Do Computers Really Get Tired?
Q: I’ve always heard that computers get slower when they’re left on for long periods of time, but I wonder if that’s technically true. Often in claiming this, people refer to “electron buildup.” Does this actually exist? Do computers really get fatigued in any dimension?
A: Electronic devices experience some sort of fatigue, but it usually happens after a lot longer than several days. In the IT world, this is called mean time between failures (MTBF) and it is used as a measure of how reliable a product is.
MTBF is usually given in units of hours; the higher the MTBF, the more reliable the product is. Typical MTBF values for parts of a computer vary between different vendors but, on average, a CD-ROM drive is about 15,000, while hard drives are rated as 500,000 MTBF (that’s 57 years!). MTBF is a calculated average and should be used to predict more than as an actual proven number.
If electronic devices were actually subject to fatigue, would a satellite be able to operate for several years? Probably not. The satellite manufacturing company knows that no one will be able to reset it every two weeks, so it designs and tests it to last.
But this is definitely not the issue here, since the concern is about several days, not months. The problem with personal computers is more related to misbehaving applications, heat and related issues.
One common contributing factor is memory consumption. If an application does not release the memory space that was allocated to it and keeps doing so over and over again, in a given span of time the computer’s RAM will be fully occupied and additional applications will be sent to the swap space or fail. This is commonly referred to as a memory leak. Another kind of misbehaving application can start a process every few minutes and not terminate the previous process it was using.
The framework that applications run in — the operating system — can also generate the same kind of CPU and memory issues, and Windows support teams know that a common solution or preventative measure is to restart a computer every week or so.
Another major factor is related to environmental issues, or one particular issue, which is temperature. Electronic devices are designed to operate in an optimal way by residing in a cold environment (which is usually too cold for humans), and if they aren’t kept cold enough, then things start to get weird. Computers that work for several days might accumulate excessive heat and might have problems getting rid of it without being turned off.
So what solutions exist for dealing with such problems? Reset your computer every three or four days. Don’t wait until it starts to act weird, because this will happen in the middle of a presentation or another inconvenient moment. Put a reminder in your calendar or just make it a habit to reset every now and then. By reset, you should not use the standard restart from Windows, you should shut down, wait five to 10 minutes and turn your PC back on (lunchtime might be a good opportunity for that).
There are software tools that can assist in keeping things under control on the memory and CPU utilization fronts. Memory managers are more common, but there are also some CPU managers out there. Google the keyword “memory manager” or “CPU manager” with your operating system version (XP, Vista, etc.) and you should find many appropriate applications.
Some operating systems are better at controlling malfunctioning applications than others. UNIX used to be the leader in that, especially the commercial server platforms such as Solaris, AIX and HPUX. A server’s operating systems are considered better at running stable for long period of times; they incorporate more security and protection for the hardware resources then workstations.
Avner Izhar, CCIE, CCVP, CCSI, is a consulting system engineer at World Wide Technology Inc., a leading systems integrator providing technology and supply chain solutions. He can be reached at editor (at) certmag (dot) com.