Power & Cooling
Increasing data centre cooling efficiency for higher density IT loads
The system of removing heat from IT equipment using cooled air and a raised floor environment has served the data centre market well throughout its 40-year life. However, moving large volumes of air over big distances in which hot and cool airstreams mingle, can add to infrastructure inefficiency. With today’s IT racks requiring a greater volume of cooling more precisely delivered, it’s time to consider enhancements to cooling architecture which will improve PUE, improve cost-effectiveness and mitigate risk of hot spots. By APC, By Schneider Electric.
The first commercial blade server was launched by Compaq less than a decade ago and the high density data centre revolution was sparked. It already feels like a long time since data centre designers first started pointing out the inadequacy of perimeter cooling solutions with a raised floor plenum for reliably cooling anything much beyond the 1.75kW/ rack norm. Although with care, densities of 4kW/ rack can be achieved, perimeter mounted CRAC units have been demonstrated to be able to deliver reliable cooling redundancy in the event of the loss of a unit from the system. That is, unless units are drastically oversized. Physical characteristics of raised floor cooling designs which can contribute to inefficiency in today’s data centres include bypass leakage, poor air coupling to the IT load, uneven distribution of cooled air, hot spots that require a lowered supply temperature, the need to over- provision computer room air handling (CRAH) units, under floor obstructions and blockages, inability to adjust to dynamic power variation, and other problems.
Oversizing of infrastructure, over compensation with low air inlet temperatures and ‘stranded capacity’ all escalate capital equipment cost as well as data centre operating expenses. Worse still, they do not necessarily ensure a reliable and adequate cooling solution.
A lot of these problems are overcome with newer approaches that are effective even without a raised floor and traditional CRAH units. Such approaches result in improved PUE along with improved compatibility with high-density and dynamic loads. Without doubt, traditional cooling approaches will continue to be used, but there is an increasingly good business case for aisle containment. Containment prevents the mixing of hot and cold air streams to offer more efficient data centre cooling infrastructure. It can also be fitted into new designs or added retrospectively to existing data centres.
Keeping the Cold aisle contained Containment of the cold aisle is typically retro fitted in traditional perimeter-based data centre
30 | DATA CENTRE SOLUTIONS |
www.datacentresols.com
environments. Traditional cooling environments use the entire room as a hot air return plenum and use cold air via the raised floor plenum to the cold aisles. This allows the rest of the data hall to become a large hot air return plenum while separating the hot and cold air streams.
Whilst a considerable step forward from legacy cooling solutions, cold aisle containment does have drawbacks such as the inefficiencies which result from distances and pressures required for adequate air distribution. Inefficiencies are caused in a room-based approach because there is a requirement to move cold air from the perimeter CRACs to the load. A row-based approach brings the source of the cooling in close proximity to the IT equipment and therefore less energy is required to deliver cold air to its destination. There are also density limitations when using cold air containment with distribution through a raised floor. On average the limit is around 6 kW per rack and higher densities can only be achieved through the use of customised designs. Additional fan
Winter 2010
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44