Avoid using as the identification string type column:
- Consumption of space.
- Slower than the digital type (MyISAM strings for use in compression index, the query will be slower).
For MD5 (), UUID (random string generation), these values are distributed in a large space, leading to insert, select statements slow:
- insert randomly written to different locations, so that insert slow; cause a page split, random disk access for clustered indexes also produces engines index fragmentation.
- select slower, logically adjacent lines in different location of the disk, memory.
- Random value will make the effect of variation cache, cache locality principle failure; memory is too small, resulting in frequent refresh, do not hit.