Checksum Transformation
(http://www.sqlis.com/default.aspx?21)Any details on how "unique" a checksum value is?
The inbuilt SQL Server CHECKSUM() function seems to have a pretty horrible collision rate due to being only a INT. It's not very useful in determining unique rows amongst anything more than a small handful of rows.|||I have no real numbers on how unique it is, only my testing, which seemed rather good. It uses the .Net Hash function (C#), and the internals of that are not documented.
Feel free to try it.
Using an Int makes it easier to combine multiple columns, where as most other methods would produce a very large value, or just not be capable of handling multiple values.
Darren
No comments:
Post a Comment