 |
 |
Archives of the TeradataForum
Message Posted: Tue, 31 Jan 2006 @ 15:34:14 GMT
| Subj: | | Re: Primary keys without a Unique Index. |
| |
| From: | | Geoffrey Rommel |
| | Another approach that was used in experimental RDBMS projects was via a perfect hashing algorithm (a perfect hash has no clashes, so when
you get one, you know that the key has been used). | |
I don't want to get into too much detail, but how is this possible? In order to know whether you have a clash, you have to compare the hashed
value to a list of existing values -- so isn't that the same as a unique index? I guess you could use storage addresses, but then you would need
as many storage addresses as keys, which hardly seems to help.
Also, if the hash has no clashes, it has to map 100 million input values to 100 million output values, so how does this gain anything?
| |
