I do not think that using dblink can improve performance. Using dblink will get busy two servers instead of one, not to mention data that are send through network.
You need to specify more information about table (number of rows, indices, query plan),
and then someone might be able to help You.
In some(most) cases VACUUM ANALYZE improves performance.
----- Original Message -----
Sent: Tuesday, May 27, 2003 2:36 PM
Subject: [GENERAL] Clustering using dblink
Dear Milist,
I Think this is the hard part for me,
I've been porting from Ms-SQL to PostgreSQL, and
All done succesfully, but the speed is still became
our apprehension right now. we have tried to add the
Hardisk(SCSI) with a controler for each. And tried to
spliting a view table or schema for helping I/O proccess.
It's helpfull just for a while. But it's getting
Slow again right now.
And the last one i had is just using dblink for spliting
my schema and table to another server. This's my part
of view's script...;
create view v_stock as
select * from
dblink('hostaddr=192.168.0.220
port=5432
dbname=d_inventory
user=inventory
password=',
'select * from t_stockreal')
as t1(
fc_branch character(6),
fc_stockcode character(20),
fn_lastdisc3 numeric(18,0),
ft_note text );
But It's more slow. Please somebody help me Please.
Thank's and GOD Bless You All.