Some data centre trends and practices meant to improve cooling and efficiency may actually do more harm when implemented during or following a power outage, according to a whitepaper published an energy management firm.
The paper titled Data centre temperature rise during a cooling system outage, was published by Schneider Electric, a multinational company that specializes in electricity distribution and automation management with facilities in Canada. The paper was written by Paul Lin, senior research analyst at Schneider Data Centre Science Centre, and Simon Zhang, senior research engineer for data centre design and management software and James van Gilder, responsible for Schneider’s data centre cooling software development and related research.
The paper focused on four data centre trends:
- Right sizing cooling capacity
- Increasing power density and virtualization
- Increasing IT inlet and chiller set-point temperatures
- Air containment of racks and rows
In normal conditions, these practices are beneficial to a data centre, but could have negative effects on cooling after a power outage, the authors said.
For example, right-sizing, which aligns cooling capacity to actual IT load, results in energy efficiency and lower capital cost.
However, excess cooling capacity is needed when a data centre is under very high temperatures such as during or right after an outage, according to a post on the online publication ITWeb. In such a situation, if total cooling capacity were matched to the heat load the facility could theoretically never be cooled to its original state and there would always be heat in excess of IT load, the authors said.
Recent trends towards compact IT equipment, has resulted in the increase of rack power densities. Blade servers can result in rack power densities that exceed 40Kw per rack.
Virtualization has also increased the ability to use and scale compute power bumping up CPU use from the typical five per cent in non-virtualized servers to 10 per cent of over 50 per cent.
The potential downside, is that virtualization and rack power density can produce more heat in a given space. This could result in data centre operators having a smaller window of time before IT inlet temperatures reach critical levels during a power outage.
Increasing IT inlet and chilled water set point temperatures increases the number of hours that cooling systems can operate on economy mode. It becomes more expensive to cool chilled water the more the set point temperature is reduced below the fixed ambient temperature. Higher IT inlet temperatures give operators less time to react during a power failure.
Air containment or racks and rows enhance predictability and efficiency of perimeter cooling systems in raised floor or hard floor environments.
Containment systems, however, prevent air from mixing with the rest of the data centre room and will tend to raise temperatures during a power outage.
The authors suggest that data centre operators take these scenarios into consideration. They said it is possible to design cooling systems that take into account longer runtimes during emergencies or power outages.