I need your advice on this matter:

[1]. What is the different between specifying 'primary key constraint' during table creation (together in the same schema) and creating 'unique index key' after table table has been created?

[2]. Referring to the question [1] above, which provide better performance e.g.
(a) --> create empty table (without any primary
key or index)
--> load data into table
--> create index.

-or-

(b) --> create empty table (with primary key
specified
--> load data into table
--> create index

Hi,

I suggest you that:
For this case, create table with primary key "disabled", next load data, next run "set constraints for "table" enabled, next create all index and next run update statistics for "table" with distributions.

Gustavo.

What is the purpose of "set constraints for "table" enabled you've mentioned. And also what is the purpose "with distributions" means? Actually, I'm still beginner in this term.

Hi,

Ok, for example:
create table xxxx
( col1 smallint not null primary key pk_col1 disabled,
col2 integer,
.......
)

This create the table with the primary key but not create the index.
Next you can load data and next enabled the primary key, then now the index is created.
During load of data, none index is used, then load fastest.

Gustavo.

Hi, how to enable primary key back?