Hi!
On Aug 30, Mark kirkwood wrote:
> Dear List,
>
> I then LOAD 300 rows into it ( about 350Mb)
> This takes about 2m30s on my ( fairly old ) hardware ( I think this is quite
> fast ).
>
> CREATE UNIQUE INDEX fact0_pk ON fact0(d0key,d1key,d2key) ;
>
> Which unfortunatly takes about 20m
Mark kirkwood wrote:
>
> Dear List,
>
> I have a (standard table handler ) table defined as follows :
>
> CREATE TABLE fact0 ( d0keyINTEGER,
> d1keyINTEGER,
> d2keyINTEGER,
> val INTE
>
> Distribution wise the d0key values are loaded in order. there are 300
> distinct d0keys, 300 distinct d0keys and 100 distinct d2keys.
>
> The next step is creating an index as follows :
>
> CREATE UNIQUE INDEX fact0_pk ON fact0(d0key,d1key,d2key) ;
>
> Whichunfortunatly takes about 20m (
Mark kirkwood ([EMAIL PROTECTED]) writes:
> CREATE UNIQUE INDEX fact0_pk ON fact0(d0key,d1key,d2key) ;
>
> Whichunfortunatly takes about 20m ( which I think is a bit slow )
Have you tried building the table structure with the index already in place,
then importing the data?
--
Ed Carp, N7EKG
