[TYPO3-50-general] UTF-16

Robert Lemke robert at typo3.org
Fri Nov 10 10:25:39 CET 2006

Hi Martin,

Martin Kutschker schrieb:

> After some thinking I think that space does matter in the DB 
> performancewise. The larger the data in the rows the larger the buffers 
> for joining and sorting must be.

I guess that the opposite is true. What helps a database to find data 
more quickly is (often) not the size but the fact if creating an index 
is easy or not. There is no golden rule when an index is easy to create 
but one big aspect is a fixed row width.

Imagine you have a phone book and want to create an index for the first 
and the second character of the last name. If we use UTF-8 we will have 
a variable size for the characters for names with special characters, 
with UTF-16 it is very unlikely that a name uses all the 4 bytes.

But that is only guessing and as I learned, guessing performance 
problems is a very bad idea. We'll have to measure it if we really want 
to know.

Maybe someone has the time and expertise to set up such a test case?


More information about the TYPO3-project-5_0-general mailing list