The EU-sponsored GreenDataNet project is so committed to reducing the energy demands of Europe's data centres by as much as 80%, that it is willing to commit €2.9 million to research. The project may be on the verge of realising its first substantial milestone thanks to a Swiss venture being undertaken at Ecole Polytechnique, Federale de Lausanne (EPFL). Project researchers believe they can increase server capabilities and deal with subsequent heat problems by using a revolutionary liquid cooling system.
The system, which was recently revealed during a GreenDataNet exhibition, involves creating a stack containing a CPU and multiple memory chips layered on top. By stacking the memory on top of the CPU, signal delays are reduced, thus allowing for faster processor times without the need for substantially more power. However, such a stack would generate a tremendous amount of heat.
To solve that problem, researchers have developed a revolutionary two-phase cooling system that utilises tiny channels running through the heart of the silicon chips, through which a cooling liquid could flow. The two-phase system indicates that a portion of the liquid would evaporate in order to take the heat with it. Nevertheless, that creates new problems of backflow and dry spots that still need to be solved. The researchers have said that they will design and build a system that works as intended; it is just a matter of time.
Should the project be successful, researchers believe there will be an added benefit for European cities. That benefit comes by way of harnessing the heat produced by the high-powered data centres of the future.
Because European cities tend to have very high population densities, municipal heating systems are the norm. What's more, European data centres tend to be concentrated in high population areas, unlike their American counterparts. Combining the two paradigms would mean excess heat from data centres being put to use to provide for municipal heating needs.
An Important Breakthrough
Those who are intimately familiar with how computers operate know that the problems associated with power and cooling are the bane of European data centres. Cooling more so, to the extent that inadequate cooling systems are largely responsible for holding back the development of more powerful computers. Simply put, we have not found cooling methods that are effective enough to push high-powered computing into the future. If the Swiss project is successful, what it accomplishes could be an important breakthrough.
Research has been ongoing for several years now, so it is safe to assume that the project is not on the doorstep of success quite yet. It is still going to be some time before it is able to produce a product with broad commercial appeal. We will be waiting with great anticipation.
In the meantime, it would be interesting to see if this project could be combined with another plan we wrote about last week; a plan to harness used electric car batteries to provide backup power for data centres. It is certainly something worth thinking about.