Data Centres vs the Laws of Physics
Q : How much further can we take data centre efficiency before we start bumping up against the laws of physics ? Scott Wilson
Data Centres vs the Laws of Physics
Q : How much further can we take data centre efficiency before we start bumping up against the laws of physics ? Scott Wilson
TITLE : GLOBAL HEAD OF BUILDING PHYSICS & SUSTAINABILITY
COMPANY : BLACK & WHITE ENGINEERING
as total power consumption or water consumption , that would be necessary to be able to verify any claims is completely lacking ,” he says .
The issue is , therefore , that unless the metrics are in place to accurately judge whether or not the data centre industry is actually becoming more sustainable , it ’ s impossible to put pressure in the places where it ’ s needed to drive change . “ What gets measured gets fixed ,” he says matterof-factly . “ Transparency of environmental data will inevitably lead to improvement , to the development of new innovations and solutions that will address the main drivers of negative environmental impact and may lead to supportive regulatory action . But without that information , change can not happen . So , it starts with transparency .”
As energy can neither be created nor destroyed , all electrical energy that enters the IT racks will become another form of energy . Namely , heat .
As the demand for data increases so does the demand for data centres . As a result , data centres are being built in less-thanideal climates where free cooling is not possible or only possible for a portion of the year . If the ambient temperature is warmer than the internal temperature work must be done to discharge that heat . While we can reduce PUEs through improved efficiencies , there will be a theoretical limit to what we can achieve , particularly in summer .
To avoid this , data centre locations need to be more carefully chosen to allow for passive / free cooling . For example , located in cold climates or close to a water source where heat can be discharged . Smart technologies can also play a part in reducing annual consumption . Through machine learning a data centre can tune a cooling system to match predicted consumption patterns and weather conditions .
datacentremagazine
. com 85