Data Center Environmental Monitoring Recommendations
Stable environmental conditions are vital to reliable operation of your data center. This article explains the areas that you should be concerned about and provides best practice guidelines.
Environmental Monitoring and control is crucial to the smooth running of your data center. To maximise system reliability you should aim for a stable environment free from wild fluctuations. The key areas of concern are:
Too high a temperature reduces life of components while too low a temperature is uncomfortable to work in. The optimum range for operator comfort and equipment reliability is normally between 21 to 23 degrees Celsius (70 to 74 degrees Fahrenheit). Not only are these temperatures comfortable, but they allow a safe buffer zone in case of air conditioning or cooling failures. In the event of failure equipment will continue to work well outside the range giving you time to effect repairs or install a temporary cooling solution.
For a variety of reasons you need to take measurements at different points in the data center.
Air temperature entering and leaving key components, Servers, Routers, Switches, is important. Typically air intake temperature measurements are taken about 5cm (2 inches) in front of the equipment.
Temperature measurements taken in different parts of the data center can vary significantly. Concentrations of equipment can produce hot spots. Measuring the temperatures between equipment racks can give you early warning of approaching problems.
If the temperature gradient in the room should exceed more than a few degrees you risk overheating the devices higher in the racks. Place sensors at 2 (0.6 metres) feet and 5 feet (1.5 metres) off the floor to measure the differential.
You should place sensors 10 to 25 metres (30 to 50) feet apart, avoiding draughts or areas of turbulent airflow.
Measure temperature over an extended period, 24 hours or longer. This will allow you to check for fluctuations and to analyse trends or patterns in the data.
Aim to avoid temperature changes greater that 5 degrees Celsius (10 degrees F) per hour and humidity changes of + or -10% in the same period.
Temperature is not the only factor to take into account. Humidity is also important.
Relative Humidity (RH) is a ratio of the amount of moisture in a given sample of air at a given temperature to the maximum amount of moisture the sample could contain at that temperature.
Warm air can hold more water than cold air. So if the temperature rises the relative humidity deceases. Conversely if the temperature falls the relative humidity increases, ultimately to the point were condensation arises. High humidity, and particularly condensation, and computer equipment do not mix. Not only are electrical failures possible, but corrosion may appear over time.
As the load on Servers, Switches, and Routers varies so does the heat output. Maintaining both temperature and humidity within acceptable limits is vitally important.
Relative humidity levels between 45% and 60% are best for safe server operation. As with the recommended temperatures this range allows a safe buffer zone in case of air conditioning or cooling failures.
Maintaining relative humidity levels that are too low can cause problems with static. Static does not leak away readily in very dry conditions. With humidity levels of around 35% static discharge becomes a serious problem and build up is likely. Humidity levels below 30% should be considered critical.
Water, computer equipment and electricity do not mix. Flooding, though possible, is unusual, however air conditioning failures frequently also result in water leaks.
Wet floors and water dripping on equipment are to be avoided in data centers. Wetness detectors can be used to alert you to puddles or damp patches on floors.
Calculating the airflow requirements for a data center can be complicated. Clearly your heating, ventilation, and air conditioning (HVAC) equipment needs ample capacity for the amount of heat you expect to generate, but you also need to take into account the possibility of hot spots and the layout of equipment. Most devices draw cold air in at the front and exhaust hot air at the back, so airflow needs to be sufficient to prevent warmed air being drawn in to the next equipment rack.
Airflow recommendations suggest 3 to 4 metres per second (10 to 13 feet). Avoid turbulent airflow as this will be felt as a draught. The number of air changes per hour in high density data centers may be several times greater than that required in an office environment. Where an office needs about 2 air changes per hour, data centers may need a much as 25 or 30 changes depending on the heat load of the installed equipment.
Poor control and monitoring of the conditions in your data center may shorten the life of the equipment, overheating can cause intermittent faults, and in extreme cases cause equipment to fail catastrophically. The cost in time, money and lost business productivity can be considerable.
Having an independent system check on the air conditioning unit and monitoring devices is a wise precaution.
Careful environmental monitoring using strategically placed probes will give you a clear picture of the conditions in different parts of the data center. Coupled with alerts such a system should give you ample warning before conditions become critical or out of control.