Why don't you always use a high sampling rate? I was just curious, as I was reading about aliasing.
As far as I understand, aliasing comes from the fact, that you use a bad sampling rate, resulting in getting a wrong waveform compared to the one you had from the beginning.
So why is it you just don't use a fast sampling rate all the time ? Or maybe you do that, and I just don't know ?
 A: 
As far as I understand, aliasing comes from the fact, that you use a
  bad sampling rate

Aliasing can also come from a 'bad' anti-aliasing filter.

So why is it you just don't use a fast sampling rate all the time

For the same reason that we don't use a sledge-hammer to crack a nut.
The problem isn't so much that the signal of interest is aliased, it is that high frequency, out of band information is aliased.
One approach would be to use an anti-aliasing filter to properly band-limit the input to the ADC.
Another approach is to oversample and then use a digital filter to reduce the bandwidth followed by decimation to reduce the storage requirements.
Another approach is a hybrid of the above.
Essentially, you don't want* to store any more samples than are necessary to properly analyze and reconstruct the signal of interest.
*Unless, of course, you're a high-end audiophile.
A: Data storage.  If you started to sample a signal at three times the rate, it would take three times the memory to store.  Unfortunately, if you only have a small amount of memory to use for storage, then sampling at three times the rate means that you can only sample for one-third of the time you would have been able to at a lower sample rate.
A: You may want to consider the actual processing of the digital signals. If the rate of samples is too high the system may not be able to process them fast enough - it runs out of processing time.
