about the impact of initiatives such as SuperMUC to data centers today. The technology cools active system components such as processors, absorbing heat to as much as 65 degrees Celsius. SuperMUC has 150,000 cores—and can do the work equivalent of about 110,000 personal computers. As a side benefit, as much as 90 percent of the captured heat can be used to heat several hundred homes, allowing both energy savings and energy re-use to reduce the carbon footprint.
The idea of liquid cooling has been around for some time. In fact, the same cooling technology was previously implemented in Aquasar, a supercomputer project in Zurich, with the captured heat being used to heat campus buildings.
Especially for large enterprises and service providers, energy consumption and waste in their data centers will rise steeply. According to Dr. Michel, liquid cooling is a key step toward achieving the goal of a green data center. But the researchers’ future aims are even larger, with the goal to dramatically reduce the size of servers in the data center to the size of a sugar cube, increasing efficiency by more than a thousand fold. The combination would be a giant step toward a much more efficient data center that has zero-net carbon emission while it operates.
More on SuperMUC and other IBM Research projects: http://www.research.ibm.com/articles/superMUC.shtml
To effectively compete in today’s changing world, it is essential that companies leverage innovative technology to differentiate from competitors. Learn how you can do that and more in the Smarter Computing Analyst Paper from Hurwitz and Associates.