Seq scan with a like operator

Поиск
Список
Период
Сортировка
От Jouneau Luc
Тема Seq scan with a like operator
Дата
Msg-id 006d01c470b3$aa0381f0$4703668a@pc502jlo
обсуждение исходный текст
Ответы Re: Seq scan with a like operator  (Peter Eisentraut <peter_e@gmx.net>)
Список pgsql-admin
Hi,
I don't understand why with a like operator, PostgreSQL 7.4.2 does not behave the same as with an equal operator :
 
Here's a table with about 200.000 tuples :
CREATE TABLE public.annu_pers2
(
  nom varchar(50),
  prenom varchar(50),
  nom_int varchar(50),
  ....
) WITH OIDS;
CREATE INDEX personne_fk3
  ON public.annu_pers2
  USING btree
  (nom_int);
======
Here are the two explain analyze requests :
select nom_int
from annu_pers2
where nom_int ='X'
 
Index Scan using personne_fk3 on annu_pers2  (cost=0.00..125.66 rows=31 width=11) (actual time=0.359..0.359 rows=0 loops=1)
  Index Cond: ((nom_int)::text = 'X'::text)
Total runtime: 0.490 ms
-------
select nom_int
from annu_pers2
where nom_int like 'X'
 
Seq Scan on annu_pers2  (cost=0.00..5885.60 rows=31 width=11) (actual time=1213.594..1213.594 rows=0 loops=1)
  Filter: ((nom_int)::text ~~ 'X'::text)
Total runtime: 1213.729 ms
 
Could someone give me a reason of the sequential scan on the request with like operator
(I know that without % a like operator is not usefull, but it was to restrain as much as possible probability of mistakes - the problem remains the same with or without %).
 
Thanks for your answer(s).
 
Luc

В списке pgsql-admin по дате отправления:

Предыдущее
От: Terry Hampton
Дата:
Сообщение: HELP ! With Money type
Следующее
От: Terry Hampton
Дата:
Сообщение: [Fwd: Re: HELP ! With Money type]