Archives of the TeradataForum
Message Posted: Thu, 08 Oct 2009 @ 18:41:01 GMT
| Subj: | | Re: Identity column experiences |
| |
| From: | | Victor Sokovin |
| | I would like to hear some real stories about the usage of identity columns for the surrogate key generation. We are especially interested
in cases when keys being generated in various physical environments are being merged. | |
Boris,
one generic negative advice for the use of identity columns is that they will have to be re-designed if the customer ever wants to move on to
Dual Active architecture.
I am not a fan of using generated IDs as surrogate keys if it is possible to find business keys. Surrogate IDs are a bit of a waste of
resources, then, because they would help with the nice distribution across AMPs but won't help with any table scans or joins.
Of course, if the data is not going to be used for scans or joins and it has no natural keys then why not use the generated IDs? But there is
now the option of having tables without PI. It may give some even distribution as well, and if it does, then why bother about generated IDs?
Victor
| 
