Storage vs. performance
Both UTF-8 and UTF-16 are substantially more compact than UTF-32, when averaging over the world's text in 
computers. UTF-8 is currently more compact than UTF-16 on average, although it is not particularly suited for 
East-Asian text because it occupies about 3 bytes of storage per code point. UTF-8 will probably end up as about
the same as UTF-16 over time, and may end up being less compact on average as computers continue to make 
inroads into East and South Asia. Both UTF-8 and UTF-16 offer substantial advantages over UTF-32 in terms of 
storage requirements.

Code-point boundaries, iteration, and indexing are very fast with UTF-32. Code-point boundaries, accessing code 
points at a given offset, and iteration involve a few extra machine instructions for UTF-16; UTF-8 is a bit more 
cumbersome. Indexing is slow for both of them, but in practice indexing by different code units is done very 
rarely, except when communicating with specifications that use UTF-32 code units, such as XSL.

This point about indexing is true unless an API for strings allows access only by code point offsets. This is a very 
inefficient design: strings should always allow indexing with code unit offsets. Moreover, because code points do 
not, in general, correspond to end-user expectations for characters, it is often better to use grapheme (user 
character) boundaries instead, and to store text in strings rather than as single code points. See The Unicode 
Standard, Version 3.0 for more information.

Conversion between different UTFs is very fast. Unlike converting to and from legacy encodings like Latin-2, 
conversion between UTFs doesn't require table lookups.

Overall performance may also be affected by other factors. For example, if the code units match the machine 
word size, access can be faster; but if they use up more memory, more page faults and cache faults may occur, 
slowing down performance.

All of the most common characters in use for all modern writing systems are already represented with 2 bytes. 
Characters in surrogate space take 4 bytes, but as a proportion of all world text they will always be very rare.