Re: Why would a scan take so long?

These three queries may help you to understand what's going on.

EXPLAIN ANALYZE SELECT count(catalog.id) FROM catalog;
EXPLAIN ANALYZE SELECT count(catalog.id) FROM catalog WHERE flag=false;
EXPLAIN ANALYZE SELECT count(DISTINCT catalog.id) FROM catalog WHERE flag=false;

Regards,
--
NAGAYASU Satoshi <satoshi(dot)nagayasu(at)gmail(dot)com>

------元のメッセージ------
送信者 : Michal Politowski
送信者: pgsql-general-owner(at)postgresql(dot)org
To: pgsql-general(at)postgresql(dot)org
件名: [GENERAL] Why would a scan take so long?
送信: 2010/10/2 12:13 AM

EXPLAIN SELECT count(DISTINCT catalog.id) FROM catalog WHERE flag=false;
QUERY PLAN
-----------------------------------------------------------------------------------------
Aggregate (cost=1615927.27..1615927.28 rows=1 width=8)
-> Seq Scan on catalog (cost=0.00..1603214.56 rows=5085084 width=8)
Filter: (NOT flag)

SELECT pg_size_pretty(pg_relation_size('catalog'));
pg_size_pretty
----------------
9380 MB

Nothing else is going on the system, during the query disk reads rise from
around 0 to > 100MB/s, so I would assume it should take a couple minutes
and it takes ten times longer:
Time: 1495549.716 ms

What am I missing?

--
Michal Politowski