There is plenty of evidence that dirt, dust and debris poses a real risk in data centers.
We will look at some examples we have found through our work as data center consultants.
Figure 1: Reasons for failure of electronic equipment (Courtesy BCC Inc. Report GB-185R)
This chart illustrates why it is important to maintain a good control over temperature, humidity and dust in the data center. Vibration is also important but is less likely to be an issue unless you are in a region with significant seismic activity or your computer room is located too near to roads, rail transport or moving or rotating heavy loads.
Dirt, dust and debris can cause failures and so it is important to understand that contamination comes from many sources including;
- External dust
- Internal dust
- ‘Zinc’ whiskers
Dust may also be conductive. Insulating dust is more of a problem because it is electrostatically charged and so sticks to electronic components. This effectively thermally insulates them so they fail from overheating, regardless of the actual temperature of the data center.
Here are some of the failure mechanisms caused by dust;
- Insulation of electronic components causing thermal overload
- Corrosion caused by acidic or caustic particles
- Short circuit caused by conductive particles
- Clogging of motor and fan drives and drive belts
Figure 2: Failed power supply in an IBM server cause by a cockroach short-circuiting the PCB. Courtesy of – Why do Power Supplies Fail, and what can be done about it? IBM 2005
Insects also conduct electricity!
To prevent insects and dust from entering the room we must;
- Filter incoming air
- Seal the room
- Maintain a positive air pressure in the room.
Unfortunately, in some countries such as Saudi Arabia dust is a big problem. Desert sand in such countries is very fine and will find its way into almost everywhere. It is therefore even more important to filter incoming air, seal rooms and maintain positive pressure in technical rooms.
Unfortunately dust will still get in in spite of all our efforts and so we must put in place regular maintenance schedules which include cleaning and replacing filters and cleaning contracts with specialist cleaners.
If you want to know more about cleaning data centres and the precautions which need to be taken with the processes and materials that are used then please take a look at our DCCS Data Centre Cleaning Specialist Course.
The cost of a data center outage
According to recent survey the cost of data center outages is about $9,000 per minute with an average outage cost of $740,000. Cost is not always measured in dollars; there are reputational costs, brand damage and possible legal ramifications from not supplying a contracted service.
Bearing those figures in mind it makes sense to put the relatively small investment needed into taking appropriate precautions against dust, debris and dirt contamination by putting in place filtration, proper maintenance processes and training cleaning employees in the correct techniques and materials to use to avoid risk of outages.
Read more about The Importance of Cleaning the Data Centre