I actualy need to join from 2 tables. Both of them similar and has more than 4 millions records.
CREATE TABLE prdt_old ( groupnum int4 NOT NULL, sku varchar(30) NOT NULL, url varchar(150), ); CREATE TABLE prdt_new( groupnum int4 NOT NULL, sku varchar(30) NOT NULL, url varchar(150) NOT NULL, ); The query returns group number and sku from old table where has no url in prdt_new table. INSERT into prdtexpired SELECT pn.groupnum, pn.sku FROM prdt_old po LEFT OUTER JOIN prdt_new pn ON (pn.groupnum = po.groupnum and pn.sku = po.sku) WHERE pn.url is null or pn.url= ''; I already have resolution for this problem where I seperate the query for each group. But when I address this question, I hope that Postgresql has some kind of table optimazion for large records. Based my experience it is faster to query from chopped smaller table rather than query from single huge table. I heard Oracle has some kind of table partition that acts like single table. ---------------------------(end of broadcast)--------------------------- TIP 7: don't forget to increase your free space map settings
