Over the last 10 years or so, great progress has been made in improving the efficiency of data centres. However, because businesses are constantly adapting in the face of new innovation, demands and challenges, change is often forced upon IT and the data centre facility in turn. As a result, facilities managers are faced with evolving demands and shifting goal posts from a business perspective, and the journey towards an ideal state of optimisation is continuous. In short, business demands are fluid, and facilities must be equally as fluid also to cope.
One of the current challenges many facilities managers are experiencing is pressure from the C-suite to reduce the costs and environmental impact associated with power and cooling. One of the methods used to achieve this is to simply raise the temperatures at which data centres are operate. But whilst doing so may reduce the financial and ecological cost of compute, the corresponding risk of failure increases. So is running data centres hotter really worth the risk?
Turning up the heat
The argument for operating facilities at higher temperatures is a simple one – a higher heat tolerance inside the building reduces the need for cooling air. And the less cooling your data centre requires, the less power it will use.
Whilst operating at higher temperatures has traditionally been thought to increase the risk of failure, recent research suggests that other factors, such as humidity, are more important in safeguarding against failure than overall temperature. Revised guidelines from ASHRAE, which allow for servers to run at hotter temperatures, lend further support to this argument.
Keeping your cool
On the other hand, it is important to remember that energy costs are only one factor in reducing overall business expenditure, albeit a significant one. If a data centre is run hot to the point where systems end up failing, all it delivers is a false economy for the business due to the cost and impact of the disruption it will cause. Also, there will always be a ceiling of some kind on the overall temperature at which facilities can run, so cooling remains an important factor in data centre infrastructure management (DCIM) even if temperatures are higher.
The underlying issue here is that any change to the operating temperature of a data centre can upset the delicate balance between capital expenditure, operational expenditure and unnecessary risk. In addition, the effectiveness of raising operating temperature will be specific to an individual facility. Lessons learned from one data centre are not necessarily applicable to another, and this makes the ability to predict the impact of any changes under unique circumstances invaluable.
The good news is that the impact of temperature changes to the data centre are eminently predictable, through the use of engineering simulation. Techniques that use 3D modelling to represent the data centre, power system simulation (PSS) and computational fluid dynamics (CFD) allow facilities manages to build an accurate picture of the impact of change on a data centre, and strike the perfect balance between temperature, expenditure and risk.