Re: Best practices for Large import in empty database? - Mailing list pgsql-general

From Andy Colson
Subject Re: Best practices for Large import in empty database?
Date
Msg-id 542C128A.80905@squeakycode.net
Whole thread Raw
In response to Best practices for Large import in empty database?  (Daniel Begin <jfd553@hotmail.com>)
List pgsql-general
On 10/1/2014 9:13 AM, Daniel Begin wrote:
> I am about to feed an empty database with a large import of data (the
> size of csv files is about 500 GB). Tables are created but I haven't
> added any constraints or indexes yet. I wonder whether the best practice
> is to add them before or after the import. Are there other good
> practices that would ease the import process?
>
> Cheers,
>
> Daniel
>

create index after, as long as you dont need it for uniqueness checking.

then:

begin;
truncate table junk;
copy ...  with freeze;
commit;

Even though the table is empty, the begin,trunc,copy lets PG skip a
bunch of extra WAL.

-Andy




pgsql-general by date:

Previous
From: Tom Lane
Date:
Subject: Re: Best practices for Large import in empty database?
Next
From: jlrando
Date:
Subject: Postgres tcp_keepalive_xxxx parameters.