Re: Is 292 inserts/sec acceptable performance ? - Mailing list pgsql-performance

From Rajesh Kumar Mallah
Subject Re: Is 292 inserts/sec acceptable performance ?
Date
Msg-id 200304301309.53150.mallah@trade-india.com
Whole thread Raw
In response to Re: Is 292 inserts/sec acceptable performance ?  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-performance
Ooops Sorry ,

Actually the query finished in approx 4 mins not 23 mins.
That performance must have been under some crazy circumstances.
So the insert Rate now is 1608 inserts/sec not 292 as stated
earlier.

Here is the EXPLAIN ANALYZE anyway


tradein_clients=# begin work;EXPLAIN analyze INSERT INTO general.profile_master
(email,country_code,city,title1,fname1,mname1,lname1,website,address,source,ifimporter,ifexporter,ifservice,ifmanu,creation_date)
SELECT
email,country_code,city,title1,fname1,mname1,lname1,website,address,source,ifimporter,ifexporter,ifservice,ifmanu,creation_date
from  general.email_bank_import where not exists (select * from general.profile_master where
email=general.email_bank_import.email); rollback; 
BEGIN
Time: 993.07 ms

+---------------------------------------------------------------------------------------------------------------------------------------------------------+
|                                                                       QUERY PLAN
                                 | 

+---------------------------------------------------------------------------------------------------------------------------------------------------------+
| Hash Join  (cost=8.07..2395887.30 rows=279296 width=129) (actual time=2.56..151083.30 rows=394646 loops=1)
                                 | 
|   Hash Cond: ("outer".country = "inner".name)
                                 | 
|   ->  Seq Scan on email_bank a  (cost=0.00..2390293.31 rows=279296 width=109) (actual time=0.36..41475.08 rows=394646
loops=1)                         | 
|         Filter: (NOT (subplan))
                                 | 
|         SubPlan
                                 | 
|           ->  Index Scan using profile_master_email on profile_master  (cost=0.00..31.66 rows=7 width=678) (actual
time=0.05..0.05rows=0 loops=558731) | 
|                 Index Cond: (email = $0)
                                 | 
|   ->  Hash  (cost=7.46..7.46 rows=246 width=20) (actual time=1.11..1.11 rows=0 loops=1)
                                 | 
|         ->  Seq Scan on countries b  (cost=0.00..7.46 rows=246 width=20) (actual time=0.06..0.73 rows=246 loops=1)
                                 | 
| Total runtime: 196874.70 msec
                                 | 

+---------------------------------------------------------------------------------------------------------------------------------------------------------+
(10 rows)

Time: 198905.62 ms
ROLLBACK
Time: 1481.41 ms


Regds
mallah.


On Tuesday 29 Apr 2003 7:30 pm, Tom Lane wrote:
> Rajesh Kumar Mallah <mallah@trade-india.com> writes:
> > Hi Can anyone tell if the case below is an acceptable
> > performance ?
>
> Not with that info.  Could we see EXPLAIN ANALYZE results for both
> the faster and slower cases?
>
>             regards, tom lane
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
>
> http://archives.postgresql.org

--
Rajesh Kumar Mallah,
Project Manager (Development)
Infocom Network Limited, New Delhi
phone: +91(11)6152172 (221) (L) ,9811255597 (M)

Visit http://www.trade-india.com ,
India's Leading B2B eMarketplace.


pgsql-performance by date:

Previous
From: "Jim C. Nasby"
Date:
Subject: Why LIMIT after scanning the table?
Next
From: "SZŰCS Gábor"
Date:
Subject: Re: Query Plan far worse in 7.3.2 than 7.2.1