pg.ddx.io  pgsql-admin@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Prasanth <dbadmin@nqadmin.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-admin@postgresql.org
Subject: Re: Query Plan - Index Scan & Seq Scan
Date: Thu, 12 May 2005 10:44:32 -0500
Message-ID: <428379E0.4010709@nqadmin.com> (raw)
In-Reply-To: <886.1115910870@sss.pgh.pa.us>
References: <428370AC.3070207@nqadmin.com>
	<886.1115910870@sss.pgh.pa.us>

Thanks for the prompt reply.

Table a has about 6 million and table b had a little more than half a million.

Sorry I wasn't exact about my numbers before.

I will be having the where conditions on both the tables that would bring down
the count drastically. Even in this case the planner is going for a seq scan.

A where condition I always use is shown below. This is bringing down the number
of rows from 6.5m to 1210. I have an index on code also. Even here it is going
for seq scan.

EXPLAIN ANALYZE SELECT count(*) fROM a where Code >2;
                                                          QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------------
 Aggregate  (cost=141321.09..141321.09 rows=1 width=0) (actual
time=6454.063..6454.064 rows=1 loops=1)
   ->  Seq Scan on a  (cost=0.00..141318.40 rows=1076 width=0) (actual
time=15.687..6453.292 rows=1210 loops=1)
         Filter: (code > 2)
 Total runtime: 6454.140 ms
(4 rows)


Below is example where I have where conditions on both the tables this in effect
is limiting the number of rows from each tables a & b to 171 & 1076 respectively.

EXPLAIN (SELECT fund_value FROM b INNER JOIN a ON a.id = b.id WHERE code >2 AND
b.account_id = 16221);
                                                             QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------------------
 Merge Join  (cost=141372.58..141462.28 rows=1 width=8)
   Merge Cond: ("outer".id = "inner".id)
   ->  Index Scan using id_idx on b  (cost=0.00..14415.96 rows=171 width=4)
         Filter: (account_id = 16221)
   ->  Sort  (cost=141372.58..141375.27 rows=1076 width=12)
         Sort Key: a.id
         ->  Seq Scan on a  (cost=0.00..141318.40 rows=1076 width=12)
               Filter: (code > 2)
(8 rows)


Thanks,
-Prasanth.

Tom Lane wrote:
> Prasanth <dbadmin@nqadmin.com> writes:
> 
>> Hash Join  (cost=13865.30..326413.23 rows=6451 width=18)
>>   Hash Cond: ("outer".id = "inner".id)
>>   ->  Seq Scan on a  (cost=0.00..125076.37 rows=6450937 width=18)
>>   ->  Hash  (cost=10168.64..10168.64 rows=500664 width=4)
>>         ->  Seq Scan on b (cost=0.00..10168.64 rows=500664 width=4)
>>(5 rows)
> 
> 
>>The planner used to perform a index scan. I have added a lot of data in those
>>two tables. Right now both tables have millions of records.
> 
> 
> It is highly unlikely that you want an index scan for joining millions
> of records ...
> 
> However, if you do have millions in both tables, why does the planner
> think there are only 500664 rows in b?  Maybe you are overdue for ANALYZE.
> 
> 			regards, tom lane
> 
> 



view thread (14+ messages)  latest in thread

Message-ID: <428379E0.4010709@nqadmin.com>
Permalink:  ../428379E0.4010709@nqadmin.com/
Also on:    postgresql.org/message-id/428379E0.4010709@nqadmin.com

 · 

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients from your mail client; the link fills in
  the headers below: reply via email
  Some clients, webmail especially, drop In-Reply-To and References from
  a link, and the reply then starts a new thread: use the mbox below.

  To: pgsql-admin@postgresql.org
  Cc: dbadmin@nqadmin.com, tgl@sss.pgh.pa.us
  Subject: Re: Query Plan - Index Scan & Seq Scan
  In-Reply-To: <428379E0.4010709@nqadmin.com>
  References: <428370AC.3070207@nqadmin.com> <886.1115910870@sss.pgh.pa.us> <428379E0.4010709@nqadmin.com>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox