Standby Measures

12 November 2008

Nick Campey explains the hot topic of temperature control in data centre environments and the need to plan ahead to reduce the risk of data melt down.

FOR FM'S MANAGING DATACENTRES HIGH EXTERNAL temperatures present major challenges
....The equipment is often so sensitive that even a very small rise in temperature can affect it, causing potential damage to the equipment and loss of data.

....The equipment itself generates heat, and when that much equipment is held together in a single space the incremental rise in atmospheric temperature caused by just a few pieces of equipment overheating slightly can be very significant indeed.

....Blade servers can produce up to 20 per cent more heat than traditional rack-mount servers.

....If data centre air conditioning fails, temperatures can rise by as much as 15 degrees within 20 minutes.

The cost of a data centre overheating can be measured in £100Ks per hour, with single pieces of equipment representing a significant IT investment and irreparable damage happening very quickly indeed following a rise in temperature of just a few degrees. And the costs don’t end there. Massive amounts of data can be wiped within hours involving significant costs in order to restore both the equipment and the data and a company’s reputation.

Purpose-built data centres are designed with temperature control built in, from the ventilation quality of the building itself through to the type of cable management and shelving used. The fixed air conditioning systems are both technically advanced and designed to provide additional capacity to address the demands of environmental factors. Sudden increases in atmospheric temperatures, as well as fluctuations in data centre activity may affect the temperature of individual pieces of equipment.

Specification of data centre air conditioning systems also needs to take account of the fact that the number of pieces of IT equipment on site may grow, and the heat collectively generated by the equipment will therefore increase too. Unfortunately, the amount of equipment on site and the heat generated may grow gradually over time, and it is easy to find that the air conditioning system has reached the limits of its capacity without any significant or sudden changes taking place.

If this happens, not only is the air conditioning system more vulnerable to faults but it may also allow the temperature to rise by a degree or so, enough to cause damage to highly sensitive electronic equipment. Regular maintenance of the air conditioning system is also part of the ABC of data centre FM and if the system is vulnerable to break down because it is being run at or beyond full capacity this will be flagged
during routine maintenance. A back-up system provides an additional safeguard against the risk of the equipment overheating.

Sometimes, however, these standard safeguards are not enough. There is always a possibility, no matter how remote, that the main air conditioning system will reach capacity and the back-up will fail. If this were to happen, the data centre could suffer business critical losses within as little as a couple of hours, resulting in
huge financial costs to the client’s business in terms of IT spend and a reverberating commercial fall out in terms of data loss and damage to the client’s reputation.

For an increasing number of data centre FM specialists, dealing with an HVAC hire specialist in advance of any problem, the FM team can not only assess the risk more thoroughly but have an agreed plan of action in place that will ensure they have the best-fit temporary air conditioning units on site and in position quickly and efficiently if the need arises. The HVAC hire specialists will then survey the most business critical sites to assess the cooling load requirements and plan the most appropriate equipment and best positioning for units for a variety of fixed air conditioning failure scenarios.

These site survey visits also provide the HVAC hire specialist with an opportunity to investigate any conditions that could potentially cause delays in the event of an emergency, such as access restrictions or limited delivery times.

Assessing all these factors in advance saves vital time in the event of an emergency because it means that a business continuity action plan is in place and ready to be rolled out. In general terms, there are two types of cooling units that are most suited to data centre environments: portable water-cooled ‘split’ air conditioning units or larger chiller units. At Andrews Sykes, the PAC 22 6.5 KW unit is by far the most
commonly specified air conditioning unit for data centre applications because its condenser unit can be positioned up to 30m away from the cooling unit, ensuring that warm air is diverted well away from the data centre. For many fast turnaround data centre applications, putting several PAC 22 units in place is often the most flexible solution. For more long-term hire requirements when significant chilling power is required, larger chillers may be more appropriate and sited outside the building with services routed through air handling units.

Some datacentres may choose to pay a ‘standby’ hire rate to reserve the relevant kit or even to have it on site in a dedicated ‘holding area’. This may seem a little ‘belt and braces’ but, it could prove the difference between a heated situation and data centre melt down.

A typical standby emergency air conditioning programme for one of the UK’s foremost telecommunications and media companies covers several hundred data centre sites spread across the country, ranging from single servers to enormous data hubs with business critical equipment controlling its telecoms, email,
internet and TV services to customers. Andrews Sykes was one of a number of HVAC hire suppliers that had provided air conditioning units on a responsive basis, but the FM provider wanted to put a business continuity programme in place and invited the company to assess the problem areas and draw up a plan.

Andrews Sykes began by working with the FM contractor to draw up a list of sites for survey, establishing the special requirements of each of these sites, as well as using this research to aid with other similar standby plans. It wasn’t long before the business continuity plan was tested. Within days of the programme beginning, Andrews Sykes was called in to put temporary hire equipment in place at a data centre in South
Wales – without it, services would have been affected for customers across Bristol and South Wales. Currently Andrews has 287 individual pieces of kit on site at 17 different data centre locations, helping the FM contractor to react quickly to rises in temperature.

● Nick Campey is National Accounts Manager at HVAC hire specialist, Andrews Sykes