How to Manage Sudden Server Room Temperature Rises

Server Room Thermal Image

With weather patterns in the UK forecast to become more erratic we are seeing a greater focus being placed on how to design a server room to ensure it is adequately cooled. This summer saw a large rise in demand for emergency air conditioning and cooling units as some server rooms struggled to maintain a comfortable and safe working ambient.

Summertime temperatures in the UK are forecast to both rise and become more erratic. This presents server room designers with an even greater need to design the right cooling system for any server environment, whether it is an on-site server room, edge computing facility or cloud service co-location datacentre. Server rooms need cooling solutions that are flexible and adaptable and not just to weather conditions. Whilst server manufacturers strive to make their hardware as energy efficient as possible, the actual power they draw is rising in line with their processing power. Typical populated racks are drawing higher amounts of power (up to 30kW or more) and this means a higher internal ambient and warm air to cool down at the rear of the racks.

What is for sure is that a sudden spike in ambient temperature or prolonged period of higher temperatures can lead to IT hardware failure. Put simply, heat kills electronics. A warm ambient also makes for an uncomfortable working environment for anyone having to work the facility. Smaller server rooms, network closets and comms rooms possibly have the worst of all worlds. As more equipment is piled in there is simply less space within which to plan for an optimum ambient environment.

Server room environmental monitoring can assist the situation. What you don’t monitor, you cannot control. Monitoring room temperatures allows for corrective actions to be taken when an alarm threshold is reached, and potential disasters can be avoided. Best practice for a server room environment is a temperature of 20-25˚C and with a humidity of 45-50%.

Whilst most servers can work above 30˚C, other hardware items within the IT space can be slowly damaged. Consider the lead acid batteries in a typical UPS system. This type of battery has a recommended working life of 3-5 years at 20-25˚C for a 5-year design life battery. For every 1degree rise above 30degrees, the design life halves.

With adequate computer room monitoring (for temperature and humidity) most overheat situations can be avoided. If you do find yourself needing to take emergency actions here is a typical checklist you can follow:

Overheating Server Room Design Checklist

  1. Check server air flow: if your server room temperature has been constant for a period, a sudden spike in temperature can mean several things. The first thing to do is check the actual air conditioner system to ensure they are operating i.e. green light on and emitting cool air at the temperature you have set. Check air intakes and return vents for blockages and that any doors are closed into the room. Check the temperatures within the room at various points and with a thermal monitor if possible, to build a room profile. Check for thermostat failures.
  2. Check the server racks: equipment can overheat within a server rack leading to a build-up of heat. Server rack air flow can also be less efficient if there are large spaces where servers and equipment have been removed or that part of the rack has not been populated. Install blanking panels to improve the air flow where this is the case and ensure that warm air is not blocked from leaving the rear of a cabinet.
  3. Check the floor tiles: if you have a raised access floor and this forms part of your air flow channelling, ensure that no floor tiles are missing or fitting poorly. As with any window that doesn’t close and seal properly, this can lead to air (whether cool or warm) leaking into an area where it is not planned for.
  4. Check the room layout: has any part of the server room design been changed in terms of additional racks or even the introduction of a large centralised UPS system? If there are multiple rack cabinets, are they arranged into a hot-aisle and cold-aisle configuration? The worst set-up would be the warm air from the rear of one rack being drawn into the front of another cabinet. In smaller rooms and network closets air flow is just as important even if you are not using multiple server racks. The path for warm air from a server must be routed away from other servers and IT hardware to avoid over temperatures.
  5. Check the weather: this may seem obvious, but it may be that the overall ambient temperature is hot outside the building and over a prolonged period. This is important for remotely operated server facilities. As in the UK this summer (2018) a prolonged heatwave lead to increased day and night time temperatures. The warmer air outside buildings lead to an overall increase internally.
  6. Consider emergency and portable air conditioning: if you can isolate the reason for the temperature spike you may need more time to consider a preventative action for the future. In the meantime you may need to consider using portable air conditioners to boost that of your onsite HVAC systems. Whilst heating and ventilation air conditioning systems are generally maintained annually, the system could be as old as the building and the premise used for HVAC design calculations could have changed.
  7. Seal off the server room: check to see that the server room is isolated from the rest of the building and ensure adequate access control. Building temperature needs change over a year with respect to weather and occupation. Autumn and winter can see HVAC systems focused on raising temperatures and the provision of heating. This will often be to a higher temperature than that required within the server room. Access control is also important to both secure the room and ensure no-one can add additional equipment or run non-IT systems in the room.
  8. Check the lighting: whether the room is large or small, the lighting could be automatically set to switch off if no one is in the room or to operate in a less intensive eco-mode. Modern LED lighting is very responsive and aside from being more energy efficient will emit less heat into the room than traditional fluorescent tubes and light fittings.

The right cooling solution is often specified and chosen during the server room design phase of many projects. It is subsequent events that lead to operational issues. Even in a well-designed server room or datacentre facility, there will always be the need for adequate environmental temperature and humidity monitoring. There are several manufacturers who specialise in this type of system and at Server Room Environments we provide a complete server room design and installation service for the manufacturers we work with.

These types of environmental monitoring systems can be connected via the IT network to a DCIM (datacentre infrastructure management) package and/or a BEMS (building energy management system). Alerts and alarms need to be configured for pre-set thresholds and response plans formulated to ensure that any changes in ambient temperatures are formally investigated by on-site staff and your local air conditioning and HVAC maintenance company.

comments powered by Disqus

Related blog posts

Prev Article
How to Improve Datacentre Energy Efficiency Using Aisle Containment
Blog box fixed 860766 1607528581

The problem of how to efficiently cool server rooms and datacentres is become more complex. Today’s IT environment can include a mixture of IT servers including high density racks leading to ‘hot-spots’ within racks and spread throughout a server environment. Edge computing will increase the complexity of the problem using remote compute facilities that may be more exposed to external ambient temperatures.

Read more ...