final thought the argument could be made that relocating facilities to the Arctic is ultimately a waste of time and resources – especially as the amount of cooling required for data centres is likely to be reduced further still over the next few years .
Then there is the simple fact that ambient temperatures don ’ t need to be arctic in order to be efficient . With ASHRAE ’ s expansion of the recommended operating envelope , supply temperatures as high as 27˚C are classed as acceptable . This will allow for considerable utilisation of outside air across many areas of mainland Europe for most of ( if not all of ) the year with no additional mechanical cooling required .
Constructing data centres in such remote locations can cause further complications – for instance , latency is a hugely important issue for financial organisations that rely heavily on the speed at which transactions can take place . Secondly , providing the skills and the services required for maintenance and management of these sites becomes an additional headache . For corporations such as Facebook that do not have any direct customers who rely on 100 per cent uptime and immediate access to data , the above issues are not as critical to the business . While it works for them , many of us need a solution closer to home .
The real barrier to efficiency The problem with building data centres in extreme environments is that doing so is a convoluted way of solving what is essentially an engineering problem . Whilst Arctic data centres are interesting , the focus shouldn ’ t be on saving energy costs through moving data centres to somewhere where energy is ‘ free ’ – be that the Arctic for free cooling or even the desert for free solar power . Instead , making facilities more efficient in practical environments is a more logical approach .
Data centres are all about balancing practicality against the needs of the business . Whilst this sounds fairly straightforward , problems arise when risk is introduced into the equation . Facilities managers may be reluctant to make changes to a data centre due to the potential for failure . For instance , undertaking initiatives to drive up utilisation , reduce energy costs or refurbish an existing data centre increases the risk of hardware failing , which would interrupt business processes and cause serious disruption .
So how can these variables and factors be tested safely and with minimal risk , and without simply resorting to building a new data centre ? Through the use of engineering simulation techniques such as 3D modelling , power system simulation ( PSS ) and computational fluid dynamics ( CFD ), engineers are able to predict the impact that introducing change will have on the data centre environment – but without the usual associated risk . For example , a lot of the hot air reflows which currently cause data centres to waste energy through overcooling could be eliminated by testing optimisation techniques in a virtual environment .
So before considering outlandish data centre locations , or resorting to overengineering to solve a problem , there are easier , cheaper and more convenient ways to optimise existing facilities .