Pg takes at least 2 hours to retrieve 7650 rows - Mailing list pgsql-hackers

From José Soares
Subject Pg takes at least 2 hours to retrieve 7650 rows
Date
Msg-id 372818B6.924C610@sferacarta.com
Whole thread Raw
List pgsql-hackers
Hi all,

I have two tables MOVIMENTAZIONI with 7650 rows
CAPI with 7650 rows, when I try to join this two tables PostgreSQL
takes more than 107 minutes to retrieve rows,
the same query in:
I have installed Oracle-8 and Informix-se in the same computer
and the same query takes:
- Informix about 6 seconds.
- Oracle   about 2 seconds.

I tried it also in:
- M$-Access about 3 seconds.

I'm sure this is not a vaccum problem because I executed vacuum before
I ran the query.


CREATE TABLE capi (       matricola               CHAR(15)    NOT NULL,       specie                  CHAR(2)     NOT
NULL,      nascita                 DATE,       sesso                   CHAR(1) DEFAULT 'F',       razza
 CHAR(3),       madre                   CHAR(15),       padre                   CHAR(15),       azienda_origine
CHAR(08),      fiscale_origine         CHAR(16),       paese_origine           CHAR(03),       iscritto
BOOLEAN,      data_aggiornamento      TIMESTAMP DEFAULT CURRENT_TIMESTAMP,       PRIMARY KEY (matricola,specie)
);

CREATE TABLE movimentazioni (       azienda                 CHAR(11)    NOT NULL,       specie                  CHAR(2)
   NOT NULL,       matricola               CHAR(15)    NOT NULL,       data_introduzione       DATE        NOT NULL,
  tipo_introduzione       CHAR(2),       azienda_provenienza     CHAR(8),       fiscale_provenienza     CHAR(16),
matricola_precedente   CHAR(15),       data_applicazione       DATE,       data_uscita             DATE,
ragione_uscita         CHAR(1),       tipo_destinazione       CHAR(1),       azienda_destinazione    CHAR(8),
fiscale_destinazione   CHAR(16),       paese_destinazione      CHAR(3),       Mattatoio               CHAR(19),
n_proprietario         INTEGER,       data_aggiornamento      TIMESTAMP DEFAULT CURRENT_TIMESTAMP,       PRIMARY KEY
(azienda,matricola,specie,data_introduzione)      );
 

$ psql -c 'vacuum'
VACUUM
$ time psql -f  test.sql  2>/dev/null >/dev/null
SELECT movimentazioni.azienda
FROM movimentazioni,capi
where ((capi.matricola = movimentazioni.matricola )
and (capi.specie = movimentazioni.specie ) );

real    107m48.354s
user    0m1.140s
sys     0m0.040s

Informix-se:
real    0m6.348s
user    0m2.250s
sys     0m0.140s

Oracle-8:
real    0m2.118s
user    0m0.780s
sys     0m0.120s

This is my environment:

- [PostgreSQL 6.5.0 on i586-pc-linux-gnu, compiled by gcc 2.7.2.3]
- Pgsql snapshot Apr 15 17:42
- postmaster -i -o -F -B 512 -S
- Linux 2.0.36 Debian
- cpu:          586
- model:        Pentium MMX
- vendor_id:    GenuineIntel
- RAM:          63112
- Swap:         102812

I tried the same query in v6.4 with best results.
real    3m45.968s
user    0m0.060s
sys     0m0.160s

I tried the same query with joins inverted as:
$ psql -c 'vacuum'
VACUUM
$ time psql -f  test.sql  2>/dev/null >/dev/null
SELECT movimentazioni.azienda
FROM movimentazioni,capi
where (capi.specie = movimentazioni.specie )
and ((capi.matricola = movimentazioni.matricola ))
real    0m4.312s
user    0m1.220s
sys     0m0.090s

PostgreSQL version 6.4:
real    0m0.600s
user    0m0.130s
sys     0m0.030s

I thougth v6.5 was faster than v6.4

Any ideas?
Jose'




pgsql-hackers by date:

Previous
From: Peter Mount
Date:
Subject: RE: [HACKERS] FYI: snapshot 4/28/1999 (fwd)
Next
From: Vadim Mikheev
Date:
Subject: Re: [HACKERS] Pg takes at least 2 hours to retrieve 7650 rows