Answer by Mrs_Fatherjack
I found this which explains it quite well. Essentially though the primary key must be unique and it's a lot harder to keep text unique than it is integers....
View ArticleAnswer by Kevin Feasel
There are two articles that I would say are must-reads on this topic. The first is [Michelle Ufford's post on what makes a good clustered index...
View ArticleAnswer by Blackhawk-17
One of the big reasons to use an INT is that it can be constantly increasing thus reducing or eliminating index fragmentation.
View ArticleAnswer by WilliamD
Don't forget that if the PK is the clustering key on the table it will appear in each and every index that is additionally created. If the PK is a CHAR(N) column, this will probably be larger in size...
View ArticleAnswer by Magnus Ahlkvist
I agree with all the points about using identity keys as clustered indexes to reduce fragmentation. But there's more to it. If you usually search on exact matches in the lookup value, I agree that you...
View ArticleAnswer by TimothyAWiseman
Kevin's answer is absolutely fantastic, but I think no discussion on primary keys can be complete without considering the pros and cons of natural, artificial, and surrogate keys, and the best place...
View Article
