Viz např.

MySQL 5.1 partitions in practice
http://dev.mysql.com/tech-resources/articles/testing-partitions-large-db.html

Citace:
---------
Testing partitions has been a tiring experience. Gaining performance improvements is not a straightforward operation. What I assumed was a painless change turned out to be a long trial-and-error process.
There is no silver bullet

Applying a partition change to a table is no guarantee for performance improvement. The gain depends on several factors:

* the column used for partitioning;
* the function used for partitioning, where the native column type is not an integer;
* the server speed;
* the amount of RAM.

Nothing should be taken for granted.
---------

Důležitá je izolace změn a důkladné měření (alespoň 4-5 měření u každého testu). U Mysql je také důležité při více stejných měření vypnout query cache (select SQL_NO_CACHE).