Archives of the TeradataForum
Message Posted: Mon, 18 Aug 2008 @ 20:42:16 GMT
| Subj: || || Column Order in Primary Index |
| From: || || Goli, Srikanth |
Does the order of columns in a primary index have any impact on the hashing algorithm? My understanding is that hashing algorithm is
insensitive to the order of columns in the index, and would always return the same AMP # regardless of the order of columns. Even if it were to
make difference, does it make sense to order your columns in such a way that you get better distribution, like ordering columns with wider range
of values to the right and columns with fewer values to the left? Is this something that Teradata commits to maintain, what if the algorithm to
hashing were to change?
Thanks for your input from the members.